the gruan manual - world meteorological … · web viewthe gruan manual establishes mandatory...
TRANSCRIPT
WMO/IOC/UNEP/ICSUGLOBAL CLIMATE OBSERVING SYSTEM (GCOS)WMO COMMISSION FOR BASIC SYSTEMS (CBS)
WMO COMMISSION FOR INSTRUMENTS AND METHODS OF OBSERVATIONS (CIMO)_________________
GCOS/CBS/CIMOWIGOS Pilot Project on
GRUAN Observing Practices and Governance
GENEVA, SWITZERLAND, 25 – 27 JANUARY 2012
WIGOS-PP-GRUAN/Doc. 2.2(1)
(17.I.2012)________
ITEM: 2.2
PROPOSED REVISION OF GRUAN MANUAL
(Submitted by Dr John Nash on behalf of CIMO)
SUMMARY
ISSUES TO BE DISCUSSED:This document provides a proposed revision of the GRUAN Manual that aligns the document with CBS and CIMO’s vision for GRUAN, takes into account a WMO desire for the introduction of an operational component to GRUAN, and shapes the document overall for easier adaptation of parts of it for use as input to WMO’s WIGOS Manual and Guide.
DECISIONS/ACTIONS REQUIRED:
The meeting is requested to consider this input as a CIMO contribution to revision of the GRUAN Man-ual.
REFERENCES: N/A
1
12345
67
12
THE GRUAN MANUAL
Version 1.0.0.67jn
Purpose of this ManualThis The GCOS Reference Upper Air Network (GRUAN) manual is designeddescribes standard-ized to ensure adequate uniformity in operating protocols, standardization in data reduction, meta-data collection, and data dissemination which shall be employed among participating sites and the GRUAN Lead Centre as it supportsin order for the goals of GRUAN to be achieved. These goals are agreed between and the mission of GCOS and WMO.. The primary goals of GRUAN are to provide vertical profiles of reference measurements suitable for reliably detecting changes in global and regional climate on decadal time scales, initially for temperature and water vapour, with the aim of expanding to other essential climate variables in liaison with other exist-ing scientific networks such as in Global Atmosphere Watch. The measurements will provide a calibrated reference standard for global satellite-based measurements of atmospheric essential cli-mate variables, will be a reference standard for the measurements of the existing GCOS Upper Air network.fully characterize the properties of the atmospheric column, and ensure that potential gaps in satellite measurement programmes do not invalidate the long-term climate record.
The GRUAN manual establishes mandatory practices and operating protocols (distinguished by use of the words ‘must’ or ‘shall’) having the status of requirements in a technical resolution, which it is necessary that sites within GRUAN follow or implement. It defines the requirements for GRUAN site operations, including requirements on uncertainty and long-term stability. The manual also establishes recommended practices, procedures and specifications (distinguished by use of the words ‘should’ or ‘could’) which participating sites are urged to comply with, but which are not mandatory. In this way the manual establishes the operational philosophy under which GRUAN shall operate and informs current and future GRUAN sites of the expected modus operandi for GRUAN.
While it is not planned that this document becomes part of the WMO Technical Regula-tions, which would require its approval by the WMO Congress , its description as a ‘manual’ is consistent with the WMO nomenclature i.e. it is a document that provides higher level dir-ectives and indicates where underlying ‘guides’ provide more detailed and specific informa-tion.Relevant information from this manual will be incorporated in the WMO Manual on the Gobal Observing System(WMO-No. 544) and the Guide on the Global Observing System(WMO-No.488). GRUAN Observing systems and operating practices will be referred to in the WMO Guide to Instruments and Methods of Observation (WMO-No.8) and this document will be linked to the relevant scientific documentation for the specialised scientific sounding systems. A GRUAN station may be a scientific observing site outside of the WMO operational Global Ob-serving System, but the long term observational procedures shall follow the guidelines laid down in this GRUAN manual whether it is a purely scientific site or whether it is already part of the Global Observing System, and the WMO Manuals shall reference these practices. Therefore, rather than providing detailed specifications for various systems within GRUAN, t
The manual provides defines higher level the principles that are intended to direct the develop-ment of the methods, techniques and processes needed to achieve the stated goals of GRUAN. The initial essential climate variables to addressed in detail have been temperature and water va-pour, and the scope of GRUAN will be extended as resources permit to cover a wider range of es-sential climate variables in liaison with existing scientific networks.. In essence, the manual spe-cifies what climate variables are to be observed and how those observations are to be made in or-
2
89
1011
12
13141516171819202122232425
262728293031323334
3536373839404142434445464748
495051525354
34
der to meet the relevant observational requirements of the various end-user communities of GRUAN data products. Where possible, the document does provide more in-depth detail on spe-cific methodologies , and these shall be appropriate for incorporation into or reference by exist-ing WMO literature such as the WMO Guide to Meteorological Instruments and Methods of Ob-servation.
3
5556575859
56
Table of ContentsPurpose of this Manual 1Table of Contents 2Executive Summary 4
1. INTRODUCTION 111.1. GRUAN heritage 111.2. The purpose of GRUAN 111.3. Organisation and design of GRUAN 121.4. Implementation of GRUAN 131.5. Links to partner networks 141.6. Link to satellite-based measurement programmes 19
2. REFERENCE MEASUREMENTS 212.1. Terminology 212.2. The concept of a reference measurement 212.3. Managing Change 22
3 MEASUREMENT UNCERTAINTY273.1 Estimating measurement uncertainty 273.2 Reporting measurement uncertainty 293.3 Reducing measurement uncertainty 293.4 Reducing operational uncertainty 293.5 Validating measurement uncertainty 30
4 ESSENTIAL CLIMATE VARIABLES MEASURED IN GRUAN 314.1 Justification and context for Essential Climate Variables 314.2 Priority 1 ECVs 324.3 Moving beyond priority 1 variables 36
5 GRUAN SITES 375.1 Introduction 375.2 Mandatory Operating Protocols 385.3 Criteria for Assessing Added Value 395.4 The Assessment and Certification Process 425.5 Site Auditing 44
6 INSTRUMENTATION 456.1 Instrument selection 456.2 Measurement redundancy 466.3 Surface measurements 466.4 Upper-air measurements 47
4
60
61626364
65666768697071
72737475
767778798081
82838485
868788899091
9293949596
78
6.5 Instrument co-location 476.6 Calibration, validation and maintenance 47
7 METHODS OF OBSERVATION 507.1 Guiding principles 507.2 Measurement scheduling 507.3 Operation and maintenance, quality standards 56
8 DATA MANAGEMENT 578.1 Overview of GRUAN data flow 578.2 GRUAN data policy 588.3 Collation of meta-data608.4 Data format 608.5 Data submission 618.6 Data dissemination 618.7 Data archiving628.8 Quality control at the instrument/site level 62
9 POST-PROCESSING ANALYSIS AND FEEDBACK 63
10 QUALITY MANAGEMENT64
ACRONYMS 66
Appendix A – Expanded details on additional GRUAN Essential Climate Variables 67
REFERENCES 70
5
9798
99100101102
103104105106107108109110111
112
113
114
115
116
910
Executive SummaryThe executive summary as it currently stands is excessively long. It will be condensed in sub-sequent versions of the manual.
The development and current operation of the GRUAN network is described by a number of dis-tinct but often overlapping documents, including GCOS-112, GCOS-121, GCOS-134, GCOS-140, web-based material, reports from GRUAN task teams and papers published in the interna-tional peer reviewed literature. The purpose of this manual is not to supersede that documentation. It is also not the purpose of this manual to provide authoritative, comprehensive information in GRUAN – that will continue to be provided through the GRUAN body of literature as a whole. Rather, tThe purpose of this manual is to provide a vehicle for communicatingcomprehensive doc-umentation of information and documenting messages important essential to the ongoing opera-tion of GRUAN but which may not find a natural home elsewhere in the GRUAN body of docu-mentation.. Therefore, this document is neither complete nor comprehensive in its coverage of GRUAN. The high level messages emerging from this GRUAN manual are summarized in this executive summary.
1. The purpose of GRUAN
The purpose of GRUAN shall be to:
i) Provide vertical profiles of reference measurements of temperature and water vapour suit-able for reliably detecting changes in global and regional climate, on multi-decadal time scales, for all climatically distinct regions of the globe, acting as a reference to the stabil-ity of the observations provided by the operational GCOS .Upper Air Network.
ii) Provide a calibrated reference standard for global satellite-based measurements of atmo-spheric essential climate variables., starting with temperature and water vapour.
iii) Measure sufficient additional atmospheric and surface variables, necessary to allow satis-factory simulation of the radiances observed by satellite sounding systems, at as many stations as possible..
iv) Fully characterize the properties of the atmospheric column.[v)] Ensure that any interruptions in satellite-based measurement programmes do not in-validate the long-term climate data record..
GRUAN should also provide observations in real-time for incorporation in meteorological ana-lysis, provided this is not detrimental to achieving the primary purposes of the network, as defined above.
1.1.
1.2. In the context of other WMO observing systems, GRUAN shall effectively be the climate reference backbone of the existing global operational upper-air network.2
1.3. Organisation and design of GRUAN
2.1 The GRUAN reference network will operate under the joint governance of GCOS and WMO.
2.2 Working oversight of the network will be performed by the GCOS Working group on Atmospheric Reference Observations, incorporating representatives at working level from WMO.
6
117
118
119120
121122123124125126127128129130131132
133
134
135136137138139140141142143144145146147148149
150
151152
153
154155
156157158
1112
2.3 A GRUAN Lead Centre agreed by GCOS and WMO, will be responsible for integrat-ing best practices into GRUAN operations, managing the network systems, including data management.
2.4 GRUAN sites shall use a designated system of methods, techniques and facilities, im-plemented for making and archiving best quality upper air observations on a global scale. At any site, this system will not be changed without permission from the GRUAN Lead Centre.
2.5 GRUAN network operations will incorporate an assurance programme to validate the stability and uncertainty of the measurements, agreed with WG-ARO, and managed in detail, by the GRUAN Lead centre
2.6 GRUAN shall also be responsive to the latest technological and scientific progress in measurement techniques and observational climate requirements. Development work can continue at a site until mature and validated, when it could be introduced into GRUAN operations with the agreement of the GRUAN Lead centre.
1.4. GRUAN shall be organized as part of GCOS, and advised by the WMO Commission for Basic Systems and the WMO Commission for Instruments and Methods of Observation.
1.5. GRUAN shall be constituted as a coordinated system of methods, techniques and facilities, im-plemented at GRUAN sites, for making upper air observations on a global scale.
1.6. GRUAN shall be designed as a flexible and developing network capable of continuous improve-ment, responsive to the latest technological and scientific progress in measurement techniques, and in accordance with changing requirements for observational data.
Sites comprising GRUAN shall be guided directly by a GRUAN Lead Centre designated by WMO.
[1.7.] The Lead Centre shall be responsible for implementing GRUAN operational protocols at network sites, for managing various systems that apply to GRUAN as a whole, and for collecting and integrating best practices across the network.
1.7.[1.8.] 3. Implementation of GRUAN
1.8.[1.9.]
GRUAN task teams shall contribute to the governance and operation of GRUAN. 3.1 The implementation of GRUAN shall be guided by the Working Group on Atmospheric Refer-ence observations (WG-ARO)
3.2
Specific issues in support of network design and other decisions shall be performed by GRUAN task teams shall support theas agreed by WG-ARO . and the Lead Centre in im-plementing GRUAN by considering specific issues in support of decision making, an-dThese will entrain ing operational and other relevant expertise in support of GRUAN.
3.3 A GRUAN Analysis Team for Network Design and Operations Research (GATNDOR) shall undertake focused, short-term research to address specific topics identified by the the WG-ARO. The work will be conducted in coordination with other relevant GRUAN task teams.
7
159160161
162163164165
166167168
169170171172
173
174175
176177
178179180181
182183
184185186
187
188
189190191
192
193194195196
197
198199200201
1314
3.4 The WG-ARO shall agree on the appropriate method of establishing standard operational procedures for observing systems within GRUAN. This could be a new task team, an instrument mentor, or an existing instrument team within other associated WMO projects/ operational groups.
As new measurement systems become operational within GRUAN, the WG-ARO shall give con-sideration to the establishment of a task team/instrument mentor to facilitate the implementa-tion of standard operating procedures for the new system across the whol network.
3.5 The GRUAN Lead Centre shall identify sites where instrument operators need training, and organise cost-efficient training courses for the network at appropriate locations, as advised by the appropriate task team/instrument mentor,, together with other GRUAN sites where the re-quired expertise is available, shall act as centres of excellence for certain instruments and shall conduct targeted training programmes for instrument operators across the network to encourage uniformity of instrument operation between sites.
A GRUAN Analysis Team for Network Design and Operations Research (GATNDOR) shall undertake focused, short-term research to address specific topics identified by the GRUAN science, the WG-ARO, and the Lead Centre.
GATNDOR activities shall be coordinated with the GRUAN task teams and with national GCOS programmes when appropriate.
3.6 All activities associated with the implementation of GRUAN shall be the responsibility of the countries hosting GRUAN sites and should, as far as possible, be met through national fund-ing.
4. Partner networks
4.1 GRUAN shall not operate in isolation of existing networks but shall collaborate and li-aise with existing networks to leverage skills and expertise available from those networks, avoid undue duplication of effort, and ensure that GRUAN data products are tailored to best meet the needs of partner networks.
4.2 The WG-ARO shall establish work in close coordination between GRUAN andwith the governing bodies of partner networks, with respect to the development of GRUAN opera-tions.. the GRUAN sites where GRUAN operations coincide with other network opera-tions. Specifically, the WG-ARO shall then identify WG members as liaisonsmethods of improving liaison with partner networks to ensure close communication between GRUAN and those networks. , particularly at expert team level.
4.3 In addition contact with expert teams from existing networks shall be made by WG-ARO, GRUAN task teams, and GATNDOR to support GRUAN operations.
Where possible, GRUAN shall identify, adopt, and extend if necessary, tools and methodolo-gies that have been developed in existing networks that can serve the needs of GRUAN. In particular, data QA/QC procedures developed in existing networks shall, where suitable, be adopted in GRUAN. Where networks are working towards QA/QC procedures, GRUAN should partner with these networks to develop systems that meet the operational requirements of both parties. Where GRUAN develops QA/QC techniques that are super-ior to those used elsewhere, those techniques shall be shared with partner networks.
4.4 The GRUAN Lead Centre shall facilitate identify methods of access to measurement data-base from partner networks to enable cross-calibration with GRUAN measurements and
8
202
203204205
206207208
209
210211212213214215
216217218
219220
221222223
224
225226227228
229230231232233234
235236
237238239240241242243
244245
1516
to quantitatively link GRUAN measurements to similar measurements made within part-ner networks.
4.5 The WG-ARO shall establish close connections to the satellite communityrelevant satel-lite technical programmes, such as GSICS.. Because GRUAN measurements are likely to serve a wide range of end-users within the satellite measurement community, WG-ARO members shall be assigned to liaise with key clients within the satellite community to en-sure that GRUAN data products are tailored, where possible, to best meet the needs of this community.
4.6 GRUAN shall be operated in such a way that homogeneity of measurements across the network will ensure that any significant site specific differences between GRUAN data and those from satellite-based instruments result from spatial biases in those systems an-dare not in from the GRUAN data products.
4.7 GRUAN shall provide a reference-standard that will serve as a common baseline for spli-cing satellite-based measurement time series to create CDRs.
4.8 GRUAN shall collaborate with the WMO SCOPE-CM programme to generate CDRs of upper air ECVs and in this way contribute to Action C10 defined in the GCOS imple-mentation plan (GCOS-92) viz. 'Ensure continuity and over-lap of key satellite sensors ...undertaking reprocessing of all data relevant to climate for inclusion in integ-rated climate analyses and reanalyses'.
4.9 The WG-ARO shall establish maintain active links to with the following partner net-works:
4.8.1 GUAN/WIGOS 4.8.2 GAW (Global Atmospheric Watch 4.8.3NDACC (Network for the Detection of Atmospheric Composition Change) 4.8.4 SHADOZ (Southern Hemisphere Additional Ozonesondes)4.8.5 ARM (Atmospheric Radiation Measurement) Programme 4.8.6 BSRN (Baseline Station Radiation Network)4.8.7AERONET (AErosol RObotic NETwork)BSRN (Baseline Station Radiation Network)GAW (Global Atmospheric Watch)SHADOZ (Southern Hemisphere Additional Ozonesondes)AERONET (AErosol RObotic NETwork)ARM (Atmospheric Radiation Measurement) ProgrammeNMS (National Meteorological Services)Reanalyses Centres
5 Reference measurements and managing change
5.8 All GRUAN measurements systems shall make reference quality measurements, i.e that, at a minimum, the observations are tied to an internationally accepted traceable standard, that the uncertainty on the measurement (including corrections) has been de-termined, and that the entire measurement procedure and set of processing algorithms are properly documented and accessible.
5.9 For GRUAN measurement systems making vertically resolved measurements, measure-ment uncertainties shall also be vertically resolved such that each measurement in a pro-
9
246247
248249250251252253
254255256257
258259
260261262263264
265266
267268269270271272273274275276277278279280
281
282283284285286
287288
1718
file shall be treated as a single measurement result requiring both the measurement and its uncertainty.
5.10 The methods by which the measurements are obtained and the data products de-rived shall be reproducible at any time in the futuredocumented for future reference.. Meta-data shall be archived that describe how the measurements were made, which cor-rections were applied, what changes occurred during the observation and post-observa-tion periods to the instruments and the data reduction algorithms.
5.11 GRUAN shall operate in a way such that changes in instrumentation, changes in operat-ing procedures, changes in data processing algorithms and changes in operators do not introduce unidentified temporal or spatial discontinuities or biases into GRUAN data products.
5.12 When changing instrument types, operating procedures, or data processing algorithms, GRUAN shall develop the necessary proceduresthe old and new measurement systems shall be operated simultaneously for a sufficient period of time to fully characterize any systematic biases between the old and new measurement systems so that corrections can be made to the historical data to maintain a long-term homogeneous measurement series. The length of overlapprocedures used shall be informed by robust scientific investiga-tions including a detailed understanding of the error limitations of the instrumentation in use. .
5.13 Data reduction processes and data archiving within GRUAN shall be designed such that reprocessing of historical data can be easily and quickly conducted i.e. that the original raw data can be efficiently reprocessed, as required, to form a single homogeneous time series.
5.14 Every reprocessing generating a new homogeneous time series shall be reflected in an increment in the data version and such updates shall be communicated to users who have accessed earlier versions of the data.
5.15 The GRUAN Lead Centre, and GRUAN instrument teams, shall work with CIMO and CBS actively engage with instrument manufactures to make manufacturers aware of GRUAN’s needs and to understand the constraints on instrument performance that man-ufactures face.
5.16 The GRUAN Lead Centre shall foster work with CIMO in fostering instrument inter-comparisons to develop the in-depth understanding required to manage changes from one instrument to another and to inform decisions on the relative advantages and disad-vantages of changing instrumentation.
6 Measurement uncertainty
6.1 The three primary steps for managing measurement uncertainty in GRUAN shall be:i) Describe/Analyze all sources of measurement uncertainty, as far as possible.ii) Quantify/Synthesize the contribution of each source of uncertainty to the total meas-
urement uncertainty.iii) Verify that the derived net uncertainty is a faithful representation of the true uncer-
tainty.
6.2 Every GRUAN station shall measure, collect, and provide all information necessary to establish an uncertainty budget for every measurement.All GRUAN observations must be reported in association with an uncertainty budget. The Lead Center shall liaise with
10
289290
291292293294295
296297298299
300301302303304305306307
308309310311
312313314
315316317318
319320321322
323
324325326327328329
330331332
1920
the GRUAN site in developing this uncertainty budget, as agreed with relevant instru-mentation experts.
6.3 The uncertainty in the geo-location and time coordinates associated with each measure-ment shall also be considered when identifying and describing sources of measurement uncertainty.
6.4 To reduce operational measurement uncertainties, optimal standard operating procedures shall be developed atidentified by the GRUAN Lead Centre and disseminated to all sites making that particular measurement, and adopted where practical, with exceptions clearly documented and justified.
6.5 The Lead center shall be responsible for ensuring that All estimates of measurement un-certainty shall beare validated through comparison of redundant measurements and/or laboratory analysis of the measurement systemsuitable procedures, by nominated instru-ment experts..
7 Site assessment and Certification
7.1 Site assessment and certification shall fall within the mandate of the Working Group on Atmospheric Reference Observations (WG-ARO) together with the GRUAN Lead Centre and .This shall be consistent with the guidance developed and standards agreed with the WMO Commission for Instruments and Methods of Observation (CIMO; WMO-No. 8) and the WMO Commission for Basic Systems (CBS; WMO-No. 488) .
7.2 Sites shall propose specific measurement programmes for inclusion in GRUAN and it is these that will be required to conform to the operating protocols defined in this docu-ment, and will be agreed as appropriate for GRUAN by the WG-ARO. and which will form the basis for assessing the added value that the site brings to the network as a whole.
7.3 While based on a body of sound scientific research, in assessing the value which a spe-cific site adds to the network, the The WG-ARO shall exercise its discretion in evaluat-ing the proposed inclusion of a site against the criteria defined in this document.
7.4 To identify potential issues early, the quality of the data received from the sites shall be reviewed annually.
7.5 Where site reassessments identify measurement programmes that consistently fall short of GRUAN operating standards, GRUAN certification of that programme shall be sus-pended.
7.6 If all measurement programmes at a site lose their GRUAN certification the site shall be suspended from the GRUAN network.
GRUAN is a heterogeneous network that includes sites from both the research community and the operational meteorological community. The mandatory requirements for sites reflect GRUAN’s primary goal of providing reference quality observations of the atmospheric column while accom-modating the diverse capabilities of sites within the network. GRUAN sites shall:
7.7 Provide reference quality observations, including uncertainty estimates for each datum. Profile measurements require uncertainty estimates for each measurement point on the profile.
7.8
7.9 Provide access to raw data and assure long-term storage of the raw data either at the site, another GRUAN facility, or at another internationally accessible archive.
11
333334
335336337
338339340341
342343344345
346
347348349350351
352353354355356
357358359
360361
362363364
365366
367368369370
371372373
374
375376
2122
7.10 Provide complete metadata for each measurement.
7.11 Provide Perform regular traceable pre -launch ground checks for balloon borne systems and record the results. Other , independent of the manufacturer, for any instruments which provide vertical profiles extending from the surface require regular checks to as-sure correct operation..
7.12 Work with the GRUAN Lead Centre to perform checks as requested to Provide redund-ant reference observations of the essential climate variables selected for measurement at the site at intervals sufficient to validate the derivation of the uncertainty on the primary measurement.
7.13 Provide annual reports summarizing GRUAN operations at the site.
7.14 Conduct measurement programmes with an operational philosophy of continually striv-ing to sustain the measurement quality at a given level. If improvements to measurement accuracy can be obtained these need to be documented and the introduction agreed with the GRUAN Lead Centre..
7.15 Manage changes in instrumentation, operating procedures, and processing algorithms pro-actively to avoid the introduction of spatial or temporal biases in GRUAN data products.
7.16 Actively communicate with the Lead Centre, WG-ARO, Task Teams and/or other sites, (e.g. through attendance of meetings, blog postings etc.).
Once a site has committed to operating a set of measurement programmes under these protocols the added value that a site brings to the GRUAN network will be a function of:
7.17 The extent to which a site can fulfil the following measurement programmes expected of a candidate GRUAN site as defined in this documentabove.
7.18 The extent to which the measurement programmes promoted for inclusion in GRUAN meet the requirements for random error and bias defined in this document.
7.19 The extent to which the site measurement programmes provide measurements in re-gions, or of atmospheric phenomena, which were not previously sampled.
7.20 The extent to which a site brings unique observational and/or analysis capabilities to the network as a whole and the likelihood of being able to propagate those capabilities across other sites in the network.
7.21 The extent to which a site is prepared to forgo locally established operating procedures and adhere to the standard operating procedures established by the Lead Centre.
7.22 The availability of historical measurements that conform to the GRUAN standard.
7.23 The extent to which a site can commit to a multi-decade programme of measurements.
7.24 The extent to which a site can provide redundant observations of the priority 1 variables.
7.25 The extent to which a site is capable of measuring other ECVs identified in GCOS-112 as being desired quantities.
7.26 The level of institutional support for the site and commitment to maintaining long-term reference quality measurement programmes.
7.27 The level of institutional support for the site (and any partner institutions) to undertake fundamental scientific research of the measurements from the site and other GRUAN sites.
12
377
378379380381
382383384385
386
387388389390
391392393
394395
396397
398399
400401
402403
404405406
407408
409
410
411
412413
414415
416417418
2324
Certification of GRUAN sites will not be a single event. Periodic (e.g. every 3-4 years) complete auditing of the measurement programmes included in the GRUAN certification for a site shall be conducted to ensure that the programmes continue to meet GRUAN standards.
8 Instrumentation
8.1 Periodic (every 3-4 years) reviews of instrumentation likely to be of use within GRUAN shall be undertaken by all Task Teams with an instrument focus. A written review sub-mitted to the Lead Centre shall then inform discussion related to deployment of new in-strument types across the network. New equipment shall not be introduced to GRUAN operations without the permission of the Lead Centre. If new instrumentation is pro-posed by a manufacturer/ scientific institution the performance of this system shall be reviewed by appropriate instrument experts and a case for incorporation in GRUAN then submitted to WG-ARO/ Lead Centre/CBS/CIMO.
8.2 To the extent possible, the number of different instrument types employed across the network shall be minimized to encourage homogeneity of data products and to reduce administrative overhead. However, in the case of radiosonde consumables it is essential to use radiosonde from more than one manufacturer to minimise the fluctuation in the consumable production quality/ manufacturing changes /weaknesses in the radiosondes.
8.3 A number of factors shall guide the selection of instruments for use in GRUAN includ-ing, but not limited to, instrument heritage, sustainability of measurement systems, ro-bustness of uncertainty, information content, manufacturer support, and site location.
8.4 Redundancy in core measurement systems at sites shall be used to validate derived measurement uncertainties, and also to detect changes in the systematic bias between the observing systems. The cross-checking of redundant measurements for consistency shall be an essential part of the GRUAN quality assurance procedures, but may not be required at all sites, because of logistical limitations.
8.5 .
8.6 Independent measurements of the same (or related) variables shall be reported in a con-sistent way.
8.7 Where a cluster of instruments at different locations is operating as a single GRUAN site, robust scientific analysis of the resultant increase in apparent measurement uncer-tainty shall be undertaken to clearly document how the lack of co-location contributes to the net measurement uncertainty. It must be very clearly indicated which observations are made from each location, and the data should not be merged as though it were all from one location.
8.8 Calibration proceduresOperational procedures should be harmonised shall not com-promise the goal of achieving homogeneity across the GRUAN network so that a meas-urement of a given parameter at one site is directly comparable as far as the stability of the equipment allows to a measurement of the same variable at a different site. When two identical instruments are deployed at two different sites, they shall also use the same calibration procedures. {Ed. Note: {This does not ensure all the measurements are the same as the equipment may operate differently in different conditions. }{ wWhere there are 300 sites using the same equipment around the world using one set of software , shouldn’t GRUAN use the same if it is supposed to be the backbone of the networks???})
13
419420421
422
423424425426427428429430
431432433434435
436437438
439440441442443
444
445446
447448449450451452
453454455456457458459460461462
2526
8.9 GRUAN sites shall maintain a “GRUAN site working standard” for each basis unit, e.g. a thermometer periodically calibrated to a National Metrology Institute or other accred-ited agency standard to ensure traceability to an SI standard.
8.10 The Lead Centre call shall implement a mechanism to address the compatibility with the rest of the network of those systems not be traceable to SI standards.
8.11 Each station shall maintain accurate meta-data records and provide these to the GRUAN archives. Copies of calibration certificates shall be submitted to the GRUAN meta-data-base.
9 Measurement Scheduling
9.9.1 Core radiosonde measurements.
In order to sustain good quality, and provide a backbone to GUAN, a programme of regular high quality radiosonde measurements needs to be maintained throughout the year. The nature of this programme needs to be negotiated with the GRUAN Lead centre, and will depend on the availab-ility of suitably skilled staff. The minimum programme should be one radiosonde observation per day at least 5 times per week. However, the preferred option is two radiosondes per day continu-ously. For temperature the most reliable measurements can be obtained in the dark, and this is not usually a problem at most sites regularly performing radiosonde observations .This section of the executive summary has not yet been written.
The programme of operation of remote sensing system at a site needs to be agreed with the Lead Centre and those sites which are included in GRUAN because of the deployment of special sci-entific systems( balloon borne, or remotely sensed), may be allowed to fly lower numbers of ra-diosondes than those sites making core radiosonde measurements.
Selected GRUAN sites are to fly one special scientific sounding system for measuring strato-spheric water vapour once per month. Where several GRUAN sites are located close together in the same latitude band there is no point in all making this expensive measurement and the indi-vidual site flight schedules should take this into account
Schedules of twin radiosonde flights for assurance purposes are to be agreed with the GRUAN Lead centre. E.g. the schedules for a tropical site may be different for those at midlatitides and there will be a limit on the number of cross-checking flights required in mid-latitudes, given the need to conserve limited financial resouces .
Data Management
This section of the executive summary has not yet been writtenGRUAN data circulation shall be based on the use of the WMO WIS system.. Similarly access to the GRUAN data base shall be through the WIS system.
10 Post Processing Analysis and Feedback
This section of the executive summary has not yet been written.
11 Quality Management
ARM quality radiosonde measurements shall be subject to the same quality evaluation procedures against numerical weather prediction fields as is applied to GUAN measurements.
14
463464465
466467
468469470
471
472
473474475476477478479480
481482483484
485486487488
489490491492
493
494
495
496497498
499
500
501
502503
504
2728
This section of the executive summary has not yet been written.
15
505
2930
1. INTRODUCTION
1.1. GRUAN heritageThe need for a reference upper-air network to better meet the needs of the international climate re-search community has long been recognized (Trenberth, 2003). In response to this need, the in-ception of the GCOS Reference Upper-Air Network (GRUAN; GCOS-112, GCOS-134) was formalized between 2005 and 2007 when a reference upper-air network, envisaged to eventually include 30-40 sites, worldwide was planned. In contrast to the GCOS Upper Air Network (GUAN), which is based on weather observing stations, GRUAN is specifically designed for cli-mate research. Therefore, rather than being an a purely operational network like GUAN, it is a re-search network that ultimately serves the international climate community through a combination of research and operational sites, giving high quality operational network observations and ele-ments of research and development for the future. GRUAN provides reference observations of up-per-air essential climate variables (ECVs), through a combination of in situ measurements made from balloon-borne instruments and from ground-based remotely sensed observations. Further-more, unlike other GOS networks, management decisions in GRUAN are driven by the a variety of climate requirements of for long-term climate trend detectiontime series of measurements of assured measurement stability, but also by the need for good operational practices to ensure stabil-ity in the measurements.. Nonetheless, there are aspects of GRUAN operations that clearly link to the global operational upper-air network. As a result, GRUAN has a somewhat split personality with a dual-purpose nature.So, Oon one hand GRUAN is partly a research network constantly striving to improve measurement techniques, and quantify and reduce measurement uncertainties by improving precision and accuracy. , but On on the other hand the network measurements need to be made in a stable way over multi-decadal time scales to achieve data homogeneity in time and spatially between measurement stations. In this sense GRUAN will operate more like a long-term monitoring network for the detection of climate change. This differentiates GRUAN from operational WMO observing systems. These two aspects of GRUAN operations are not mutu-ally exclusive, but do need to be carefully balanced. This dual-purpose nature of GRUAN has been accommodated in this manual.
1.2. The purpose of GRUANTThe purpose of GRUAN shall be to:i) Provide vertical profiles of reference measurements of temperature and water [and additional
essential climate variables], suitable for reliably detecting changes in global and regional cli-mate, on multi-decadal time scales, for all climatically distinct regions of the globe. The uni-formity and coherence of standard operating procedures at GRUAN stations and the resultant homogeneity of GRUAN climate data records not only provides a global reference standard for operational upper-air network stations, but improves the detection of changes in the cli-mate of the troposphere and stratosphere.
ii) Provide a calibrated reference standard for global satellite-based measurements of atmo-spheric ECVs. This facilitates the creation of seamless, stable, and long-term databases of satellite-based measurements suitable for detection of trends and variability in climate in the upper troposphere and stratosphere on all time scales.
iii) Fully characterize the properties of the atmospheric column. This is necessary for process un-derstanding and for radiative transfer modelling.???
iv) Ensure that any interruptions in satellite-based measurement programmes do not invalidate the long-term climate data record.
16
506507
508
509510511512513514515516517518519520521522523524525526527528529530531532533534
535
536537538539540541542543544545546547548549550551
3132
GRUAN should shall also provide radiosonde observations in real-time for incorporation in met-eorological analysis, provided this is not detrimental to achieving the primary purposes of the net-work, as defined above in order to fulfil the requirement of providing a reference to the opera-tional observations..
In achieving the four goals stated above, GRUAN will address some of the current deficiencies of the GUAN network. The reliable detection of the vertical structure of changes in climate variables in the atmosphere requires high quality atmospheric observations with well characterised meas-urement uncertainties. While GUAN provides upper air measurements over large regions of the globe, using radiosondes that are similar to those used in GRUAN. However GUAN sites these are primarily for operational weather forecasting and as a result seldom include additional sys-tems to guarantee validate data quality stability, and rely on the assumption of stability in the ra-diosonde quality with time. If GRUAN can identify the changes that occur in production consum-ables, this will benefitthose using GUAN measurements.such that the data are suitable
for long-term trend detection. For example GUAN does not include, as part of its design, regular intercomparisons between measurements at different sites to ensure homogeneity in data quality and traceability.
In the context of the other WMO observing systems, GRUAN will effectively need to be be the climate reference backbone of the existing global operational upper-air network. As noted in GCOS-112, GRUAN sites need not necessarily be current GUAN sites. Because GUAN sites of-ten operate with different equipment, sensors, and operating protocols, the different requirements of GRUAN and GUAN operations may require careful management. The envisaged capabilities of a fully-implemented GRUAN are detailed in GCOS-112. The scientific justification and re-quirements for GRUAN are summarized in Section 3 of GCOS-112 and in Seidel et al. (2009) and are not repeated here. Continued implementation of GRUAN is specifically called for under Ac-tion A16 of the 2010 update to the implementation plan for GCOS (GCOS-138).
17
552553554555556557
558559560561562563564565566
567568569
570571572573574575576577578
3334
1.3. Organisation and design of GRUAN{Ed Note: [this needs to be replaced}
Figure 1: Schematic outline of the structure of GRUAN. GRUAN elements are shown in red while exter-nal support structures are shown in black. GATNDOR=GRUAN Analysis Team for Network Design and Operations Research.The GRUAN reference network will operate under the joint governance of GCOS and WMO. A schematic outline of the GRUAN governance structure is given in Figure 1.
Working oversight of the GRUAN network will be performed by the GCOS Working group on Atmospheric Reference Observations(WG-ARO), including representatives at working level from WMO. GCOS and WMO will select those groups ( e.g. GCOS/WCRP Atmospheric Observations Panel, or WMO Technical Commission working groups/ex-perts) through which WG-ARO will report.
A GRUAN Lead Centre agreed by GCOS and WMO, will be responsible for integrating best practices into GRUAN operations and managing the GRUAN network systems, in-cluding data management. This Lead Centre is currently hosted by the German Weather Service at the Lindenberg Meteorological Observatory in Germany. The GRUAN Lead Centre acts as the interface between GRUAN and the community of users of GRUAN products. For example, data transfer to end-users is not made from GRUAN measurement sites but is first shared within the GRUAN community, subjected to the QA/QC proced-ures developed within GRUAN (Section 10), and then submitted by the GRUAN Lead Centre to the GRUAN data repository (NCDC, Section 8.6).
GRUAN sites shall use a designated system of methods, techniques and facilities, imple-mented for making and archiving best quality upper air observations on a global scale. At
18
579580
581582583584
585
586587588589590
591592593594595596597598599
600601
3536
any site, this system will not be changed without permission from the GRUAN Lead Centre. GRUAN operations shall integrate where possible and when feasible with other in-ternational climate monitoring programmes.
GRUAN network operations will incorporate an assurance programme to validate the sta-bility and uncertainty of the measurements, agreed with WG-ARO, and managed in detail, by the GRUAN Lead centreGRUAN shall also be responsive to the latest technological and scientific progress in measurement techniques and observational climate requirements. Development work can continue at a site until mature and validated, when it should be introduced into GRUAN operations with the agreement of the GRUAN Lead centre. GRUAN shall be organized as part of GCOS, and advised by the WMO Commission for Basic Systems (CBS) and the WMO Commission for Instruments and Methods of Observation (CIMO). A schematic outline of the GRUAN governance structure is given in Figure 1.
GRUAN shall be constituted as a coordinated system of methods, techniques and facilities, implemented at GRUAN sites, for making upper air observations on a global scale, integ-rating where possible and when feasible with other international climate monitoring pro-grammes.
GRUAN shall be designed as a flexible and developing network capable of continuous im-provement, responsive to the latest technological and scientific progress in measurement techniques, and in accordance with changing requirements for observational data. The design of GRUAN shall recognise the heterogeneity of the network of sites, many of which will have primary responsibility to networks other than GRUAN.
The sites comprising GRUAN shall be guided directly by a GRUAN Lead Centre designated by WMO who also sponsors the GCOS Steering Committee. This Lead Centre is currently hosted by the German Weather Service at the Lindenberg Meteorological Observatory in Germany. The Lead Centre shall be responsible for implementing GRUAN operational protocols at network sites, for managing various systems that apply to GRUAN as a whole, and for collecting and in-tegrating best practices across the network.
The GCOS steering committee guides the GCOS/WCRP Atmospheric Observation Panel for Cli-mate (AOPC). The AOPC in turn guides the Working Group for Atmospheric Reference Observa-tions (WG-ARO) which guides the development of GRUAN, is responsible for GRUAN site se-lection (Section 5), develops guidelines for observations and data, and ultimately guides the GRUAN Lead Centre.
The GCOS Secretariat provides additional support to the GCOS Steering Committee, the AOPC, the WG-ARO and the GRUAN Lead Centre. The GRUAN Lead Centre acts as the interface between GRUAN and the community of users of GRUAN products. For example, data transfer to end-users is not made from GRUAN measurement sites but is first shared within the GRUAN community, subjected to the QA/QC procedures developed within GRUAN (Section 10), and then submitted by the GRUAN Lead Centre to the GRUAN data repository (NCDC, Section 8.6).
1.4. Implementation of GRUAN
The implementation of GRUAN shall be guided by the Working Group on Atmospheric Refer-ence observations (WG-ARO).
Specific issues to be investigated in support of GRUAN implementation shall be performed by GRUAN task teams set up by WG-ARO .These will entrain operational and other relevant ex-pertise in support of GRUAN and will work in coordination with the GRUAN Lead Centre.
19
602603604
605606607608609610611612613614
615616617618
619620621622623
624625626627628629
630631632633634
635636637638639640
641
642643
644645646
3738
A GRUAN Analysis Team for Network Design and Operations Research (GATNDOR) shall un-dertake focused, short-term research to address specific topics identified by the WG-ARO. The work will be conducted in coordination with other relevant GRUAN task teams. GATNDOR activities shall be coordinated with the GRUAN task teams and with national GCOS programmes when appropriate.
The WG-ARO shall agree on the appropriate method of establishing standard operational proce-dures for all observing systems within GRUAN. This could be a new task team, including investi-gations at the GRUAN Lead centre, an instrument mentor, or an existing instrument team within other associated WMO projects/ operational groups. The task teams shall evaluate the appropri-ateness of uncertainty estimates, the usefulness of particular measurements and operational pro-cedures, synthesize the available knowledge and develop recommendations to improve GRUAN measurements and operations. These task teams shall confer regularly to evaluate the current status of GRUAN observations, to identify weaknesses, and to incorporate new scientific under-standing into GRUAN. The expertise of these teams shall also be used to support the Lead Centre in guiding individual stations through instrumental and operational changes without impacting long-term measurement time series.
The GRUAN Lead Centre shall identify sites where instrument operators need training, and or-ganise cost-efficient training courses for the network at appropriate locations, as advised by the appropriate task team/instrument mentor, to encourage uniformity of instrument operation between sites.
In addition to the Lead Centre and GRUAN measurement sites, GRUAN task teams, and a GRUAN Analysis Team for Network Design and Operations Research (GATNDOR) shall contribute to the governance and operation of GRUAN. GRUAN measurements are made at the Lead Centre and at GRUAN measurement sites which should, as far as practical, implement the same measurement practices recommended by the Lead Centre. All activities associated with the implementation of GRUAN are the responsibility of the countries hosting GRUAN sites and should, as far as possible, be met through national funding.
To best serve the needs of climate monitoring and research, it is essential that GRUAN be in-formed by a good understanding of the evolving science issues that drive the measurements and accuracy of the GRUAN data. Therefore, the establishment of an internal or external science ad-visory panel should be considered (GCOS-134)
As discussed in Section Error: Reference source not found, there may be cases where, for very good reasons, the measurement protocols recommended by the Lead Centre cannot be employed. The extent to which this might detract from the value a site might add to the GRUAN network shall be left to the discretion of the WG-ARO. The GRUAN Lead Centre, together with other GRUAN sites, which may act as centres of excellence for certain instruments, shall conduct tar-geted training programmes for instrument operators across the network to encourage uniformity of instrument operation between sites.The instrumentation deployed and the observing schedules may differ between sites , as agreed with WG-ARO, but the methods of observations used with the main observing systems, i.e. radiosondes and balloon borne Specialised Sounding Instruments shall be uniform between all the GRUAN sites.
20
647648649650651
652653654655656657658659660661662
663
664665666667
668669670671672673674675676677678679680681682683684685686687688689690691692
3940
GRUAN task teams shall support the WG-ARO and the Lead Centre in implementing GRUAN by considering specific issues in support of decision making, and entraining operational and other relevant expertise. The task teams shall evaluate the appropriateness of uncertainty estimates, the usefulness of particular measurements and operational procedures, synthesize the available know-ledge and develop recommendations to improve GRUAN measurements and operations. These task teams shall confer regularly to evaluate the current status of GRUAN observations, to identify weaknesses, and to incorporate new scientific understanding into GRUAN. The expertise of these teams shall also be used to support the Lead Centre in guiding individual stations through instrumental and operational changes without impacting long-term measurement time series.
Following ICM-1 (GCOS-131), the GATNDOR team was established to undertake scientific in-vestigations (in additional to the more operational investigations undertaken by the task teams) in support of GRUAN decision making and to report at subsequent ICM meetings. GATNDOR shall undertake focused, short-term research to address specific topics identified by the GRUAN sci-ence, the WG-ARO, and the Lead Centre. GATNDOR activities shall be coordinated with the GRUAN task teams and with national GCOS programmes when appropriate. To best serve the needs of climate monitoring and research, it is essential that GRUAN be informed by a good un-derstanding of the evolving science issues that drive the measurements and accuracy of the GRUAN data. Therefore, the establishment of an internal or external science advisory panel should be considered (GCOS-134).
Links to partner networksEstablishing GRUAN is not just an exercise in adding another acronym label to existing measure-ment sites. While in the charter for GRUAN (GCOS-92) it is stated that ‘where feasible, these ref-erence sites should be co-located and consolidated with other climate monitoring instrumenta-tion’, GRUAN will require a mode of operation, and the establishment of measurement pro-grammes, currently not available anywhere. The purpose of this section is to provide, as early as possible in this document, a context for GRUAN in the broader community of climate monitoring networks. For instance, in the charter for GRUAN (GCOS-92) it is stated that ‘where feasible, these reference sites should be co-located and consolidated with other climate monitoring instru-mentation’,
GRUAN shall not operate in isolation of existing networks and GRUAN is not intended to replace in any way existing networks. Many GRUAN initial and candidate sites already belong to existing networks such as GUAN, GAW, NDACC, BSRN and SHADOZ. One of the essential character-istics of a successful GRUAN is close coordination with the user community and many of these networks are also likely to be users of GRUAN data. Similarly, complementary measurements from these other networks should be collated in a database to enable cross-calibration and to quantitatively link GRUAN measurements to similar measurements made within other net-works. As a result, close coordination between the governing bodies of these networks and with the WG-ARO is required on a continuous basis. This close coordination can be achieved by hav-ing members of the WG-ARO attend steering group meetings of partner networks and by inviting co-chairs from partner networks to attend WG-ARO meetings.
There is a wide range of tools and methodologies that have been developed in existing networks that GRUAN can adopt, extend if necessary, and learn from. Similarly, existing networks will have skills and expertise likely to be useful to GRUAN and its operations. As a result, contact with expert teams from existing networks shall be made by WG-ARO, GRUAN task teams, and GATNDOR to support GRUAN operations and to avoid duplication of effort by utilizing existing scientific knowledge.
A number of networks currently in operation make measurements which fall within the scope of GRUAN. Of particular interest areFor instance, those stations that make upper air measurements
21
693694695696697698699700701
702703704705706707708709710711
712713714715716717718719720721
722723724725726727728729730731732
733734735736737738
739740
4142
that are not part of the typical meteorological measurements of temperature, pressure and humid-ity. Many of these networks have developed systems for assuring the quality of their measure-ments, including GUAN.. Where the systems currently in place are sufficient to meet the opera-tional requirements of GRUAN, they should be used by GRUAN. Where networks are working towards QA/QC procedures, GRUAN should partner with these networks to develop systems that meet the operational requirements of both parties. In some cases, sites within these partner net-works may also become GRUAN sites. This is encouraged since it facilitates a traceable link between GRUAN measurements and measurements made at all other sites within the partner net-work (assuming that the measurements within the partner network are cross-calibrated and can be quantitatively linked).
Existing networks and potential resources from within those networks likely to be of value to GRUAN are discussed below.
1.4.1. GAW (Global Atmospheric Watch)
The GAW programme of WMO is a partnership involving 80 countries, providing reliable sci-entific data and information on the chemical composition of the atmosphere, and the natural and anthropogenic drivers of changes in chemical composition. In this way, GAW improves under-standing of the interactions between the atmosphere, the oceans and the biosphere. GAW has strong linkages to GCOS and so is likely to have skills and resources that could be used to support GRUAN.
1.4.2. NDACC (Network for the Detection of Atmospheric Composition Change)
NDACC reports to GAW. The NDACC comprises more than 70 high-quality, remote-sensing re-search stations for observing and understanding the physical and chemical state of the stratosphere and upper troposphere and for assessing the impact of stratospheric changes on the underlying tro-posphere and on global climate. Primarily based on lidar measurement techniques. NDACC incor-porates 5 water vapour measurement sites and maximum of 8 temperature measurement sites.Be-cause GRUAN and NDACC share a number of common science goals, it has been debated whether GRUAN is necessary and whether NDACC could achieve the goals of GRUAN. There are a number of key differences between NDACC and GRUAN that require GRUAN to operate as a new and independent network, including: NDACC aims to observe and understand the chemical composition of the stratosphere and
upper troposphere. For GRUAN the highest priority observations are the atmospheric state variables of temperature, pressure and humidity.
The primary focus of NDACC is on ozone and the chemicals responsible for ozone depletion. The primary focus of GRUAN is on climate and the factors driving changes in climate. This is evidenced by the fact that NDACC does not include measurements of incoming and outgo-ing longwave and shortwave radiation, nor measurements of various cloud parameters such as cloud amount/frequency, base height, layer heights and thicknesses, cloud top height, cloud top pressure, cloud top temperature, and cloud particle size. While GRUAN does include measurements of a few trace gases (ozone, methane, and carbon dioxide) it excludes the wide range of trace gases measured within NDACC.
The vertical domains for GRUAN and NDACC are quite different. NDACC focuses primar-ily on the upper troposphere and stratosphere whereas GRUAN focuses primarily on the tro-posphere and lower stratosphere. .
NDACC operates as a federation of independent measurement sites. While NDACC does have in place stringent standards which must be met for measurement programmes to become part of the network, . However, large numbers of unlike GRUAN, the NDACC network is
22
741742743744745746747748749750
751752
753
754755756757758759
760
761762763764765766767768769770771772773774775776777778779780781782783784785786
4344
not controlledballoon borne measurements in GRUAN requires by a Lead Centre that aims to implement a minimum set of standard operating procedures across the network as a whole..
One of the primary goals of GRUAN is to detect long-term climate trends above the Earth's sur-face. NDACC does aim to make long-term measurements of changes in chemical composi-tion in the upper troposphere and stratosphere but this is not the primary focus of the network.
Because NDACC does not operate under the control of GCOS, it does not have the institutional mandate to act as the reference standard for the GUAN which is a key purpose of GRUAN.
There are, however, a number of measurements and operational procedures common to both net-works and every effort should be made to avoid duplication of effort and to ensure that the lessons learned within NDACC are assimilated into GRUAN. For example: The NDACC has established 'working groups' that are primarily centred on specific instru-
ments used within the NDACC. GRUAN task teams currently include a mix of teams focus-sing on specific measurements systems (radiosondes and precipitable water from GNSS) and on network wide operational issues. As more measurement systems are incorporated into GRUAN operations, consideration should be given to later expanding the ‘Ancillary Meas-urements’ Task Team to include specific measurement systems in addition to the 'cross-cut-ting' task teams that focus on issues common to the network as a whole. This could be achieved through assigning ‘instrument mentors’ as recommended in GCOS-112 and utilized operationally in the ACRF programme. Task teams focussing on specific measurement sys-tems or on specific ECVs would better link to advisory groups within partner networks e.g. the Scientific Advisory Groups within GAW (see Section 1.4.5). SCOPE-CM (see Section 1.5) intends to establish one or two centres to lead the generation and provision of funda-mental climate data records for each ECV and so establishing task teams within GRUAN fo-cussed on specific ECVs or groups of ECVs would mirror the structure within SCOPE-CM and thereby facilitate interactions with the satellite-based measurement community (one of the key clients of GRUAN).
Measurements of vertical ozone and water vapour profiles made within NDACC will be com-mon to measurements made within GRUAN. This includes both balloon-sonde and lidar measurements.
Techniques have been developed within NDACC to manage changes in instrumentation. GRUAN should build off the expertise developed in this community over the past two dec-ades e.g.
i) The JOSIE ozonesonde intercomparisons (Smit et al., 2007).ii) Regional ozone profile intercomparisons from multiple instruments (McDermid et al.,
1998a; McDermid et al., 1998b).iii) Intercomparisons of vertical water vapour profile measurements.
Measurement redundancy in the NDACC network sites has been a strength of the network since it allows intercomparisons of supposedly identical measurements by different instru-ments which often highlight previously unknown deficiencies in the measurements (Brinksma et al., 2000). GRUAN will include similar measurement redundancy (see Section 6.2).
1.4.3. BSRN (Baseline Station Radiation Network)
The BSRN provides a worldwide network to continuously measure radiative fluxes at the Earth's surface. The network comprises about 40 stations between 80°N and 90°S many of which began operation in 1992 and each year more stations are added to the network. These stations provide data for the calibration of measurements made within the GEWEX Surface Radiation Budget (SRB) Project and other satellite-based measurements of radiative fluxes. BSRN data are also
23
787788789790791792793794
795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827
828
829830831832833
4546
used to validate radiative flux models. BSRN data are archived at the Alfred Wegener Institute (AWI) in Bremerhaven, Germany. In 2004, BSRN was designated as the global surface radiation network for the GCOS. The BSRN stations also contribute to GAW (see Section 1.4.5).
The primary goal of BSRN is to monitor the shortwave and longwave radiative components and their changes with the best methods currently available. Therefore the measurements of longwave and shortwave incoming and outgoing radiation within GRUAN will overlap with the measure-ments made within BSRN. Access to the BSRN calibration facilities at the Physikalisch-Meteoro-logisches Observatorium Davos (PMOD)/World Radiation Centre (WRC) would be highly ad-vantageous to GRUAN. The BSRN includes a working group on measurement uncertainties that could be used to provide guidance for establishing the radiation measurement uncertainties within GRUAN.
1.4.4. WOUDC (World Ozone and UV Data Centre)
The WOUDC is one of the World Data Centres which are part of the GAW (see Section 1.4.5) programme of WMO. The WOUDC, operated by the Experimental Studies Section of Environ-ment Canada in Toronto, is not so much a network as an international repository for ozone and UV data. There are many practices employed within the ozone measurement community that are likely to be useful to GRUAN. For example, the management of the Dobson Spectrophotometer and Brewer Spectroradiometer networks, both of which provide data to the WOUDC, demonstrate many of the principles that form the foundation for GRUAN. These include: Undertaking regular regional intercomparisons of instruments which always include a travel-
ling standard which facilitates standardization of instrument performance between regions. Archiving of raw data to permit later reprocessing should new improved ancillary data be-
come available e.g. the shift to the Bass and Paur ozone absorption cross-sections in the late 1980s. A similar process is now underway to evaluate a possible change from the Bass and Paur cross-sections to e.g. the Daumont (Daumont et al., 1992) cross sections.
Careful QA/QC of data before archiving and strict version control of data submitted to inter-national archives.
These principles have resulted in ground-based total column ozone time series of sufficient qual-ity to allow detection of the multi-decade decline in ozone until the end of the 20 th century and the onset of ozone increases thereafter.
1.4.5. GAW (Global Atmospheric Watch)
The GAW programme of WMO is a partnership involving 80 countries, providing reliable sci-entific data and information on the chemical composition of the atmosphere, and the natural and anthropogenic drivers of changes in chemical composition. In this way, GAW improves under-standing of the interactions between the atmosphere, the oceans and the biosphere. As with the NDACC, the primary focus of GAW is on changes in atmospheric composition, although GAW focuses on different atmospheric species to NDACC. GAW has strong linkages to GCOS and so is likely to have skills and resources that could be used to support GRUAN.
1.4.6. SHADOZ (Southern Hemisphere Additional Ozonesondes)
The SHADOZ project was initiated to remedy the lack of consistent tropical ozonesonde observa-tions. This was done by increasing the frequency, and improving the quality, of ozonesonde launches at selected tropical ozone observing stations (Thompson et al., 2003). Rather than estab-lishing an entirely new network, SHADOZ aims to enhance ozonesonde launches at existing facil-ities on a cost-share basis with international partners. The geographical coverage of the network was specifically designed to address target research questions.
24
834835836
837838839840841842843844
845
846847848849850851852853854855856857858859860
861862863
864
865866867868869870871
872
873874875876877878
4748
1.4.7. AERONET
AERONET (AErosol RObotic NETwork) is a federation of ground-based remote sensing aerosol networks with contributions from national agencies, institutes, universities, individual scientists, and research partners. The programme provides a long-term, continuous and publically accessible database of aerosol optical, microphysical and radiative properties. The standardization of instru-ments, calibration procedures, and data processing and distribution is well aligned with the needs of GRUAN.
The AERONET programme provides globally distributed observations of spectral aerosol optical depth (AOD), products derived from the raw measurements, and precipitable water in diverse aer-osol regimes. Aerosol optical depth data are computed for three data quality levels: Level 1.0 (un-screened), Level 1.5 (cloud-screened), and Level 2.0 (cloud-screened and quality-assured). It is primarily the level 2.0 data that are likely to be of interest to GRUAN since these data are quality-assured. Inversions, precipitable water, and other AOD-dependent products are derived from these levels and may implement additional quality checks.
1.4.8. Atmospheric Radiation Measurement (ARM) Programme
The goal of the Department of Energy ARM programme is to study alterations in climate, land productivity, oceans or other water resources, atmospheric chemistry, and ecological systems that may alter the capacity of the Earth to sustain life. This includes improving the atmospheric data sets used in regional and global climate models. A primary objective of the ARM user facility is improved scientific understanding of the fundamental physics related to interactions between clouds and radiative feedback processes in the atmosphere.
Four of the 15 initial GRUAN sites are also ARM sites in part because the radiation measure-ments made at these sites satisfy many of the ECV measurement requirements within GRUAN. The dedicated Data Quality (DQ) Office which ARM established in July 2000 to coordinate and implement efforts to ensure the quality of the data collected by ARM field instrumentation will likely provide a number of tools which could be implemented across the GRUAN network to en-sure the quality and network homogeneity of the radiation measurements. The DQ Office has the responsibility for ensuring that quality control results are communicated to data users so that they may make informed decisions when using the data, and to ARM's Site Operations and Engineers to facilitate improved instrument performance and thereby minimize the amount of unacceptable data collected. The ARM DQ Office has developed a suite of sophisticated data quality visualisa-tion tools that are likely to be of interest to GRUAN.
Another ARM organizational structure that is likely to be relevant for GRUAN is the assignment of instrument mentors. Because GRUAN task teams are not structured by instrument (as is the case for NDACC where each working group focuses on one instrument), having ARM-type in-strument mentors that advise on instrument operation, maintenance and calibration across the net-work as a whole may be beneficial. Instrument mentors have an excellent understanding of in situ and remote-sensing instrumentation theory and operation and have comprehensive knowledge of the scientific questions being addressed with the measurements made. They also possess the tech-nical and analytical skills to develop new data retrievals that provide innovative approaches for creating research-quality data sets.
1.4.9. Partnership with Meteorological agencies
Meteorological agencies producing global real-time analyses or historical reanalyses (e.g. NCEP/NCAR, ECMWF, JMA, Met Office, DWD and NASA) are likely to be users of the high quality data produced by GRUAN. Well developed systems for monitoring the quality of opera-tional observations ( whether it is the performance of individual radiosonde stations or the bias
25
879
880881882883884885
886887888889890891892
893
894895896897898899
900901902903904905906907908909910
911912913914915916917918919
920
921922923924
4950
corrections required by current satellite observations In turn, diagnostics obtained from the 4D-Var assimilation schemes used in such activities will provide information on the consistency of the GRUAN measurements with other data used in the operational analyses (thereby facilitating easier comparisons of GRUAN measurements with e.g. satellite-based measurements) and on the representativeness of the uncertainty estimates on the GRUAN data. If the GRUAN data are to be used in the 4D-Var assimilation schemes used by the reanalysis centres, it is essential that the pre-cise 4D (latitude, longitude, altitude and time) coordinates associated with any measurement are available (see Section 8.3). Reference sites will prove essential for helping to characterize obser-vational biases and the impact of observing system changes, as well as to understand model er-rors, all of which are important aspects in creating high-quality reanalyses (Schubert et al., 2006). The additional value provided by the GRUAN measurements in such data assimilations should be quantified since this would provide additional scientific justification for GRUAN operations. Once a sufficiently large database of GRUAN measurements has been accumulated, such a study could be undertaken through collaboration between GATNDOR and perhaps the SPARC data as-similation activity. Such a study would also be of interest to organisations undertaking global met-eorological reanalyses. Because GRUAN will make profile measurements at vertical resolutions much higher than can be retrieved from satellites, it will provide valuable insights into the poten-tial limitations of satellite-based measurements for the analyses of specific atmospheric phenom-ena.
Some GRUAN sites may also be National Meteorological Service (NMS) sites, or may be paired with an NMS site to extend the range of measurements performed, with the result that NMSs are likely to provide partial or full support for a site.
1.5. Link to satellite-based measurement programmesGRUAN will provide data sets, not currently available, that will be useful to the satellite measure-ment community for calibrating and validating satellite-based sensors, for providing input to radi-ative transfer calculations used in satellite-based measurement retrievals, and for removing offsets and drifts between satellite-based data sets when creating merged data products. Because the GRUAN measurements are likely to serve a wide range of end-users within the satellite measure-ment community, WG-ARO members shall be assigned to liaise with key clients within the satel-lite community, and with other data providers (e.g. the Radiation Panel within GEWEX), to en-sure that GRUAN data products are tailored, where possible, to best meet the needs of this com-munity. Once GRUAN datasets are available, pilot studies on enhanced datasets using these refer-ence measurements need to be undertaken.
1.5.1. Forward modelling for satellite-based measurement retrievals
{Ed Note: [This retrieval process for temperature and humidity is not required by numerical mod-els, so probably should not be used in any argument here.]]Satellite-based measurements of atmospheric parameters often rely on an optimal estimation ap-proach (e.g. Rodgers 2000) to derive profile or slant column density information of these paramet-ers. Optimal estimation employs a forward model that is used to simulate the radiance field that a satellite-based sensor would sample for a given state of the atmosphere. To determine a state vec-tor (the true values of the atmospheric parameters of interest), together with uncertainties, from the observed satellite-based radiance measurement, typically in the form of a spectrum, the for-ward model calculations need to be inverted. Such an inversion is typically poorly constrained, i.e. does not have a unique solution, and, as a result, known a priori (background) information about the variables to be retrieved is usually required as input to the forward model. GRUAN measure-ments may provide such a priori information. Furthermore, GRUAN measurements of atmo-spheric state variables such as temperature, pressure and humidity that partially define the radiat-
26
925926927928929930931932933934935936937938939940941942943
944945946
947
948949950951952953954955956957
958
959960961962963964965966967968969970971
5152
ive transfer properties of the atmosphere, and which are required as input to the forward model, can significantly reduce the uncertainties on other retrieved atmospheric parameters.
1.5.2. Calibration and validation of satellite-based sensors
In addition to being used as part of the satellite retrieval, gGround-based reference profile meas-urements may also provide an independent standard against which the satellite retrievals may be validated. For example, Vömel et al. (2007a) demonstrate how reference-quality in situ water va-pour measurements can be used to validate satellite-based observations of stratospheric water va-pour. In addition to validating retrieved products, satellite radiances require calibration against a ground truth to unambiguously remove biases (Ohring et al., 2005) in order to be useful for cli-mate monitoring. GRUAN and the GSICS are complementary in meeting this need.
New satellite missions have higher resolution and better station-keeping, resulting in better con-trol of diurnal sampling. Global Positioning System Radio Occultation (GPS-RO) measurements are also highly promising,in use at least for upper-tropospheric and lower-stratospheric temperat-ure, and validated by comparison with numerical weather prediction fields. Even though they rep-resent a significant step forwards, these more recent satellite observing systems will not be ad-equate for climate purposes unless they can be suitably validated. To this end GRUAN will also provide shorter-term quality assured measurements for the validation of satellite-based retrievals.
The need for inter-station homogeneity within GRUAN has special significance for validation of satellite-based measurements. If satellite-based measurements agree well with gGround-based measurements made at one all GRUAN stations but disagreeshall be made in similar fashion with measurements made at another, this will significantly weaken the utility of GRUAN measure-ments for satellite instrument validation (unless of course there is a spatial bias in the satellite re-trieval), so that differences in the soundings of temperature and water vapourbetween GRUAN sites are as small as possible . If this is achieved. Satellite-based measurements can, in this way, identify potential inter-station inhomogeneities, differences in collocation biases between GRUAN measurement and satellite radiance will then primarily be a function of systematic bias in the satellite radiance or caused by a difference between sites in other conditions, e.g. thin clouds in the satellite field of view, surface emissivity, etc..
The issue of measurement scheduling within GRUAN to accommodate satellite validation activit-ies is discussed further in Section 7.2.
1.5.3. Creating global homogeneous atmospheric climate data records
While satellite-based measurements have the advantage of providing global or near-global geo-graphical coverage, the quality and usefulness of the measurements is compromised by an inabil-ity to conduct regular calibrations, limited vertical resolution, difficulties in continuity due to drifting orbits (which, for species showing strong diurnal variation can alias into apparent trends), and limited instrument lifetimes which require data series from multiple instruments to be spliced together to form long-term data records. Discontinuities between satellite-based measurements of climate variables, while not important for weather forecasting purposes, can be ruinous for detect-ing variability and long-term changes in climate. The reference measurements that GRUAN will produce can be used to remove offsets and drifts between these separate satellite-based measure-ment series within the limitations imposed by the uncertainties on the GRUAN measurements. In this way GRUAN shall provide a reference-standard that will serve as a common baseline when splicing satellite-based measurement time series. Specifically, differences between a given satel-lite-based data set and the GRUAN reference-standard can be analyzed using the algorithms de-tailed in Alexandersson et al. (1997) and Khaliq et al. (2007) to automatically detect steps and drifts in the differences. The underlying systematic structure in such differences can then be used
27
972973
974
975976977978979980981
982983984985986987988
989990991992993994995996997998999
10001001
1002
100310041005100610071008100910101011101210131014101510161017
5354
to homogenize the satellite-based measurements with the GRUAN reference-standard. Similar ap-proaches using the global ground-based Dosbon and Brewer spectrophotometer networks to create long-term global total column ozone records from multiple satellite-based measurements have been developed (Bodeker et al., 2001) as have techniques for homogenizing radiosonde temperat-ure time series (Haimberger, 2007).
By contributing to the creation of global homogeneous trace gas data bases, GRUAN will connect to the WMO SCOPE-CM (Sustained, Co-Ordinated Processing of Environmental Satellite Data for Climate Monitoring) programme. The aim of SCOPE-CM is to establish a network of facilities ensuring continuous and sustained provision of high-quality satellite products related to ECVs, on a global scale, responding to the requirements of GCOS. GRUAN and SCOPE-CM shall collabor-atively contribute to Action C10 defined in the GCOS implementation plan (GCOS-92) viz. 'En-sure continuity and over-lap of key satellite sensors ...undertaking reprocessing of all data relevant to climate for inclusion in integrated climate analyses and reanalyses' (Action C8 in the 2010 up-date of the GCOS implementation plan; GCOS-138).
28
10181019102010211022
102310241025102610271028102910301031
5556
2. REFERENCE MEASUREMENTS
2.1. TerminologyThe following terminology is used throughout this manual to describe the uncertainty components of a reference measurement:Measurement uncertainty: Every measurement has imperfections that give rise to some uncer-tainty in the result, hereafter referred to as the measurement uncertainty. Traditionally, a measure-ment uncertainty is viewed as having two components, a random component and a systematic component.Random error: The random error component of any measurement is the result of stochastic vari-ation in quantities that influence that measurement. While random errors cannot be designed out of a system, the random error on the mean of multiple measurements is reduced since, by defini -tion, the expected value for the random error is zero. The term ‘random error’ is preferred over the term ‘precision’ since precision is often used to designate the number of bits or significant digits to which a value is specified.Bias: The bias represents the difference between the measured value and the true value. It results from systematic errors that do not average to zero as the number of measurements increases. However, if biases can be identified and quantified, they can be corrected for. The term ‘bias’ is preferred over the term ‘accuracy’ since it denotes more clearly that the deviation is systematic-ally in one direction. Stability: Stability refers to the consistency of random errors and biases with time. Undetected trends in biases induce artificial trends in measurement time series.
2.2. The concept of a reference measurementAs denoted by its title, the primary objective of GRUAN is to provide reference measurements for a range of upper-air climate variables. Reference quality atmospheric observations are based on key concepts in metrology (measurement science), in particular traceability. Metrological traceab-ility is the process whereby a measurement result, i.e. a measurement and its error, can be related to a reference through a documented, unbroken chain of calibrations, each of which contributes to the measurement error.
A reference measurement does not refer to a measurement that is perfect, nor to a measurement that will never change. Rather it refers to our current best estimate of the value for some atmo-spheric parameter, as well as a best estimate for the level of confidence that is associated with this value, recognising that future improvements in measurement techniques and/or reprocessing fol-lowing new knowledge may lead to refinements in that reference value. In most cases it will be the best technology available that will achieve the best estimate of the value for some target atmo-spheric parameter. Reference measurements accommodate the unavoidable sources of uncertainty in the compilation of the net measurement uncertainty while excluding those sources of uncer-tainty that can be avoided. For example, in the pre-deployment calibration of a sensor, there will be some unavoidable uncertainty in the accepted measurement standard and hence some unavoid-able uncertainty in the calibration which must then be included in the net measurement uncer-tainty. However, contributions to measurement uncertainty from e.g. an improperly documented traceability chain, proprietary methods, appeal to physical principles without experimental verific-ation, or the use of an improper calibration standard must be avoided. Similarly, when the instru-ment is later deployed, there will be numerous, unavoidable contributions to the total measure-ment uncertainty from e.g. uncertainty in the input data, data processing constants, the data re-
29
10321033
1034
1035103610371038103910401041104210431044104510461047104810491050105110521053
1054
105510561057105810591060
1061106210631064106510661067106810691070107110721073107410751076
5758
trieval algorithm, and in the physical/chemical model of the measurement system used to convert raw measurements into data. However, contributions to measurement uncertainty from the use of ‘black box’ software, undocumented or unvalidated measurement adjustments, or the disregard of known biases must be avoided.
A reference measurement may not necessarily be the outcome of a measurement using a single in-strument but may be an average of measurements from one instrument or an average of results from multiple instruments. This highlights the importance of measurement redundancy (see Sec-tion 6.2) in that access to coincident multiple measurements of the same quantity often leads to a more robust estimate of the true value and a better estimate of the uncertainty on that value.
The estimate for the level of confidence on any measurement is expressed as the measurement un-certainty and is a property of the measurement that combines instrumental as well as methodolo-gical uncertainties. The measurement uncertainty describes the current best knowledge of instru-ment performance under the conditions encountered during an observation, it describes the factors impacting a measurement as a result of operational procedures, and it makes all factors that con-tribute to a measurement traceable. Within GRUAN this uncertainty shall be vertically resolved and each measurement in a profile shall be treated as a single measurement result requiring both the measurement and its uncertainty. To provide the best estimate for the instrumental uncertainty, a detailed understanding of the instrumentation is required for the conditions under which it is used. Specific requirements that an observation must fulfil to serve as a reference for calibrat-ing or validating other systems, have been defined in Immler et al. (2010).
A reference measurement typically results from a measurement procedure that provides sufficient confidence in its results by relating to well-founded physical or chemical principles, or a measure-ment standard that is calibrated to a recognized standard, in general a standard provided by a Na-tional Metrological Institute (NMI). For GRUAN, a reference measurement is one where the un-certainty of the calibration and the measurement itself is carefully assessed. This includes the re-quirement that all known biases have been identified and corrected, and, furthermore, that the un-certainty on these bias corrections has also been determined and reported. An additional require-ment for a reference measurement is that the measurement method and associated uncertainties should be accepted by the user community as being appropriate for the application.
The methods by which the measurements are obtained and the data products derived shall be re-producible by any end-user at any time in the future. It should be kept in mind that these end-users are likely to use GRAUN data for decades to come. They shall be able to reproduce how measurements were made, which corrections were applied, and be informed as to what changes occurred during the observation and post-observation periods to the instruments and the al-gorithms. Hence maintenance of comprehensive rich metadata regarding data provenance and pro-cessing is key.
In brief, reference within GRUAN means that, at a minimum, the observations are tied to a trace-able standard, that the uncertainty on the measurement (including corrections) has been determ-ined, and that the entire measurement procedure and set of processing algorithms are properly documented and accessible.
2.3. Managing ChangeThis section is being developed as a stand-alone document and once completed will be reinteg-rated into the manual.
GRUAN recognizes that change is inevitable – changes in instrumentation, changes in operating procedures, changes in data processing algorithms and changes in operators. Such changes intro-duce sources of operational uncertainty into GRUAN data products. Rather than designing a sys-
30
1077107810791080
10811082108310841085
10861087108810891090109110921093109410951096
109710981099110011011102110311041105
1106110711081109111011111112
1113111411151116
1117
11181119
112011211122
5960
tem that is resistant to change, GRUAN appreciates that without change, improvement is im-possible. Therefore, the goal is to manage changes that have scientific value in a way that does not compromise the integrity of the long-term climate records being measured while, as far as prac-tical, minimizing those changes that have no likely scientific benefits. To this end the GRUAN network shall develop robust protocols for managing change. One of the core tasks for GATNDOR is to provide the basis for developing such protocols which will be provided by the Lead Centre.Given the wide range of observing systems that potentially may be deployed as part of GRUAN operations, the protocols for (a) assuring high stability and (b) deciding when an im-provement merits a change to the GRUAN methods of observations will need to be developed as required by WG-ARO and developed by the appropriate instrumentation specialists, given guid-ance on the user requirements when required by GANTDOR.
With the radiosonde observations, the standard procedures recommended by the LEAD centre shall be used, and the equipment and methods of observation in daily use shall not be changed, without agreement from WG-ARO, as advised by the GRUAN Lead centre. Improvements to performance can be developed at GRUAN sites, but the evidence that the improvement justifies changing the GRUAN radiosonde protocol must be rigorously assessed, before any change to GRUAN observations is considered by WG-ARO.
Managing change within GRUAN is a key component of maintaining network homogeneity. If changes are implemented unilaterally at a single location, and even if those changes are imple-mented such that the long-term homogeneity of the measurement record at that station is pre-served, the change may introduce inconsistency with other stations in the network. Therefore, changes in measurement systems at GRUAN stations shall be conducted in such a way that the homogeneity of the resultant GRUAN data products is not compromised.
The first focus in managing change as detailed in the GCOS climate monitoring principles is to ensure that when transitioning from older to newer instrumentation, that a sample of coincident measurements, sufficient to quantify any biases between the two systems, is obtained before the older system is retiredevidence is presented to the appropriate GRUAN task team. For example flying dual ozonesondes has proven to be useful when shifting from one ozonesonde system to an-other or from one standard operating procedure to another (Boyd et al., 1998). The length of time for which the older and newer systems should be run in parallel, and the frequency with which co-incident measurements should be made, will depend on the instruments used and on an in-depth understanding of the measurement technique. The overlap period may also depend on the site since seasonal variability may differ between sites such that a site experiencing greater atmo-spheric variability may take longer to derive a robust estimate of differences between measure-ment systems than a site experiencing lower atmospheric variability. Such decisions shall be in-formed by robust scientific investigations. Until the results of such research are available, sites should err on the side of caution and undertake an approach that maximizes overlap so that sub-sampling can be undertaken later to determine a minimum safe level of overlap required to pre-serve the record (see approach used at Tateno in Section 2.2.1). Where it may not always be feas-ible to operate older and newer instruments side-by-side for extended periods of time e.g. with balloon-borne instruments, alternating between the newer and older instruments is particularly useful in diagnosing and correcting systematic inter-instrument differences; regression analysis techniques including a basis function that is set to 1 for one measurement system and to 0 for an-other can be used to extract biases between the two systems. These biases can be derived as func-tions of other variables such as air pressure, temperature, time of day, solar zenith angle etc..
31
11231124112511261127112811291130113111321133
1134
113511361137113811391140
1141
114211431144114511461147
1148114911501151115211531154115511561157115811591160116111621163116411651166116711681169
6162
As new and more in-depth knowledge of various measurement systems is gained, reprocessing of historical data will be necessary. Data reduction processes and data archiving within GRUAN shall be designed with this in mind i.e. that the original raw data (which must always be archived) can be easily and regularly reprocessed, as required, to form a single homogeneous time series that is then provided to end-users. Each change in instrumentation, operating procedure or data processing algorithm is likely to require reprocessing of historical raw data to ensure a consistent product. Protocols shall be established to indicate when reprocessing of the full measurement re-cord is justified. Every reprocessing generating a new homogeneous time series over the complete measurement period shall be reflected in an increment in the data version as prescribed in the data versioning protocols developed by the Lead Centre. Such updates shall be communicated to users who have accessed earlier versions of the data (see Section 8.6). For this reason it is also import-ant that all older versions of any data set are always made available through the GRUAN archives.
A discussion of specific sources of changes is presented below but in general this requires dealing with breakpoints in the measurement time series. It is far more preferable that these changes are identified a priori through the available meta-data that identifies such changes. However, it is also possible to identify breakpoints in measurement time series based on the statistical behaviour of the data themselves. Significant resources and techniques have already been developed within the surface climate community (see e.g. http://www.homogenization.org) and upper air climate com-munity around this issue (e.g. Haimberger, 2007) although homogenizing upper-air records is more challenging that homogenizing surface climate records.
These techniques must be grounded in quantitative understanding of the causes of offsets and drifts between two different measurement systems i.e. the reliance should not be on the imple-mentation of signal processing techniques that identify and correct for offsets and drifts in time series. This quantitative understanding in turn should emerge from the meta-data associated with each measurement and from in-depth knowledge of each measurement system.
2.2.1. Changes in instrumentation
Changes in instrumentation are both inevitable and desirable if they lead to more precise measure-ments of the true atmospheric state or if they can save on the cost of making the measurement (without compromising measurement quality). Instrument changes will also often be driven by the necessities of production engineering (when instrument components become unavailable or too expensive) and decisions will have to be made as to what level of component change requires ad-ditional change testing. Formal instrument intercomparisons will be essential for developing the in-depth understanding required to manage changes from one instrument to another and for in-forming decisions on the relative advantages and disadvantages of changing instrumentation. For this reason, participation in formal intercomparisons should be a pre-requisite for the adoption of any instrument within the GRUAN network. Outcomes from such intercomparisons would form an important component of the meta-data archived at the GRUAN Lead Centre. GRUAN need not necessarily organise these intercomparisons themselves. WMO and partner networks (e.g. NDACC) often run instrument intercomparison campaigns and GRUAN should participate in these and share the data where possible. Such participation would be mutually beneficial to both communities. GRUAN needs to work closely with CBS and CBS and CIMO to gain maximum benefit for all parties from these intercomparisons. In addition to intercomparisons of similar in-struments (e.g. radiosondes), intercomparisons between different instruments measuring the same ECV will also be highly informative (e.g. comparisons of ozonesondes, ozone lidars and ozone microwave radiometers at a single site). A number of case studies exist which can be used as ex-amples of how to manage changes in instrumentation. For example the impacts of changes from the Meisei RS2-91 type radiosonde to the Vaisala RS92-SGPJ type GPS sonde at Tateno were
32
1170117111721173117411751176117711781179118011811182
11831184118511861187118811891190
11911192119311941195
1196
119711981199120012011202120312041205120612071208120912101211121212131214121512161217
6364
quantified by conducting dual sonde flights during four intensive observation periods in Decem-ber 2009, and in March, June and September/October 2010.
Following a scientifically robust replacement strategy that maximises the maintenance of long-term climate records will be important for ensuring the integrity of the GRUAN data products in the face of change. A goal within the 'Management of Change' research topic of the GATNDOR team is to provide scientific bases to develop operational practices to better manage instrument changes at GRUAN sites and to accurately merge disparate data segments to create a homogen-eous time series. Consideration will need to be given to the desired strategy when more than one station in the network is making an identical (or very similar) change with respect to timing, shar -ing of data, and whether certain sites will act as pioneers. This will be especially important where the change is forced by a supply issue.
Measurement redundancy (see Section 6.2) has significant benefits for managing instrument change as a second instrument measuring the same ECV can be used as a common reference against which both old and new instruments can be compared over an extended period. This bene-fit increases further when three or more instruments measure the same ECV and any changes are substantially staggered. An ideal aim that assures the record is therefore at least triple redundancy. The same advantages could be achieved through the use of a travelling standard instrument. For in situ balloon-borne instruments, consistent ground-check routines between new and old instru-ments will minimize changes in procedural uncertainty contributions.
Dealing with changes in instrumentation will require GRUAN to establish close two-way links to instrument manufacturers. Inclusion of the Association of Hydro-Meteorological Equipment In-dustry (HMEI) in discussions of instrument change within GRUAN would be advantageous. A productive point of interaction with the different vendors and manufacturers would be the periodic GRUAN participation in the CIMO multi-sensor field campaigns. Engaging the manufacturers in these field campaigns will assist GRUAN not only in evaluating the different sensors but also as a point of interaction with the vendors apart from the limited HMEI attendance at GRUAN meet-ings. A close cooperation between GRUAN and instrument suppliers will also help GRUAN to better understand industry capabilities and to better quantify instrumental uncertainties. This co-operation will also help suppliers to better understand GRUAN requirements, and the industry would be able to advise GRUAN of its current and prospective abilities to meet these require-ments. For many of the parameters of interest (as instruments of required accuracy do not yet ex-ist), GRUAN aims to further their development in cooperation with instrument manufacturers. HMEI has suggested that a workshop specifically for manufacturers and open to all HMEI mem-bers would be helpful.
Detailed archiving of instrument meta-data will be vital to managing changes in instrumentation. This will allow later reprocessing of the raw data as 'deep' as possible. Since it is not always known in advance which meta-data are likely to be required for reprocessing at a later date, GRUAN operators should err on the side of collating as much meta-data as possible about meas-urement systems even if no immediate use for those data can be envisaged. In all cases sufficient meta-data must be available to tie the new instrument via a comparable traceability chain back to the same recognized standard as the old instrument.
2.2.2. Changes in operating procedures
Even if instruments themselves do not change, changes in the operating procedures for an instru-ment may also introduce breakpoints in a measurement time series. For the most part, changes in operating procedures should be dealt with in a fashion similar to changes in instrumentation e.g. reprocessing of historical data to homogenize the time series and redistribution of the data with an updated version number will almost certainly be required. The expectation is that standard operat-
33
12181219
122012211222122312241225122612271228
12291230123112321233123412351236
123712381239124012411242124312441245124612471248124912501251
1252125312541255125612571258
1259
12601261126212631264
6566
ing procedures for all instrument types within GRUAN will be archived at the Lead Centre and that this body of material will be used to advise sites through transitions in operating procedures. As discussed in Section Error: Reference source not found the extent to which I site is prepared to conform to GRUAN standard operating procedures will be one of the criteria used when evaluat-ing the potential inclusion of the site in GRUAN.
2.2.3. Changes in data processing algorithms
New knowledge and resultant improvements in reduction of raw data to useful measurements are likely to lead to changes in data processing algorithms. As for changes in operating procedures, such changes in data processing algorithms should be dealt with in a fashion similar to changes in instrumentation. At the very least every change in data processing algorithm must be reflected in a change in version number of the final data product. Because raw data from various GRUAN sites will be processed at one location and one location only (either the Lead Centre or some other GRUAN site with particular expertise in that measurement), changes in data processing al-gorithms shall be implemented uniformly across the network. To achieve homogeneity across the network it is important that individual sites do not independently implement changes in data pro-cessing algorithms for data submitted as GRUAN data even if those changes are well documented and follow the prescriptions listed above. This more central, 'top-down' approach to data pro-cessing is different from the more decentralized approach employed in other networks. While such enforced conformity incurs an operational cost, the advantage is that end-users of the GRUAN data products will see data homogeneity not only in time for single stations, but also between stations. In support of maintaining consistency in the use of data processing algorithms within GRUAN, the Lead Centre shall maintain an archive of data processing algorithms which then also comprises an important part of the meta-data archive for GRUAN.
Tension may arise where a site may wish to implement a non-standard (at least non-standard for GRUAN) data processing algorithm for some purpose e.g. to create a data product that is tailored for a specific need. Such eventualities can be accommodated by having a central processing facil-ity for each GRUAN product (see above) where a common data processing procedure is applied to the ‘rawest’ form of data collected. This would not preclude a site from implementing non-standard processing of the raw data and serving this for their own purposes.
2.2.4. Change in operators
Ideally the quality of the measurements should be immune from changes in operators. This is more likely achievable if standard operating procedures are developed where there is reduced op-portunity for idiosyncrasies of operators to affect the measurements. Meta-data should include codes (not names to protect the privacy of operators) to denote where different operators have been responsible for measurements.
34
12651266126712681269
1270
12711272127312741275127612771278127912801281128212831284128512861287
128812891290129112921293
1294
12951296129712981299
6768
3 MEASUREMENT UNCERTAINTY
3.1 Estimating measurement uncertaintyNo measurement can be made perfectly, and estimating measurement uncertainty is a central tenet in GRUAN’s operations. A common GRUAN definition of measurement uncertainty and a com-mon procedure to establish measurement uncertainties is required to homogenize uncertainty es-timates across the network. It is also needed to make the steps leading to the determination of measurement uncertainty traceable. This common definition should, ideally, be adopted by instru-ment providers as well.
Achieving a useful estimate of measurement uncertainty may require as much, if not more, effort than making the measurement itself. However, such effort is necessary to achieve the goal of GRUAN to provide reference measurements from the surface to the upper stratosphere. The avail-ability of an estimate of the measurement uncertainty for every measurement made within GRUAN will significantly improve the utility of the measurements and will elevate the GRUAN measurements above what is currently available.
The availability of sufficiently detailed meta-data is vital to quantifying random errors and biases in measurements. The more detailed the meta-data, the 'deeper' the measurement uncertainty can be traced. The approach that should be followed is that where some calibration, reference stand-ard, application of an operating procedure, or use of a data processing algorithm introduces a source of uncertainty into a measurement, complete details about that uncertainty source must be available through the meta-data tagged to that measurement. Such sources of meta-data may in-clude (Immler et al., 2010) previous measurement data, experience with or general knowledge of the behaviour and properties of relevant materials and instruments, manufacturer’s specifications, data provided in calibration and other certificates, and uncertainties assigned to reference data taken from handbooks. It is vital that all sources of measurement uncertainty are made transpar-ently available to end-users of GRUAN measurements.
A particular challenge for GRUAN in estimating measurement uncertainty is that for in situ meas-urements of upper-air ECVs, the instrumentation operates in conditions that are difficult to replic-ate in a controlled environment (e.g., a test chamber). Calibration of the instrument in its operat-ing environment where e.g. transient influences of changes in solar radiation and/or clouds are likely to affect sensor characteristics is generally not possible. Furthermore, the staple instruments for much of GRUAN, viz. balloon-borne sondes, are used for measurements of single profiles. The well calibrated instruments with quantified measurement errors are discarded after each pro-file measurement and re-calibration or re-characterization after a measurement is often not pos-sible even if the instrument is recovered. The emphasis is then on employing standards that ensure stability, traceability, and uniformity between instruments and across the GRUAN network as a whole.
Because one of GRUAN’s primary goals is to detect long-term climate trends in the upper atmo-sphere, the primary consideration might be to work towards reducing changes in the systematic bias the random error in measurements i.e. to emphasize reproducibility. However, because GRUAN data are likely to be used for other purposes such as satellite validation, acting as a refer-ence for GUAN, or as input to global meteorological reanalyses, reducing biases to achieve the best possible accuracy also needs to be a priority. Therefore the aim should be to identify and minimize both random errors and biases and random errors, and to include the effects of both when calculating measurement uncertainties.
35
1300
1301
1302
130313041305130613071308
130913101311131213131314
13151316131713181319132013211322132313241325
13261327132813291330133113321333133413351336
13371338133913401341134213431344
6970
The GRUAN mantra policy for dealing with measurement uncertainty shall be: Describe/Analyze all sources of measurement uncertainty. Quantify/Synthesize the contribution of each source of uncertainty to the total measurement
uncertainty. Verify that the derived net uncertainty is a faithful representation of the true uncertainty.
3.1.1. Describe/Analyze sources of measurement uncertainty
The first step in the process of deriving an uncertainty associated with any measurement is to first fully explore and describe each source of uncertainty in the form of biases and random errors. Contributions to the net measurement uncertainty are likely to include sensor calibration, sensor integration, sensor performance and external influences to operational routines such as sensor pre-paration and sensor ground-checks. While a specific sensor might perform well, if its value de-pends in some way on another sensor that performs less well, this source of uncertainty needs to be accounted for. For example, if a very precise and accurate temperature measurement is made but the vertical coordinate for that measurement is a less precise pressure measurement, in the presence of large T/p, the uncertainty in pressure can introduce significant uncertainty in the temperature measurement. Therefore uncertainty in the geo-location and time coordinates associ-ated with each measurement shall also be considered when identifying and describing sources of measurement uncertainty. A full list of sources of measurement uncertainty will be defined in the GRUAN common definition of measurement uncertainty terms. Every GRUAN station shall measure, collect, and provide all information necessary to establish an uncertainty budget for every measurement.
3.1.2. Quantify/Synthesize sources of uncertainty
The second step is, where possible, to quantify and correct for any measurement biases. Uncer-tainty in such bias corrections, which shall also be diagnosed, documented and quantified, then contributes to the random error on the measurement. Once all biases have been corrected for, and assuming all remaining random errors are normally distributed about the mean, the resultant net uncertainty on the measurement can be reported as a single value i.e. the first standard deviation of the distribution (1σ errors). Where systematic biases cannot be determined, or perhaps can be determined but cannot be corrected for, or when remaining random errors are not normally dis-tributed about the mean, a different approach is required for quantifying the net uncertainty on the measurement. In such cases, because the net uncertainty is no longer represented by a Gaussian distribution, it cannot be reported as a single value. Techniques to fully describe the shape of the error distribution must then be developed and higher order moments of the distribution (e.g. the skewness or kurtosis) would need to be reported as part of the measurement uncertainty descrip-tion. If a measurement process can be simulated, and if the probability distribution functions (PDFs) of the various sources of uncertainty are well known, a Monte Carlo approach can be used to generate a large ensemble of ‘virtual’ measurements from which measurement uncertainty stat-istics can be calculated. This approach can be used no matter how structured or asymmetrical the individual PDFs might be. This approach has been used to estimate asymmetric errors in ozonesonde measurements (Bodeker et al., 1998).
3.1.3. Verify measurement uncertainties
The uncertainty budget for every GRUAN measurement should be verified at regular intervals us-ing redundant observations from complementary instruments (see Section 6.2). If coincident ob-servations of the same ECV are available and are subjected to the same uncertainty analysis, the degree to which the measurements agree within their stated uncertainties is indicative of the valid-ity of the measurement uncertainties. If measurements agree within their uncertainties, the error
36
13451346134713481349
1350
135113521353135413551356135713581359136013611362136313641365
1366
136713681369137013711372137313741375137613771378137913801381138213831384
1385
13861387138813891390
7172
estimates on the measurements are more likely to be correct. Formal methods have been de-veloped to achieve this (Immler et al., 2010). For example, if two large sets of data are compared and more than 4.5% of the data are statistically significantly different within their error bars, then either a systematic effect in either or both measurement sets has been overlooked, or the uncer-tainties have been under-estimated. On the other hand, if much less that 32% of measurement dif-ferences are smaller than the RMS of the uncertainties, then the measurement uncertainties have probably been over-estimated. This verification by itself does not provide a statement about the usefulness of a measurement; it only provides information about the completeness of an uncer-tainty analysis. Including such comparisons in operational data processing can act as a flag for where error analysis within the processing may not be complete.
GRUAN includes both in situ and remote sensing methods. In the case of in situ methods, the in-strument is generally calibrated directly to the geophysical quantity of interest. In the case of re-mote sensing methods, the calibrated data are in physical units of radiance and/or frequency, which are then analyzed to provide estimates of the underlying climate variable of interest. Valid-ation of data products, which is equivalent to verifying measurement uncertainties, is therefore a two-step process whereby the accuracy of both the instrument calibration and the analysis al-gorithm, are validated.
3.2 Reporting measurement uncertaintyAn overarching principle for the operation of GRUAN is that no measurement should be provided without also providing an estimate of the measurement uncertainty. Where all sources of system-atic error in the measurement have been identified and corrected for, the measurement uncertainty can be quoted as the standard deviation of the random error. As discussed above, where biases re-main in the measurement, or where the net random error in the measurement does not follow a Gaussian distribution, alternative methods for reporting the measurement uncertainty must be con-sidered. This may be in the form of establishing 1σ upper and lower bounds on the measurement uncertainty to denote that the uncertainty is asymmetric – generally reported as where X is the measurement, u is the 1σ uncertainty in the positive direction and l is the 1σ uncertainty in the negative direction. For more complex distributions of measurement uncertainty it may be neces-sary to quote the most likely value i.e. the peak in the PDF for the measurement and parameters that detail the shape of the PDF (or a pointer to the PDF itself).
3.3 Reducing measurement uncertaintyChanges in instrumentation or operating procedures may lead to reductions in measurement un-certainty. In such circumstances it is important that the same detail of uncertainty analysis is con-ducted for the new instrument/operating procedure as has been done for the instrument/operating procedure to be replaced.
In some circumstances, e.g. in the presence of high natural variability (such as for temperature and water vapour), reducing measurement uncertainty has little impact on derived trends since the primary source of the variability in the trend estimate might be the measurand noise on the signal being analyzed. It is therefore important that scientific analyses guide where reducing measure-ment uncertainties is most likely to lead to reductions in uncertainties in trend estimates.
3.4 Reducing operational uncertaintyOperational uncertainty includes uncertainties related to instrument set-up, sampling rates and the application of algorithms for data analysis. The contribution of operational uncertainty to the total measurement uncertainty in GRUAN is likely to be significantly reduced if the ‘rawest’ form of
37
139113921393139413951396139713981399140014011402140314041405140614071408
1409
141014111412141314141415141614171418141914201421
1422
1423142414251426
14271428142914301431
1432
143314341435
7374
measurement data is submitted to a central GRUAN data processing facility (see Section 8.1) where a single verified, validated and well described data processing algorithm is applied to the raw data. Similarly, the adoption of an identical standard operating procedure for each instrument type across the network, would reduce the operational uncertainties related to instrument set-up. To this end, optimal standard operating procedures shall be developed at the GRUAN Lead Centre and then disseminated to all sites making that particular measurement and adopted where practical with exceptions clearly documented and justified agreed with the WG-ARO.(see Section Error: Reference source not found).
3.5 Validating measurement uncertaintyOnce the uncertainty on a measurement has been calculated, the question then becomes: how well does this measure of uncertainty represent the degree of confidence we should have in this meas-urement? Two approaches are available for validating the derived uncertainty on any measure-ment, viz.:
3.5.1. Comparison of redundant measurements
A traditional way of validating measurement uncertainty is to measure the quantity of interest through two (or more) techniques, based on physically different measurement principles. Because the different techniques are subject to unique measurement uncertainties, comparisons yield a ro-bust and continuous demonstration of measurement accuracy. Where simultaneous measurements of the same quantity are made using two different techniques, and disagree within their stated measurement uncertainties it suggests that either one or both of the measurements are erroneous, or that the measurement uncertainties are under-estimated. In this way, complementary measure-ment techniques with different susceptibilities to local conditions can be chosen to maximize the accuracy of the data record. Additionally, uncertainty budgets validated in this way may help identify other error sources that cannot be compensated for by complementary sensors, but may be monitored in situ.
3.5.2. Laboratory analysis of the measurement system
The ability to simulate a specific measurement in the laboratory can permit an in-depth investiga-tion of the various sources of uncertainty in the measurement. For example, the environmental simulation facility at the Research Centre Juelich (Smit et al., 2007) has provided information to validate measurement uncertainty in ozonesondes. Similar laboratory facilities are available for radiosonde at Lindenberg.
38
14361437143814391440144114421443
1444
1445144614471448
1449
14501451145214531454145514561457145814591460
1461
14621463146414651466
7576
4 ESSENTIAL CLIMATE VARIABLES MEASURED IN GRUAN
Since GRUAN’s goal is not only to provide long-term high quality climate records, but also the ancillary data required to interpret those records, a number of parameters in addition to the funda-mental atmospheric state variables of temperature, pressure, and humidity would add value to the GRUAN database. High quality measurements of atmospheric state variables, trace gas concentra-tions, the atmospheric radiation environment, and cloud and aerosol properties will significantly aid the interpretation of priority 1 variables (see below). Many of these parameters have been identified by GCOS as Essential Climate Variables (ECVs; GCOS-92). A subset of ECVs has been selected as the most scientifically important and most tractable for GRUAN (see Appendix 1 of GCOS-112). As scientific research into the underlying causes of observed changes in upper-air climate advances, and as the capabilities of GRUAN sites expand, this list is likely to grow.
4.1 Justification and context for Essential Climate VariablesThe complete list of ECVs targeted by GRUAN is given in Appendix 1 of GCOS-112. The pur-pose of this section is to provide additional scientific justification and context, and more general guidelines for the measurement requirements for those ECVs listed as priority 1 for GRUAN, viz. temperature, pressure, and water vapour. The complete list of ECVs targeted by GRUAN is given in Appendix 1 of GCOS-112.A summary of material for the priority 2, 3 and 4 variables is provided in Appendix A.
The desired performance requirements for each of the ECVs are based on the scientific require-ments of the data and not on current instrument performance, so they may not be currently achiev-able.Therefore, this section provides clear operational expectations for the measurement of prior-ity 1 ECVs for GRUAN sites. However, this manual recognizes the heterogeneity of the network and its state of development. Therefore, the requirements imposed on current and putative GRUAN sites, as detailed in GCOS-112, may not be immediately achievable. In such cases the WG-ARO and Lead Centre shall provide possible incremental approaches to achieving the target attributes for each measurement. Because the desired operations parameters for each of the ECVs are based on the scientific requirements of the data and not on current instrument performance, they may not be currently achievable. Therefore, as stated in GCOS-112, these GRUAN require-ments should be interpreted as eventual measurement goals of any given network site.
Setting these parameters ambitiously high may discourage potential sites from joining GRUAN since they may not be able to immediately achieve these standards. On the other hand setting the parameters low is likely to result in stagnation since once achieved there will be little incentive to advance. For this reason the requirements detailed below are somewhat different to those listed in the WMO/CBS requirements. The values in Appendix 1 of GCOS-112 describe what is required of the measurements to meet specific research goals and a distinction needs to be made between what is desirable and what is feasible. While they may not be currently achievable, as measure-ment technology advances, attaining such targets should become more likely. In no case should an inability to achieve these targets result in the exclusion of a site or a measurement programme from the GRUAN network. However, a GRUAN site shall use currently available equipment in a manner ensuring optimum performance from that system. Development and improvement of sys-tems at GRUAN sites is to be encouraged, but these developments should be performed in a man-ner that does not interfere with the stability of GRUAN network observations. This manual recog-nizes that GRUAN is less about meeting prescribed measurement standards and more about estab-lishing an approach that continually strives to improve measurement precision and accuracy, ex-tend the range of coverage, and achieve higher sampling.
39
14671468
1469147014711472147314741475147614771478
1479
148014811482148314841485
14861487148814891490149114921493149414951496
1497149814991500150115021503150415051506150715081509151015111512
7778
The measurement ranges prescribed in Appendix 1 of GCOS-112 should cover the range of values likely to be encountered over the vertical range of interest so that any proposed instrument, or set of instruments, would need to be able to operate throughout that range. Measurement precision refers to the repeatability of the measurement as measured by the standard deviation of random er-rors. However, measurement precision is closely tied to the frequency of observations since obser-vations are often averaged and the greater the sample size, the less stringent the required precision in terms of the uncertainty on the mean. Measurement frequencies are not specified because they may vary over time (see Section 7.2). Measurement accuracy refers to the systematic bias in a measurement (the difference between the measured or derived value, and the true value). It is not directly specified for many variables for which variations, and not absolute values, are needed to understand processes. Measurement accuracy is directly related to long-term stability, the max-imum tolerable change in bias over time, which is a critical aspect of the reference network. To ensure that realistic climate trends can be derived from the dataset, the effect of any intervention to the measurement system on measurement error, such as a change in instrument, should be smaller or quantified to a much greater degree than the value given for long-term stability. The re-quirements stated in Appendix 1 of GCOS-112 are largely consistent with the GCOS ECV obser-vation requirements, as detailed in the WMO observing requirements database1..
4.2 Priority 1 ECVs
4.2.1. Temperature
Scientific justification: Upper-air temperatures are a key dataset for the detection and attribution of tropospheric and stratospheric climate change since they represent the first order connection between natural and anthropogenically driven changes in radiative forcing and changes in other climate variables at the surface. Furthermore, the vertical structure of temperature trends is im-portant information for climate change attribution since increases in atmospheric long-lived greenhouse gas (GHG) concentrations warm the troposphere but cool the stratosphere steepening vertical temperature gradients in extra-tropical regions. Other drivers of atmospheric temperature changes, e.g. changes in solar output, would not have the same vertical profile fingerprint. Re-maining discrepancies between temperature trends derived from satellite-based measurements and from radiosondes weaken the attribution of changes in temperatures to changes in climate forcing agents. High quality temperature measurements within GRUAN will contribute to the resolution of these discrepancies.
Because rRadiosondes will remain the a primary workhorse within the global upper-air network for the measurement of temperature, pressure and humidity, it is imperative that GRUAN sites es-tablish state-of-the-art radiosonde measurement programmes that match the optimum stability of performance obtainable to date. In addition, efforts should continue to improve continually strive to improve the quality of radiosonde measurements, where it is known there are significant limita-tions in performance for use in climatological obervations, see WMO(2011). Other measurement techniques can and should be developed to extend the height range of the temperature profile measurements and to reduce the random error and bias on the measurements. However, these should always be quantitatively inter-compared with collocated radiosonde measurements to provide a traceable link to the radiosonde measurements made within GUAN. Temperatures measured by high-quality radiosondes are needed to: Monitor the vertical structure of local temperature trends.
1 http://www.wmo-sat.info/db/
40
1513
15141515151615171518151915201521152215231524152515261527152815291530
1531
1532
153315341535153615371538153915401541154215431544
154515461547154815491550155115521553155415551556
79
8081
Correlate changes in other parameters, especially water vapour (see below), with changes in temperature.
Provide a reference against which satellite-based temperature measurements can be calibrated and adjusted so that long-term changes can be estimated globally with greater confidence.
Validate temperature trends simulated by climate models. Provide input to global meteorological reanalyses such as NCEP, ECMWF, NASA, JMA. Provide input to numerical weather prediction models if and when submitted shortly after the
measurement. Upper-air measurements of temperature and relative humidity are two of the basic measurements used in the initialization of numerical weather prediction models for op-erational weather forecasting. In turn, feedback from the numerical analysis potentially provides a useful metadata element in the final GRUAN measurement (see Section 9).
There are a number of satellite-based measurements of this ECV. Stratospheric Sounding Unit (SSU) instruments have been flown on TIROS-N and NOAA-6 to NOAA-14 operational meteor-ological since 1979. Three SSU channels with weighting functions of 10-15 km width, peak at 29, 38 and 44 km providing sampling of the middle and upper stratosphere until October 2005. SSU provides the only near-global source of data on temperature trends above the lower stratosphere over such a long period; it has been extensively used in assessments of those trends, and their pos-sible causes (Randel et al., 2009). Global satellite observations from the Microwave Sounding Unit (MSU) Channel 4 and the Advanced Microwave Sounding Unit (AMSU) channel 9 provide a weighted layer mean temperature of the lower stratosphere between 13 and 22 km. This weight-ing primarily covers the stratosphere in the extra-tropics, but spans the upper troposphere and lower stratosphere in the tropics (the weighting function peaks near the tropical tropopause). The MSU/AMSU time series are derived by combining measurements from the series of satellite in-struments that have been operational over 1979–2007. Since the mid-1990s, radio occultation (RO) measurements of temperature have also become available (Steiner et al., 2009).
However, none of these satellite-based measurements are likely to extend deep down into the tro-posphere and so radiosonde measurements are likely to remain the primary data set for trend de-tection in this region. Recent research has shown that the RO technique has the potential to provide high resolution profiles of atmospheric refractive index in the middle to lower tropo-sphere, which combine the effects of temperature and water vapour in this region. The require-ments for random error, bias and long-term stability are detailed below and are guided, in part, by the needs of end-users and in particular the use of the measurements in detecting trends in tem-perature time series which include natural, unforced climate variability. This becomes a signal-to-noise ratio problem and climate models can be used to guide the measurement requirements given expectations of future trends in temperature and natural variability (see e.g. Figure 10.7 of IPCC 4th assessment report). { Ed Note: This isn’t true as stated and is best left out}
It is particularly important that trends in tropical cold point tropopause temperatures are accur-ately detected since this is thought to control the flux of water vapour into the stratosphere (Gettelman et al., 2002; Fueglistaler and Haynes, 2005) and changes in stratospheric water vapour influence radiative forcing and temperatures both in the lower stratosphere but also in the upper troposphere (Forster et al., 2007). At present temperature trend uncertainties in the lower strato-sphere and upper troposphere remain large, particularly in the tropics. For this ECV, addressing trends in tropical cold point temperatures should be a focus for GRUAN. To this end establishing close working ties between the tropical GRUAN sites at Manus and Nauru with the sites within the SHADOZ network (Thompson et al., 2007) and with the GUAN stations operating in the tropics would be particularly advantageous.
41
15571558155915601561156215631564156515661567
15681569157015711572157315741575157615771578157915801581
15821583158415851586158715881589159015911592
1593159415951596159715981599160016011602
8283
Measurement range: Ideally temperature measurements should cover the range 170 – 350 K to span the range of measurements encountered between the Earth’s surface and the upper strato-sphere. Currently available technology can meet this requirement.
Vertical range: The effects of elevated concentrations of greenhouse gases on atmospheric tem-peratures are seen most clearly in the upper stratosphere (Shine et al., 2003). Ideally GRUAN measurements of the vertical temperature profile should extend from the surface to ~50 km. Ver-tical temperature profiles are most routinely measured using radiosondes which seldom reach above ~35 km altitude (noting that radiosondes flown to provide input to NWP models aim only to reach ~25 km). However, if used to provide a reference standard for temperature over the lower portion of GPS-RO satellite-based measurements of the vertical temperature profile, and then if combined seamlessly with those satellite-based measurements, the goal of achieving refer-ence quality measurements from the surface to the stratopause (and even higher) would be achieved. Ideally temperature profiles from the surface to the upper stratosphere/lower meso-sphere, measured by a single instrument, should be the GRUAN goal since these would provide the most robust signal of climate change. Use of GRUAN radiosonde temperature profiles as a standard for other GUAN stations would increase the geographical coverage in the troposphere.
Vertical resolution: Given that it is primarily balloon-borne instruments that provide high resolu-tion profiles of the vertical temperature profile in the atmosphere, a resolution of 100 m or better below 30 km altitude and a resolution of ~500 m above 30 km altitude is appropriate.
Random error: ≤0.2 K in measurement repeatability.
Bias: Biases ≤0.1 K in the troposphere and ≤0.2 K in the stratosphere. This is significantly more stringent than the 0.5 K in the troposphere and 1 K in the stratosphere prescribed in the Guide to Meteorological Instruments and Methods of Observation (WMO-No. 8) and In the dark this re-quirement can probably be met by several of the better operational radiosondes but not in the day-time, see WMO (2011) and the revision of chapter 11 in the CIMO Guide, published in 2012. The most accurate radiosonde in the day is possibly is currently unrealistic since the perhaps most accurate temperature sonde, the ‘Accurate Temperature Measuring Radiosonde’(Schmidlin, 1991), claims claiming an uncertainty of 0.3K throughout most of the upper troposphere and the stratosphere, but this is not yet widely available in sufficient numbers for use throughout GRUAN..
. This suggests that while Thus, GRUAN should proceed with the best technology operational ra-diosondes available, using the methods of observation agreed with the GRUAN Lead Centre, en-suring that sufficient sites make a priority of temperature measurements in the dark.available, em-phasis also needs to be placed on the d Development of commercially availabe new technology to achieve higher accuracaccuracy in the daytime is a priorityy. It might be that reduced bias is achievable from night-time soundings where the radiation correction, the dominant source of un-certainty in the stratosphere (Immler et al., 2010), is significantly reduced. For temperature, redu-cing measurement bias may be partially accomplished by using a three-thermistor set with differ-ent radiative properties (e.g. white, black and silver) to quantify the uncertainty in the radiation correction which is the largest source of measurement bias towards the top of the flight.
Stability: Better than 0.05 K/decade. The signal of change over the satellite era is in the order of 0.1–0.2K/ decade requiring long-term stability to be an order of magnitude smaller to avoid ambi-guity.
42
160316041605
16061607160816091610161116121613161416151616161716181619
162016211622
1623
1624162516261627162816291630163116321633
1634163516361637163816391640164116421643
164416451646
8485
4.2.2. Water vapour
Scientific justification: Water vapour is the primary natural GHG and is central to global water and energy cycles. It acts primarily as a feedback, amplifying the effects of increases in other GHGs. Water vapour is the raw material for clouds and precipitation, and limited knowledge has compromised our ability to understand and predict the hydrological cycle, and understand its ef-fect on radiative transfer (Peter et al., 2006). Water vapour is also a source of OH in the upper tro-posphere and stratosphere, influencing methane, ozone and halogenated GHGs. High clouds due to water vapour in the UT/LS affect both the planet's shortwave albedo and its longwave green-house effect, and both cloud particles and water molecules are involved in chemical reactions that govern stratospheric ozone concentrations. Fully quantifying the Earth’s radiation budget depends on an accurate assessment of the radiative properties of clouds and the water vapour continuum.
For weather forecasting, boundary layer and lower tropospheric humidity measurements (or total column water vapour, which is dominated by the lower troposphere) are of primary interest. How-ever, cChanges in water vapour in the UT/LS exert a greater radiative forcing than changes else-where (Solomon et al., 2010). Unfortunately sStandard radiosonde humidity sensors have had very poor response at the low temperatures [below -50 ºC], pressures, and water vapour concen-trations of the UT/LS (Wang et al., 2003) .However, there has been a lot of progress since 2003, see the results of the WMO Intercomparison of High Quality Radiosonde Systems, WMO(2011), although no operational radiosonde can be expected to measure with sufficient accuracy for cli-matological purposes ( see GCOS-112) in the lower stratosphere.
. A number of factors, many linked to changes in climate, are likely to affect the flux of water va -pour into this climatically important region of the atmosphere, viz.:
i) Changes in the cold-point tropopause temperature (Zhou et al., 2001).ii) Changes in convection. Convective transport of ice particles into the UT/LS can provide a
path with bypasses the limitation imposed by the cold-point tropopause temperature.iii) Changes in the Brewer-Dobson circulation (Austin et al., 2006).
While most of the Earth’s water vapour is contained in the lower atmosphere where it is can be measured as absolute or relative humidity, the water vapour content of the upper atmosphere is measured in parts per million and is difficult to measure accurately; the current older generation of operationally-deployed balloon-borne instruments, and the satellite data record to date do did not allow the measurement of water vapour in the upper troposphere and lower stratosphere to the required accuracy to be useful for climate applications (Soden et al., 2004). However, accurate water vapour measurements in the upper atmosphere are critical, especially for radiative transfer modelling. Understanding the water vapour budget throughout the atmosphere is also necessary for interpreting measurements of outgoing longwave radiation.
Satellite-based solar occultation and limb-sounding instruments can measure water vapour in the upper troposphere and stratosphere but inter-satellite differences preclude the use of these earlier data in long-term trend analyses (Rosenlof et al., 2001). High precision measurements of water vapour profiles will provide valuable input data to global meteorological reanalyses and data for validating global climate models.
Instruments such as the Cryogenic Frostpoint Hygrometer (CFH; Vömel et al. 2007b), the Fluor-escent Advanced Stratospheric Hygrometer for Balloon (FLASH-B) Lyman-alpha instrument, can provide water vapour measurements in the lower stratosphere, but are very expensive compared to operational radiosondes. , theThe Snow White chilled mirror hygrometer is able to measure re-liably in the upper troposphere at night. All these instruments require a much higher skill level to to ensure reliable operation than an operational radiosonde. It does not appear justified to expect
43
1647
1648164916501651165216531654165516561657
165816591660166116621663166416651666
166716681669167016711672
167316741675167616771678167916801681
16821683168416851686
1687
168816891690169116921693
8687
every GRUAN site to fly these systems once a month, where several GRUAN sites are in the same climate region, e.g. western Europe. The variability of water vapour in the stratosphere over a given climatic region is not expected to be high, but as indicated above, a priority should be given to measurements in the tropics, when resources are available.,
or the Vaisala RS92 (Suortti et al., 2008) or RS-90 FN (Leiterer et al. 1997), may be used for reference measurements in their respective, valid altitude range. Other proven reference instru-ments may be introduced, with careful attention to data continuity concerns.
Modern operational radiosondes have much improved performance compared to those reported earlier and there has been a significant improvement between the WMO Radiosonde Intercompar-ison hosted in Mauritius in 2005 and that hosted in Yangjiang China in 2010, see WMO(2011). The better sensors now start to become slow to respond at temperatures around -70ºC. A second source of error comes from assuming the temperature of the humidity sensor in the day is the same as that reported by the radiosonde temperature sensor. However, adjustment algorithms for this slow response have been implemented and methods of reducing the solar heating error have been implemented, so the relative humidity errors in the tropical upper troposphere are very much smaller than in earlier operational radiosondes. The use of the better operational radiosondes in GRUAN will improve the capability to monitor changes in the upper troposphere day to day, al-though further development of the systems should be encouraged.
Many sites are currently developing the capability to observe and analyze data from ground-based GPS receiver, usually as part of a larger or international networks. These data provide continuous high-quality estimates of column water vapour which, in addition to being useful in their own right, can be used to partially validate the vertical humidity profile measurements; total precipit-able water calculated from the radiosonde measured temperature and humidity profiles should compare well with that measurement by the GPS receiver.
Measurement range: 0.1 – 90000 ppm. The large range in values that needs to be covered by these measurements presents a challenge for instrument development and operation since no single commercially available instrument is responsive over this range. Instrument packages may therefore need to include more than one instrument, each of which covers a particular region of the atmosphere.
Vertical range: 0 to ~40 km.
Vertical resolution: 50 m below 5 km and 100 m above 5 km altitude.
Random error: 2% in mixing ratio in the troposphere and 5% in mixing ratio in the stratosphere.
Bias: 2% in mixing ratio throughout the profile. 1% for total column. This is more stringent than the 53% standard prescribed recommended in WMO-No. 8 by CIMO as currently achievable..
Stability: 0.3%/decade in mixing ratio and for the total column.
4.2.3. Pressure
Scientific justification: Measurements of upper air temperature and water vapour must be accom-panied by the height/ pressure at which the measurement is made. Data used by WMO, e.g. in nu-merical weather prediction primarily use standard geopotential heights, but conversion from geo-metric to geopotential height is straightforward, see WMO Guide to Observations(WMO No.8). Thus, to convert from geometric height as measured for instance by a GPS radiosonde, to geopo-tential height does not require knowledge of the vertical temperature structure.
44
1694169516961697
1698
169917001701
17021703170417051706170717081709171017111712
1713
171417151716171717181719
17201721172217231724
1725
1726
1727
17281729
1730
1731
173217331734173517361737
8889
In most numerical weather prediction models the observational data are input at levels defined by a ratio of the pressure to the surface pressure. Thus, the model must convert geopotential height into pressure, if the system does not provide pressure observations. Deducing pressure from the geopotential height requires knowledge of the temperature and water vapour structure in the ver-tical, and if this is not directly available from the system, the model will compute the values using its own analysis fields..
If a radiosonde uses a pressure sensor to compute geopotential height, the Accurate measurements of pressure from the surface to the upper stratosphere are necessary for relating measurements made in different vertical coordinates e.g. radiosonde (pressure) and lidar (geometric height) measurements, or model output which is often provided with geopotential height as the vertical coordinate. Uuncertainty in calculated geopotential heights will result from uncertainties in tem-perature, pressure and water vapour measurements. However , the majority of modern radio-sondes are now using GPS navigation signals to measure height and when set up carefully can meet all the GRUAN requirements for pressure/height observations. The uncertainty in the GPS height measurements has very little variation with height in the atmosphere, see WMO (2011). So CIMO recommend that GPS radiosonde are used at all GRUAN stations. Thus , for instance problems with reproducing structure in the vertical in the stratosphere become negligible with GPS radiosondes, e.g. with ozone profiles . The extent to which calculated geopotential heights/geometric heights agree with GPS derived altitudes can provide an indirect validation of the ac-curacy of the temperature, pressure and water vapour measurements. If pressure measurements drift in the presence of a steep vertical gradient in some target trace gas, this will alias into an ap-parent trend in that trace gas. It is therefore essential that pressure profile measurements maintain long-term stability.
Measurement range: 1 – 1100 hPa
Vertical range: 0 – 50 km
Vertical resolution: 0.1 hPa[ better quoted in equivalent height]
Random error: 1 km 0.011 hPa , [equivalent height error, 10m]
16km 0.3 hPa, [equivalent height error, 20m]
32 km 0.05 hPa , [equivalent height error, 30m]
48 km 0.01 hPa, [equivalent height error,50m]
{Ed NNote: the requirements as stated previously, impose far too high a requirement on pressure measurements near the ground.}
Bias: 0.1 hPa. Half the random error quoted above.
This is more stringent than the 1 hPa to 2 hPa in the troposphere and 2% in the stratosphere re-quirements listed in WMO-No. 8.
Stability: Better than 0.1 hPa/decadea quarter of the random error quoted above, per decade..
4.3 Moving beyond priority 1 variablesThis section is being developed in a separate document which will then be merged with the manual. The content from the previous version of the manual is provided below. The final version of this section will be far more comprehensive.
45
1738
173917401741174217431744
174517461747174817491750175117521753175417551756175717581759176017611762
1763
1764
1765
1766
1767
1768
1769
17701771
1772
17731774
1775
1776
177717781779
9091
The emphasis to date within GRUAN has been on observations of priority 1 variables. This allows testing of the guiding principles for all reference observations before expanding the measurements at GRUAN sites to lower priority variables. A fully functioning GRUAN that serves all envisaged purposes will require measurements of all ECVs listed in Appendix 1 of GCOS-112. An approach to expanding site measurement capabilities to eventually cover as many of the specified variables as possible, whilst recognising that not all variables may be observed at all stations, is required.
46
178017811782178317841785
9293
5 GRUAN SITES
This section is being developed in a separate document which will then be merged with the manual once approved by the WG-ARO and 66% of GRUAN sites. The content of the latest ver-sion of that document is provided below. It must be emphasized however that this section is very much a work in progress.
5.1 IntroductionThe purpose of this chapter is to define the process by which sites are assessed for GRUAN certi-fication and the process by which that certification is maintained. Certification is essential to en-sure that the sites within GRUAN operate at a level that maintains GRUAN’s status as a premier upper air climate monitoring network (Seidel et al., 2009). GRUAN is more than a collection of measurements made at individual sites. Part of the scientific benefit that will accrue from GRUAN results from the homogeneity of the reference quality standard of the measurements made at the sites comprising the network. A shortfall in maintaining that quality standard at one site reduces the confidence that users will have in the utility of the measurements made across the network as a whole. Sites therefore need to be sufficiently consistent and scientifically sound in their operation for scientific benefits to accrue. The site certification process assures all sites that they are operating to reference quality standards, that all sites in the network are operating to the same standard, and that the scientific benefits resulting from such homogeneity of quality across the network are guaranteed. This document provides pragmatic criteria that will be used to assess and certify existing sites as well as new site offers. These criteria are designed to be as transparent as possible and to minimize the overhead involved for all parties in the certification process.
Specifics regarding the site assessment and certification include:1. Site assessment and certification falls within the mandate of the Working Group on Atmo-
spheric Reference Observations (WG-ARO) together with the GRUAN Lead Centre. As-sessment and certification of sites within GRUAN is consistent with the guidance de-veloped with the WMO Commission for Instruments and Methods of Observation (CIMO; WMO-No. 8) and the WMO Commission for Basic Systems (CBS; WMO-No. 488).
2. Sites seeking to become GRUAN sites will first be assessed according to their ability to meet the mandatory operating protocols defined in Section 5.2 and then according to the added value they bring to the network, as defined in Section 5.3.
3. Sites will propose specific measurement programmes for inclusion in GRUAN and it is these that will be required to conform to the operating protocols defined in Section 5.2 and which will form the basis for assessing the added value that the site brings to the network as a whole. This will enable sites to operate some, but not necessarily all, measurement programmes to GRUAN standards.
4. Determining whether the operating procedures for proposed measurement programmes meet the prescribed operating protocols will be done objectively against the standards out-lined in Section 5.2.
5. While based on a body of sound scientific research, in assessing the value which a specific site adds to the network, the WG-ARO will exercise its discretion in evaluating the pro-posal against the criteria defined in Section 5.2.
6. The Lead Centre and WG-ARO will provide written feedback to each site as part of the certification process.
47
17861787
1788178917901791
1792
179317941795179617971798179918001801180218031804180518061807
1808180918101811181218131814181518161817181818191820182118221823182418251826182718281829
9495
7. To identify potential issues early, sites will be reviewed annually based on their annual re-ports (see below) which must highlight any operational anomalies, and based on reports on data flow, site performance etc. from the Lead Centre. Such reviews will be ‘by exception’ reviews. More complete site audits will be undertaken every 3 to 4 years (see Section 5.5 below for more details).
Where site reassessments identify measurement programmes that consistently fall short of GRUAN operating standards, GRUAN certification of that programme will be suspended. If all measurement programmes at a site lose their GRUAN certification and where jointly developed recovery plans for the measurement programmes at the site have repeatedly failed to resolve out-standing issues, the site will be suspended from the GRUAN network. The WG-ARO and Lead Centre will work pro-actively with sites to remedy these issues wherever possible in a timely and cost-effective manner.
5.1.1. Geographical coverage
It is not necessary that GRUAN provides globally complete and spatially homogeneous coverage - rather GRUAN should provide a reference anchor for other ground- and satellite-based networks which would then provide the required global coverage.. However, it would be advantageous if GRUAN should aim to sample as many as possible could sample allof the major climatic regimes and environment types to ensure that different temperature and radiation environments are reliably calibrated. Expansion of the network should concentrate on climatic zones and regions that are under-sampled in the initial network configuration. Geographical coverage of GRUAN sites should also be tailored to meet the specific needs of end-users e.g. the satellite-based measure-ment community is likely to want validation data in key regions of the atmosphere, with relatively uniform surface properties surrounding the site.
5.1.2. Funding
Candidate GRUAN sites able to demonstrate reasonable expectations of funding to maintain oper-ations over many decades will be valued above sites that cannot. The WG-ARO should have in place procedures for supporting long-term funding applications to local funding agencies for sites seeking to join GRUAN. At present most national funding agencies are challenged by require-ments for funding over multi-decade timescales. Thus realistic aims must be set for GRUAN to ensure operating costs are kept as low aspossible.. Thus it is sensible to have a sizeable propor-tion of the GRUAN sites where the facilities are already mostly funded for weather observation-s.This may therefore requite higher level (GCOS) education of national funding agencies for maintenance of the global climate observing system. Having sites transiently joining and then leaving the GRUAN network could compromise the goal of ensuring data homogeneity across the network e.g. if trends in ECVs differ at two different stations which measured the ECV over dif-ferent time periods it is not clear whether the differences arise from the geographical separation or from different time periods being sampled.
5.2 Mandatory Operating ProtocolsGRUAN is a heterogeneous network that includes sites from both the research community and the operational meteorological community. The mandatory requirements for sites reflect GRUAN’s primary goal of providing reference quality observations of the atmospheric column (see Section 2). The emphasis is on how the measurements are made rather than specifically on what measure-ments are made. These requirements define GRUAN’s unique nature while accommodating the diverse capabilities of sites within the network. These protocols also recognize that GRUAN is not the sole stakeholder at any of the sites. Recognizing these boundary conditions, sites shall:
48
18301831183218331834
1835183618371838183918401841
1842
1843184418451846184718481849185018511852
1853
1854185518561857185818591860186118621863186418651866
1867
1868186918701871187218731874
9697
1. Provide reference quality observations as defined in Section 2. In particular every meas-urement must be traceable to fundamental standards and calibrations through well docu-mented routes.
2. Provide uncertainty estimates for each datum or collaborate with other sites, instrument developers, GRUAN Task Teams and the GRUAN Lead Centre to provide these estimates in a consistent manner for a given instrument across the network. Profile measurements re-quire uncertainty estimates for each measurement point on the profile. Documentation de-scribing the calibration methods applied to each measurement, and the sources of measure-ment uncertainty excluded and included in the uncertainty estimate, must be provided.
3. Provide access to raw data and assure long-term storage of the raw data either at the site, another GRUAN facility, or at another internationally accessible archive.
4. Provide complete metadata for each measurement as defined in the requirements document developed by the Lead Centre. Metadata need to be sufficient to allow reprocessing of raw data by an independent party and will depend on the measurement system employed.
5. Provide regular traceable ground checks, independent of the manufacturer, for any instru-ments which provide vertical profiles extending from the surface.
6. Provide regular calibration information about the measurement systems (in-situ and remote sensing).
7. Provide redundant reference observations of the essential climate variables (ECVs; GCOS-138) selected for measurement at the site at intervals sufficient to validate the derivation of the uncertainty on the primary measurement (noting that this validation is generally achieved through comparison against other recognized reference observations).
8. Provide annual reports summarizing GRUAN operations at the site, the extent to which standard operating procedures developed for the network as a whole have been adhered to, changes in instrumentation, how those changes were managed, improvements made etc. Present these reports at the annual network meeting.
9. Conduct measurement programmes with an operational philosophy of continually Selected GRUAN sites may striving strive to improve measurement accuracy. , Actively actively conduct conducing research through intercomparisons, laboratory studies, work with other GRUAN sites and/or cooperation with manufacturers to improve measurement accuracy.
10. Manage change pro-actively as defined in Section 2.3.11. Have a representative on the task team of site representatives and a nominated reserve con-
tact for GRUAN purposes.12. Actively communicate with the Lead Centre, WG-ARO, Task Teams and/or other sites,
(e.g. through attendance of meetings, blog postings etc.).
These mandatory operating protocols do not replace the target requirements defined in GCOS-112 and GCOS-121, which remain the targets for GRUAN, but rather emphasize the importance of how the measurements are made, and in particular what is required to guarantee reference quality observations, rather than what physical measurements are made.
5.3 Criteria for Assessing Added ValueOnce a site has committed to operating a set of measurement programmes under the protocols defined in Section 5.2, the added value that a site brings to the GRUAN network will be a func-tion of:
1. The extent to which the site measurement programmes provide measurements in regions, or of atmospheric phenomena, which were not previously sampled. In this case, the added
49
18751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909
1910191119121913
1914
19151916191719181919
9899
value will be a function of the locations and capabilities of the sites already participating in the network.
2. The extent to which a site brings unique observational and/or analysis capabilities to the network as a whole and the likelihood of being able to propagate those capabilities across other sites in the network.
3. The participation of that site in other international measurement networks (e.g. GUAN, GAW, NDACC, BSRN). Participation in existing networks reduces start up costs for es-tablishing the GRUAN site and it quantitatively links the GRUAN measurements to the measurements being made in those other networks
4. The extent to which a site can commit to a multi-decade programme of measurements. While it is recognised that a multi-decade programme of measurements cannot be guaran-teed, a statement of intent with documented support (e.g. from the host institution or relev-ant funding agency) will add to the assessment of the value that the site brings to the net-work.
5. 6. The extent to which a site can fulfil the following measurement programmes expected of a
candidate GRUAN site:a. At least twice daily (00 and 12 LST as a preference over UTC2) measurements of
vertical profiles of temperature from the surface to ~30 km and water vapour in the troposphere. High quality surface measurements of these same variables are also re-quired to provide a traceable link between the measurements at the lowest level of each profile. Where feasible, occasional soundings at both 00/12 LST and UTC could be used to establish climatologies of differences, including uncertainties, which could thereafter be used to relate measurements made at one standard time to measurements made at another.
b. At least monthly observations of stratospheric water vapour to ~30 km, at nominated sites.. Given that high frequency natural variability in the lower stratosphere is relat-ively small, these profile measurements should be made when most practical and when the altitude coverage can be maximized (see Section 7.1.4).
c. d. Hourly observations of integrated precipitable water vapour.
These high priority measurement programmes will be refined as the research which forms their basis progresses. At such a time this section of the manual will be updated to reflect these changes.
Achieving each of these measurement programmes is not mandatory for the inclusion of a site in GRUAN. However, the extent to which a site can meet these requirements will de-termine, in part, the additional value that that site brings to the network. For sites unable to make twice daily measurements, a compromise would be to make one weekly radiosonde measurement of temperature, pressure and humidity using the usual radiosonde used at the station but to fly this together with a second sonde, either from another manufacturer (to test network homogeneity) or from the same manufacturer (to test repeatability). Both ap-proaches would assist in validating measurement uncertainty which is equally important for these measurements (see Section 6.2).
While weekly sampling significantly underestimates monthly standard deviations in tem-perature, differences between detectable trends for weekly sampling compared to 12
2 00/12 UTC observations are no longer as important for NWP since 4D data assimilation is now more common. Where higher priority considerations require sites to measure at 00/12 UTC rather than 00/12 LST, this will not count against the site.
50
1920192119221923192419251926192719281929193019311932193319341935193619371938193919401941194219431944194519461947194819491950
195119521953
195419551956195719581959196019611962
19631964
100101102
103104
hourly sampling may be acceptably small (Seidel and Free, 2006). So, for example, a site that makes weekly reference quality radiosonde measurements of temperature, pressure and humidity in a large region of the globe containing no other GRUAN stations might be assessed as adding as much value to the network as a site making 12 hourly reference qual-ity measurements but located very close to another site making the same measurements, but operational practice suggests weekly measurements by non-experts does not give reli-able results, usually..
7. The extent to which the measurement programmes promoted for inclusion in GRUAN meet the requirements for random error and bias as defined in Section 4.2.
8. The extent to which the site measurement programmes provide measurements in regions, or of atmospheric phenomena, which were not previously sampled. In this case, the added value will be a function of the locations and capabilities of the sites already participating in the network.
9. The extent to which a site brings unique observational and/or analysis capabilities to the network as a whole and the likelihood of being able to propagate those capabilities across other sites in the network.
10. The participation of that site in other international measurement networks (e.g. NDACC, GAW, BSRN). Participation in existing networks reduces start up costs for establishing the GRUAN site and it quantitatively links the GRUAN measurements to the measurements being made in those other networks
11. The extent to which a site is prepared to forgo locally established operating procedures and adhere to the standard operating procedures established by the Lead Centre and adopted by the majority of the sites already in the network. Where sites are unwilling or unable to do this, it would diminish the added value that site would bring to the network as a whole.
12. The availability of historical measurements that conform to the GRUAN standard. All else being equal, a site that extends an existing multi-decadal time series of reference quality measurements will be assessed as adding more value to the network than a site that would initiate the same measurement programme starting at the present. Detailed documentation describing how changes in standard operating procedures, instruments, data processing al-gorithms and operators over the history of the measurement programmes have been man-aged to ensure that the historical measurements are reference quality. Where historical ref-erence quality measurements are available, consideration will be given by the WG-ARO and Lead Centre to providing these as GRUAN data through the GRUAN data archives.
13. The extent to which a site can commit to a multi-decade programme of measurements. While it is recognised that a multi-decade programme of measurements cannot be guaran-teed, a statement of intent with documented support (e.g. from the host institution or relev-ant funding agency) will add to the assessment of the value that the site brings to the net-work.
14. The extent to which a site can provide redundant observations of the priority 1 variables (see Section 4.2) or can conduct periodic intercomparisons of a large range of instrument types.
15. The extent to which a site is capable of measuring other ECVs identified in GCOS-112 as being desired quantities.
16. The level of institutional support for the site and commitment to maintaining long-term reference quality measurement programmes. If, in addition, a site can demonstrate that it is actively pursuing resources to enhance the site’s capability, such as the addition of new measurement programmes, this would also enhance the added value the site would bring to the network. It is also desirable that there is full host institution commitment to GRUAN-
51
196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012
105106
related activities and that this commitment is not dependent on a single Principal Investig-ator.
17. . The level of institutional support for the site (and any partner institutions) to undertake fundamental scientific research of the measurements from the site and other GRUAN sites. Because GRUAN includes aspects of both operational and research networks, a critical mass of scientific analyses is required on a continuing basis to ensure that GRUAN fulfils its role as a research network. Those making the measurements are arguably in the best po-sition to analyze them.
18. The degree of historical or planned cooperation with other sites both within and outside the GRUAN network including other GRUAN-relevant networks. It may be the case that while a single station might not be able to provide the full range of ECV measurements re-quired by GRUAN, a group of two or more stations, located sufficiently close together, might have the combined capability of providing the full range of measurements. Such a collection of stations may then act as a single GRUAN site. Robust analysis would be re-quired to demonstrate the extent to which such a cluster of sites might act as a single GRUAN station (see Section 6.5)
Such assessments of added value rely on the expert judgement of the WG-ARO and Lead Centre, recognize the heterogeneity of the sites within the network, and facilitate a practical approach to expansion of the network following the 2009-2013 implementation phase for GRUAN (GCOS-134).
52
2013201420152016201720182019202020212022202320242025202620272028
2029203020312032
107108
5.4 The Assessment and Certification ProcessA schematic of the site assessment and certification process is provided in Figure 2. Proposals for
the addition of new sites to GRUAN are likely to happen through two possible routes, viz.: Sites being approached by the WG-ARO and/or Lead Centre and invited to become GRUAN
certified sites. Sites being proposed externally e.g. through the National Weather Service of the host country
to the Lead Centre or WG-ARO.
53
Figure 2: A schematic representation of the site assessment and certification process.
2033
2034
20352036203720382039
109110
Once a site has been identified for possible inclusion in GRUAN, through either of the routes lis-ted above, the following sequence of events will be used to assess the site for potential GRUAN certification:
1. Provision of the GRUAN requirements document (the external document comprising this chapter), the GRUAN manual, guidelines for the operation of specific instruments in wide-spread use in GRUAN, as well as documentation describing data submission protocols and the procedures that must be followed when data are submitted to the internal GRUAN archives, to the candidate site by the Lead Centre.
2. The response from the candidate site should include:a. A list of the measurement programmes at the site proposed for inclusion in GRUAN.
This need not necessarily include all measurement programmes at the site. If a new or existing measurement programme is later proposed for inclusion in GRUAN, a similar procedure to that defined here will be used to include that programme in the GRUAN certification for that site.
b. A complete description of how those measurement programmes will be conducted. Such information would include, for example, detailed standard operating proced-ures for each of the measurement programmes, including a description of data stor-age policies, and a description of how systematic and random uncertainties in the measurements will be derived and reported. This information must be sufficient to establish the ability of the site to meet the mandatory operating protocols detailed in Section 5.2.
c. For measurement programmes for which a GRUAN data product has not yet been well defined, the site must describe their intended strategy for developing the exist-ing observational product into a GRUAN data product that fulfils the mandatory op-erating protocols defined in Section 5.2. In such instances, cooperation with other sites already in the network is highly desirable to ensure that this expertise is dissem-inated to similar measurement programmes in operation at other sites.
d. The management structure of the site and a general description of the manner in which the site is operated. This would include a description of current and expected future funding levels for ongoing operation of the site.
e. A description of which databases the measurements have previously been submitted to and are currently being submitted to.
f. A description of how measurements to date at the site have been processed. This will be used to assess whether the time series to date meet the standards for a GRUAN reference measurement. Particularly important in this regard will be detailed docu-mentation around how changes in standard operating procedures over the history of the measurement programmes have been managed to derive a homogeneous time series of measurements. Since the historical database of measurements is an import-ant aspect for assessing the added value that a site brings to the network (see Section 5.3), it is particularly important that the historical data can meet the stated GRUAN requirements for long-term homogeneity.
g. A list of the scientific experts employed at the site who would likely participate in the analyses of the data collected within GRUAN.
h. Any additional information required to assess the site against the requirements listed in Sections 5.2 and 5.3.
3. There is likely to be some iteration between the Lead Centre and the candidate site to con-firm specific details, fill in information gaps, and finalize the documentation from the can-didate site.
54
204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087
111112
4. Based on the documentation received from the candidate site, the Lead Centre will then write a short recommendation. This, together with the documentation from the candidate site, will then be submitted to the WG-ARO who will evaluate the proposal within 6 calen-dar months against the requirements listed in Sections 5.2 and 5.3. A visit to the site by members of the WG and/or Lead Centre may be required to obtain specific additional in-formation about the measurement programmes slated for inclusion in GRUAN at that site. If accepted, the measurement programmes defined in 5.2a will then be included in the GRUAN certification for the site.
5. Regardless of the outcome, the WG-ARO and Lead Centre will provide written construct-ive feedback to the candidate site outlining strengths and weaknesses of their programme for GRUAN purposes and suggestions as to future improvements for GRUAN operational purposes. This feedback is non-binding but rather intended to provide useful guidance and support to site capability development and retention of current capabilities.
Sites currently within GRUAN will be assessed and certified in a similar manner before GRUAN becomes fully operational at the beginning of 2014.
5.5 Site AuditingCertification of GRUAN sites will not be a single event. Periodic (e.g. every 3-4 years) complete auditing of the measurement programmes included in the GRUAN certification for a site will be conducted to ensure that the programmes continue to meet GRUAN standards. Such an audit may include:
1. A review of annual reports from sites on GRUAN activities.2. A written report from the site – essentially an update of the original report written to initi-
ate the assessment and certification process.3. A site visit by selected members of the WG-ARO and the GRUAN Lead Centre. Such a
visit would include discussions with the researchers responsible for the measurement pro-grammes at the site.
It is important for external perceptions of GRUAN integrity that these audits are conducted by the WG-ARO and Lead Centre and are not based exclusively on annual station reports. In the eventu-ality of identified site issues the following protocols will be followed:
1. Should a measurement programme at an existing GRUAN site show significantly reduced observational capability over more than a year, as evaluated by the criteria listed above, the WG-ARO and Lead Centre will investigate the circumstances at that site, and, if needed, exclude that programme from the GRUAN certification for that site. The WG-ARO and Lead Centre will work pro-actively with sites to resurrect such programmes providing what technical and in-kind support is practical.
2. Should the overall contribution of a site be deemed sufficiently diminished to call into question its continued presence in the network, the site will be informed immediately in writing. The site will be given six months to form, in consultation with the Lead Centre and WG-ARO, a capabilities recovery plan. Should this plan be accepted the site will have no more than two calendar years from its acceptance to implement agreed key aspects. In the eventuality that this is not achieved, the site will be suspended with an invitation to submit anew at such time as issues are remedied.
55
2088208920902091209220932094209520962097209820992100
21012102
2103
2104210521062107210821092110211121122113
211421152116211721182119212021212122
2123212421252126212721282129
113114
6 INSTRUMENTATION
6.1 Instrument selectionThe choice of what instruments should be deployed within GRUAN will not be a onetime de-cision. Periodic review of instrumentation likely to be of use within GRUAN shall be undertaken since instrument technology is constantly evolving. It must also be recognized that not all sites within GRUAN will operate the same instrumentation, e.g. a new site may decide to adopt the most recent technology while a site that has a multi-decade record using an older instrument may decide to continue to use that instrument to potentially avoid introducing a discontinuity in the measurement time series. The emphasis is therefore not on prescribing an instrument, but rather on prescribing the capabilities of an instrument and allowing individual sites to select an instru-ment that achieves those capabilities in the context of the scientific, programmatic, and practical constraints on the site. That said, the fewer the number of instrument types deployed within GRUAN, the more likely network homogeneity will be achieved.
A number of factors should be considered when selecting instruments for use in the GRUAN net-work including (Immler et al., 2010): Instrument heritage: How long has an instrument been in use by the community and for what
purpose? In what other networks is the instrument deployed? How substantial is the body of literature documenting its performance and measurement uncertainty? How widely distrib-uted is the knowledge base that facilitates the instrument’s successful operation?
Sustainability: Are the costs for operating the instrument and the demands on personnel for operating the instrument consistent with the resources available at GRUAN sites? Is the com-mercial demand sufficient, and the technology available, to support the production and use of the instrument for sufficiently long for the expected multi-decade deployment within GRUAN?
Robustness of uncertainty: Is the underlying accuracy claim for the instrument and its result-ant data sufficiently robust i.e. is it likely to be able to meet the accuracy, precision and stabil-ity standards (see Section 4.1) required by GRUAN?
Information content: Are the temporal and spatial resolution, dynamic range, and other char-acteristics of the measurements made by the instrument consistent with GRUAN require-ments?
Manufacturer support: Is the manufacturer committed to a process of improving the perform-ance of the instrument, based on findings made by the GRUAN user community? Is the man-ufacturer prepared to actively participate in instrument intercomparisons? Is the manufacturer willing to disclose the necessary information required to form a fully traceable chain of sources of measurement uncertainty, given that in some cases this information may have to be kept from public display by GRUAN lay so as not to undermine the competitive advantage of the manufacturer? A case in point regarding this last question – Immler et al. (2010) were un-able to adequately assess the radiation correction made in three different radiosondes because the manufacturer would not disclose the correction algorithm applied by the radiosonde soft-ware. For a consistent uncertainty analysis it is imperative that the algorithms used for correc-tions within the data processing software are made publicly available for GRUAN analaysis, but not necessarily for public display. by the instrument manufacturers. Unwillingness for the manufacturer to do so, should count against the selection of that instrument for use within GRUAN.
Site location: Instrumentation may have to differ by climate region. For example, high-latit-ude sites exhibit extremely low water vapour contents in winter compared to equatorial sites. Therefore, instruments such as water vapour radiometers operating at 23.8 and 31.4 GHz,
56
2130
2131
21322133213421352136213721382139214021412142
2143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176
115116
which have limited sensitivity for integrated water vapour amounts below 5 mm, would need to be augmented with more sensitive microwave radiometers operating near 183 GHz.
6.2 Measurement redundancyHaving different instruments at GRUAN sites measuring the same atmospheric parameters will be invaluable for identifying, understanding and reducing systematic errors in measurements. A pro-ject within GATNDOR has been tasked with quantifying the value of redundant measurements and assessing optimal combinations of measurements. If successive reductions in measurement uncertainty with the addition of each coincident measurement from a different instrument can be quantified in a scientifically robust way, this provides a powerful justification for measurement re-dundancy at GRUAN sites. A case study underway within GATNDOR is using vertical profile measurements of temperature and water vapour at the GRUAN sites at Beltsville, Cabauw, Lindenberg, and Potenza to quantify the error reduction resulting from increasing redundancy of measurements. This requires an assessment of the uncertainty of the temperature and water vapour vertical profiles retrieved using each of the considered techniques and then the investigation of possible sensors’ synergies to reduce the uncertainty. The investigation will be carried out focus-ing on the most common instruments at the considered GRUAN sites: for temperature, radiosonde soundings and microwave profilers; for moisture, radiosonde soundings, Raman lidars, mi-crowave profilers, and GPS receivers. The quantification of the value added by complementary observations should be assessed with respect to: Sensor calibration/inter-calibration (here the ARM Value Added Products could be con-
sidered as a model) Identification of possible biases Representativeness of measurements i.e. which horizontal and vertical region of the atmo-
sphere does the measurement represent. Quality control/assurance with a focus on instrument performance in different meteorological
conditions.
As for much of the other research underway to support the operational framework for GRUAN, this is work in progress and the true value of having multiple measurements of the same climate variables at GRUAN stations will become clear in time.
One important factor for GRUAN is that independent measurements of the same (or related) vari-ables should be reported in a consistent way. The cross-checking of redundant measurements for consistency should be an essential part of the GRUAN quality assurance procedures. Since all data are to be reported with uncertainties, a consistency check is, in principle, a straight for-ward task.
6.3 Surface measurementsWhile GRUAN is, by definition, an upper-air network, surface measurements at sites should also be made in such a way that: They are made according to WMO guidelines (WMO-no. 8), including traceability to SI
standards. The CIMO classification for stations should be applied. The surface measurements provide ground-truthing for vertical profile measurements. For ex-
ample, comparisons between ozonesonde measurements of ozone at the surface against a high precision standard provides essential information for quantifying uncertainties in the ozonesonde measurement.
The measurements can, where relevant, constrain retrievals applied to remotely sensed profile data. Some remote sensing instruments that derive vertical profile data from e.g. optimal es-
57
21772178
2179
21802181218221832184218521862187218821892190219121922193219421952196219721982199220022012202
220322042205
22062207220822092210
2211
2212221322142215221622172218221922202221
117118
timation techniques (Rodgers, 2000), are better constrained when a high precision surface measurement is included as input to the forward model used in the retrieval. In some cases re-mote sensing of column amounts of a trace gas can benefit from having collocated surface measurements of that trace gas e.g. as is done in TCCON.
While there are no formal requirements for GRUAN stations to include surface measurements, the guideline is that where such measurements would significantly add to the quality or utility of the GRUAN measurements, these surface measurements should be made.
6.4 Upper-air measurements
6.4.1. In-situ instruments
A discussed in Section 4.2.1, radiosondes will remain the primary workhorse within GUAN for the measurement of vertical profiles of temperature, pressure and humidity. The fact that these in-struments are not recovered has important implications for GRUAN operations, viz.: The instruments must be low cost, and because they are low cost, the sensors on sondes are
unlikely to be the best commercially available. Therefore, certain compromises in systemmeasurement accuracy have to be accepted by users, taking into account that radiosondeThe sensors are not usually the limiting expense in the cost of a modern operational radiosonde and good sensors can be obtained relatively cheaply. The exposure/mounting of the sensors
on the radiosonde is a limiting factor on the performance of many radiosondes, so there is still scope for improvement with the current systems without investment in very expensive re-
placement technology. manufacturers are producing systems that need to operate over an ex-tremely wide range of meteorological conditions.
Maintaining long-term stability in a radiosonde measurement time series is challenging when the instrument being used to make the measurement is discarded after each measurement. Each instrument must be individually calibrated and tied to common calibration standards to ensure long-term stability. It must also be able to retain its performance throughout an ascent, and currently this is probably one of the limitations of the best operational radiosondes where do the systematic bias does not appear stable to 0.1K during an ascent, see WMO (2011). The better manufacturers have managed to eliminate most faults that occur required by pro-duction engineering, but any given radiosonde type has shown small fluctuations in perform-ance with time, when checked on the ground, although these variations in performance during flight may have been minimised by the ground check procedures used.
Where a site may have elected to make only weekly high quality measurements of temperature, pressure and humidity rather than 12 hourly (see Section Error: Reference source not found), it may permit the use of more expensive (and hopefully more accurate) sensors. In this event, greater efforts should also be made to retrieve and refurbish reference quality sondes after use.
6.4.2. Remote sensing instruments
Material to come in here from Task Teams on ‘Ancillary Measurements’ and ‘Measurement schedules and associated instrument-type requirements’
6.5 Instrument co-locationAs discussed in Section 5.3, some of the current GRUAN sites, and many potential sites, consist of instrument clusters spread over some region rather than single compact sites. Some of them are in geographical locations that have complex orography and/or heterogeneous surface characterist-ics. There remain open questions about how physically far apart measurements can be made and still represent a GRUAN site measurement of a single column. Therefore, appropriate collocation
58
2222222322242225
222622272228
2229
2230
2231223222332234223522362237223822392240224122422243224422452246224722482249225022512252
2253225422552256
2257
22582259
2260
22612262226322642265
119120
requirements for variables and instrumentation should be established to ensure the representative-ness of measurements. These considerations should be site and parameter-specific.
6.6 Calibration, validation and maintenance
6.6.1. Instrument calibration
Establishing reliable calibration procedures for the instruments being used within GRUAN, and applying these uniformly across the network, will be an absolute prerequisite for achieving the GRUAN goals. In addition to establishing calibration procedures at individual sites that minimize the uncertainty introduced into the measurement chain (see Section 2.3) and avoid introducing discontinuities into the time series, it is equally important that calibration procedures do not com-promise the goal of achieving homogeneity across the GRUAN network as a whole so that a measurement of a given parameter at one site is directly comparable to a measurement of the same variable at a different site. A guiding principal that will achieve this goal is that when two identical instruments are deployed at two different sites, they shall also use the same calibration procedures, preferably tied to the same absolute standards, and should also employ identical data processing algorithms. While achieving a common data processing for each instrument will be fa-cilitated through processing the raw data at a single central data processing facility (see Sections 2.2.3 and 8.1), the same approach cannot be used for calibration procedures. To this end achieving inter-site homogeneity will may be improved by developing travelling calibration standards which can be taken to different GRUAN stations to be used in on-site calibration or inter-comparisons. A current example of this would be Dobson Spectrophotometer #83 which is used in the NDACC and WOUDC networks to achieve homogeneity across the global Dobson network (see Sections 1.4.1 and 1.4.4). Such travelling standards for ground-checks for radiosondes (temperature and humidity sensor checks) would be particularly valuable.
Traceability to recognized measurement standards (e.g. SI standards) that can be reproduced glob-ally and over long periods of time will be the key component enabling GRUAN to provide refer-ence measurements useful for long-term climate observations. Traceability is a property of a measurement that is manifest by an unbroken chain of measurements back to a recognized stand-ard, with fully documented uncertainty at each step. This then allows a robust calculation of the propagation of uncertainties from the fundamental standard to the final measurement. If common fundamental standards are available across the GRUAN network this will support the goal of achieving coherence across the network.
GRUAN stations shall maintain a “GRUAN site working standard” for each basis unit, e.g. a ther-mometer periodically calibrated to a National Metrology Institute or other accredited agency standard since this ensures traceability to an SI standard. A mechanism shall be implemented to address the compatibility of those systems with the rest of the network that may not be traceable to SI standards.
Use of traceable calibration standards will also aid operators to detect and quantify systematic er-rors in GRUAN measurements (see Section 3.2). Where the final data product of a reference ob-servation depends on ancillary measurements, these measurements must again be traceable to standards. Traceability will also permit the network to incorporate new scientific insights and new technological developments, while maintaining the integrity of the long-term climate record. To achieve traceability, meta-data on all aspects relating to a measurement and its associated uncer-tainty shall be collected. Each station shall maintain accurate meta-data records and provide these to the GRUAN archives. Copies of calibration certificates shall be submitted to the GRUAN meta-database.
59
22662267
2268
2269
2270227122722273227422752276227722782279228022812282228322842285228622872288
22892290229122922293229422952296
22972298229923002301
230223032304230523062307230823092310
121122
The schedule of field recalibration and validation procedures should be drawn initially from ex-perience with a given sensor type, then refined according to the results of laboratory tests and in-tercomparisons. The date and nature of field recalibrations should be included in meta-data, so that if future experiments reveal shortcomings in schedules or methods that were in use, uncer-tainty estimates can be adjusted after the fact to reflect those newly-discovered issues.
6.6.2. Instrument validation
Validation of the instruments used within GRUAN should include well documented and traceable calibration procedures, participation in regular intercomparisons with similar instruments used at other sites and/or intercomparisons with a travelling standard, and operational comparison of un-certainty estimates on the resultant measurements with those from other instruments (see Section 3.1.3). Most sites will likely not have identical instrumentation, with the result that instrument val-idation will likely be site specific. A standard recommendation for the use of redundant instru-mentation and remote sensing instrumentation should be developed to aid site specific, regularly scheduled, instrument validation. The purpose is to make sharing and communication of best practices across sites seamless and continuous.
6.6.3. Instrument maintenance
GRUAN sites are equipped with sophisticated, state-of-the-art instrumentation and should comply with strict requirements of station maintenance, exposure of instruments and calibration perform-ance to avoid degradation of the quality of the measurements. To ensure that the goal of long-term high quality climate records is reached, site scientists who are leading experts for the instruments used at the respective GRUAN sites shall take responsibility for individual instruments operated at the GRUAN site. However, because all maintenance of an instrument can also introduce dis-continuities in measurement series, maintenance shall not be conducted more frequently than is necessary. Maintenance schedules shall developed for all instruments. All maintenance actions on instruments shall be documented as part of the meta-data associated with the measurements made by that instrument.
60
23112312231323142315
2316
231723182319232023212322232323242325
2326
2327232823292330233123322333233423352336
123124
7 METHODS OF OBSERVATION
7.1 Guiding principlesGRUAN is not being established as a network of completely new stations, and many of the initial stations within GRUAN have been in operation in some cases for decades. As a result, sites cur-rently collecting data from different instruments will almost certainly use different data processing algorithms, different instrument pre-checks, different quality control/quality assurance etc.. While the goal is to achieve operational homogeneity with regard to these issues across the network as a whole, any changes in these procedures will need to be very carefully managed (Section 2.3) to avoid introducing discontinuities in the measurement record that would compromise detection of long-term trends. The sites comprising GRUAN will also not be supported by a single funding agency and therefore different levels of financial support at different sites are to be expected. A process for achieving convergence on agreed on operations and maintenance procedures that will be applied across the network therefore needs to be developed. Furthermore, many of the initial sites report to numerous networks and their governance and stated aims differ substantially. It is therefore essential to have in place protocols and agreements, such as a Manual of Operations, in-cluding common quality assurance procedures that allow the required flexibility, whilst maintain-ing the fundamental quality of the observations necessary to meet GRUAN aims.
7.2 Measurement schedulingThis section is being developed in a separate document which will then be merged with the manual once approved by the WG-ARO. The content of the latest version of that document is provided below. It must be emphasized however that this section is very much a work in progress.
7.1.1. Responsibilities
1. WG-ARO shall work with the appropriate task team to define measurement schedules that allow the resultant data products capture all important scales of temporal variability, both for trend analysis and for process understanding. These schedules should be conservative in the early stages of GRUAN, to allow the appropriate task teams to refine their studies, since currently there is a range of opinions of what is necessary.
2. When GRUAN operations have been implemented, the core measurement schedulesThe development of measurement scheduling protocols shall be undertaken by the Measure-ment schedules and associated instrument-type requirements shall be agreed bt the GRUAN Lead centre with individual sites, subject to the agreement of WG-ARO.task team, hereafter referred to in Section 7.2 as ‘the task team’.
3. Subsequent changes to the GRUAN operations at a site will have to be agreed by WG-ARO and then implemented as far as possible, by negotiation between the Lead centre and the GRUAN sites.
4. The task team shall determine measurement schedules for each measurement programme within GRUAN such that resultant data products capture all important scales of temporal variability, both for trend analysis and for process understanding. In cases where over-sampling would allow averaging of measurements to reduce the net random error, and where this is technically feasible, measurement schedules should be set accordingly.
5. Statistical studies shall inform the process of establishing measurement schedules.
61
23372338
2339
234023412342234323442345234623472348234923502351235223532354
2355
235623572358
2359
23602361236223632364
23652366236723682369
237023712372
2373
23742375237623772378
2379
125126
6. Core measurement scheduling shall remain stable unless there is a clear requirement for change, which would then have to be agreed with the relevant GRUAN sites. Measure-ment scheduling protocols shall be revised by the task team annually, or more frequently as required, in response to the changing needs of the end-users of GRUAN data products. Amendments to the GRUAN measurement scheduling protocol shall be submitted by the task team to the WG-ARO before being distributed to GRUAN sites for voluntary imple-mentation.
7. While implementation of the prescribed measurement schedules is not mandatory, sites shall be required to specify in their annual reports which measurement schedules are being adhered toIndividual GRUAN sites will agree which measurements and measurement schedule they can sustain.
7.1.2. Guiding principles
1. Measurements programmes at GRUAN sites are likely to be constrained by more than just the requirements of GRUAN. In recognition of the heterogeneity of the network, the scheduling protocols defined in this document are optional and are therefore in the form of guidelines.may not apply at every GRUAN site, but any deviation from the measurement schedule must be agreed by the GRUAN Lead centre and then accepted by WG-ARO.
2. A sound scientific basis for the measurement schedules detailed in this document are not always available. These schedules are therefore likely to change as more robust scientific analysis on measurement scheduling becomes available.Some evolution of measurement schedules can be expected in the longer term when performance requirements from the network for climate studies become clearer, but changes must be limited in time and agreed by the WG-ARO, and the GRUAN sites.
3. The highest priority is that measurement schedules are established to achieve the four primary goals of GRUAN (see Section 1.2). Measurement scheduling shall be designed not only for the purposes of quantifying variability on a range of timescales and for long-term trend detection, but to fulfil all goals of GRUAN. Where perturbations to schedules would increase the utility of the measurements without compromising the primary goals of GRUAN, measurement schedules should be adapted to meet the needs of other end-users e.g. the timing of a daily measurement may be shifted to coincide with a satellite overpass and in this way provide valuable high quality data for satellite validation; noting however that the satellite and balloon-borne measurements will never be exactly coincident in time and space. Clear protocols shall be established to ensure, to the extent possible, that meet-ing the needs of some users of GRUAN products does not compromise the quality of the data provided to other users.
4. Where possible, measurements schedules for redundant systems should be synchronized so as to avoid sampling biases when combining the measurements into a single data product. Where such synchronization is not possible, measurement schedules should be designed so that data averaging can be used to produce measurement time series that are temporally synchronized.
5. Required measurement schedules may vary regionally and seasonally. In places and sea-sons where the parameter being measured is more variable, measurements should be made more frequently so that the effects of that variability can be accounted for in trend ana-lyses.
6. Measurement schedules should will have beenbe set to permit a statistical separation of the different drivers of changes in the observed variable.
62
2380238123822383238423852386
2387238823892390
2391
23922393239423952396
239723982399240024012402
240324042405240624072408240924102411241224132414
24152416241724182419
2420242124222423
24242425
127128
7. When developing a sampling strategy for a measurement programme, aA first step is will have been to assess how the sampling affects uncertainties on derived monthly means, fol-lowed by a second step which determines how those monthly mean uncertainties affect the statistical robustness of trends derived from those monthly means.
8. Models such as atmosphere-ocean general circulation models or coupled chemistry-cli-mate models have a keywill have a useful role to play in guiding measurement scheduling. They can be used to provide initial estimates of the autocorrelation, the magnitude of vari-ability, and the trend in climate variables and the composition of the atmosphere as a func-tion of location and season. Simulating the effects of a measurement schedule by sampling model output on the same schedule as the measurements can provide an indication of how the proposed measurement schedule is likely to affect the determination of variability on a range of timescales as well as long-term trends. Where model output is not available, ana-lysis of temperatures, e.g. from reanalyses, can often serve as a valid proxy for other cli-mate variables since temperature responds to climate variability in a similar way to many other climate variables. As this work develops it is possible that the initial GRUAN net-work measurement protocols may change..
{For some measurements, scheduling with respect to UTC or Local Solar Time (LST) may be important and may result in conflicting requirements regarding different intended uses of the measurements. For example, scientifically it may be advantageous to have all GRUAN sites making measurements at the same LST (especially for variables that show strong diurnal variations or for instruments that have diurnally dependent biases that we wish to minimize), while for ensuring coincidence with GUAN stations having all measurements made at the same UTC might be more appropriate.
9. Current assimilation schemes used in NWP and reanalysis centres, e.g. 4D-var assimila-tion, are less sensitive to measurements being made at the same time each day compared to assimilation schemes used in the past. Therefore, consistent timing of measurements is not an issue for assimilation into NWP or for reanalysis. If, however, the variable being meas-ured shows a strong diurnal cycle, or if the instrument being used to make a measurement has diurnally varying biases, changing measurement times would introduce additional variability which would need to be accounted for in any analysis in order to avoid sampling bias.}
7.1.3. Factors affecting measurement scheduling for trend detection
For trend detection four factors should guide the development of measurement schedules (Weath-erhead et al., 1998), viz.:
1. The magnitude of the variability. In most cases this will vary as a function of season. Where measurements through a month are sparse, and where monthly means of those measurements will be used in trend analysis, the day-to-day variability within the month will determine the representativeness of the sparse measurements in quantifying the true monthly mean. The variability in the monthly means themselves, or, more precisely, the variability in the monthly residuals after a regression model fit, will also determine the statistical robustness of derived trends.
2. Autocorrelation. This is a measure of the ‘momentum’ or ‘latency’ in the system. When autocorrelation is high, measurements in consecutive time periods are highly correlated. When autocorrelation is low, the signal is very noisy and consecutive measurements are largely independent of each other. Autocorrelation is also likely to vary through the year. If monthly means of measurements are being used in trend detection, the auto-correlation between those monthly means constitutes an important source of uncertainty in the trend
63
2426242724282429
243024312432243324342435243624372438243924402441
2442244324442445244624472448
24492450245124522453245424552456
2457
24582459
2460246124622463246424652466
246724682469247024712472
129130
estimate (Tiao et al., 1990). One advantage of using monthly means for the calculation of long-term trends is that the uniform temporal sampling simplifies the calculation of the autocorrelation in the signal. However, individual measurements may also be used in trend detection and methods are available for determining the autocorrelation in such potentially unequally spaced measurement time series (Bodeker et al., 1998). A clear distinction must be made between: Day-to-day autocorrelation which determines, in part, the likelihood of over-sampling, Day-to-day variability which determines the robustness of the monthly mean when it is
calculated from sparse, isolated measurements through the month, Autocorrelation in the monthly means which determines, in part, the uncertainty on
calculated trends, Variability in the monthly means which also contributes to the uncertainty on calcu-
lated trends.All four of the above are likely to vary spatially and seasonally with the result that optimal measurement schedules are likely to vary between sites and with season. On initiation of a measurement programme, and where the autocorrelation is not known a priori, measure-ments should be made at the highest possible frequency so that a robust seasonal pattern of the autocorrelation can be established. Thereafter, measurement frequency can be relaxed during periods of expected high autocorrelation since momentum in the system being sampled will result in nearby measurements not being independent.
3. The precision of the measurement. When very precise measurements can be made, meas-urement frequencies can potentially be reduced, depending on the extent to which meas-urement precision is a factor in trend detection or in analyses of specific atmospheric phe-nomena. When measurement precision is poor, high frequency sampling is required to re-duce the effects of the random errors on the measurements. Measurement precision can also vary with season as a result of confounding factors (such as surface albedo, humidity and temperature) which vary through the year. The derivation of measurement uncertain-ties within GRUAN, and how these might vary with season, must therefore play a role in determining measurement scheduling.
4. The size of the expected trend to be resolved. For large trends, the signal to noise ratio will be high and measurement frequencies can be reduced (all else being equal). The magnitude of the trend is also likely to be a function of season.
Where the precision of the measurement is a significantly smaller contributor to the uncertainty in the trend estimate than autocorrelation and natural variability, for most mid-latitude locations and for most climate variables, the autocorrelation in the system results in diminishing returns for measurements made at a frequency of higher than every 3 days. On the other hand, for most cli-mate variables measured at mid-latitudes, sampling less frequently than every 10 days signific-antly increases the uncertainty on derived monthly means.
The interplay between the four factors discussed above must be accounted for when planning measurement schedules. It may be that the uncertainty on derived trends is limited by natural vari-ability rather than by the precision of the instrument, in which case more resources should be in-vested in increasing measurement frequency rather than increasing measurement precision. In some cases this may require a cost-benefit analysis where the cost to detect a putative trend of X%/decade (perhaps based on projections from models) over N years is minimized. A cheaper in-strument making a less precise but more frequent measurement might be selected over a more ex-pensive instrument making a more precise but less frequent measurement, since the greater fre-quency leads to detection of the expected trend either in fewer years or at a lower cost. A meas-
64
24732474247524762477247824792480248124822483248424852486248724882489249024912492
249324942495249624972498249925002501
250225032504
250525062507250825092510
251125122513251425152516251725182519
131132
urement strategy might have a greater cost per year than any alternative, but if that strategy can detect a statistically robust trend in fewer years, the net cost may be reduced.
7.1.4. Measurement schedules
Mandatory schedules are not defined for any of the measurement programmes within GRUAN. Rather, once a station has selected the frequency with which measurements will be made, this sec-tion provides guidelines on appropriate timing of those measurements. The frequency of measure-ments at sites will determine, in part, the added value that a site brings to the network (see Section 5.3).
Radiosondes For sites performing four radiosonde flights daily: Flights either at 00, 06, 12 and 18 UTC
or at 00, 06, 12 and 18 LST.
For sites performing twice daily radiosonde flights: One flight at 00LST and one flight between 06LST and 18LST timed to maximize coincidence with any satellite-based instru-ment overpass measuring the same variables. Since satellite-based measurements are more likely to be daytime measurements, the daytime radiosonde launch time is the one which is varied.
For sites performing daily radiosonde flights: One flight at 12LST.why??
For sites performing weekly radiosonde flights: Nominal launch times should be 12LST on the same day of the week, but allowed to vary by up to 48 hours either side to match satellite overpasses.
For sites performing monthly radiosonde flights: Nominal launch times should be 12LST on the same day of the month, but allowed to vary by up to 5 days either side to match satellite overpasses.
Frost point hygrometers {Ed Note: No site is likely to fly a frost point hygrometer more often than once per week, so the measurements will be made at 7 day intervals, but in fact most sites will only be able to make one flight per month, if at all, so the measurements will be once per month. Does this discussion add to this basic fact?}
For sites performing an average of N flights per week i.e. N×52 flights per year: Where the seasonal cycle in natural variability is not yet known, intervals between flights should be constrained by (4/N) < t < (10/N) where t is the interval in days. Once a climatology of the seasonal cycle in natural variability has been determined, during the 5 months of the year exhibiting highest natural variability, intervals between flights should be constrained by (3.5/N) < t < (6.5/N) where t is the interval in days; this should result in a total of N×30 flights through those 5 months. For the remaining 7 months of the year intervals between flights should be constrained by (7/N) < t < (13/N); this should result in a total of N×22 flights through those 7 months.
For sites performing an average of N flights per month i.e. N×12 flights per year: Where the seasonal cycle in natural variability is not yet known, intervals between flights should be constrained by (20/N) < t < (40/N) where t is the interval in days. Once a climatology of the seasonal cycle in natural variability has been determined, during the 4 months of the year exhibiting highest natural variability, intervals between flights should be constrained by (15/N) < t < (25/N) where t is the interval in days; this should result in a total of N×6 flights through those 4 months. For the remaining 8 months of the year, intervals between
65
25202521
2522
25232524252525262527
2528
25292530
25312532253325342535
2536
253725382539
254025412542
2543254425452546
254725482549255025512552255325542555
2556255725582559256025612562
133134
flights should be constrained by (35/N) < t < (45/N); this should result in a total of N×6 flights through those 8 months.
Aerosol sondes For sites performing an average of N flights per week i.e. N×52 flights per year: Where
the seasonal cycle in natural variability is not yet known, intervals between flights should be constrained by (4/N) < t < (10/N) where t is the interval in days. Once a climatology of the seasonal cycle in natural variability has been determined, during the 5 months of the year exhibiting highest natural variability, intervals between flights should be constrained by (3.5/N) < t < (6.5/N) where t is the interval in days; this should result in a total of N×30 flights through those 5 months. For the remaining 7 months of the year intervals between flights should be constrained by (7/N) < t < (13/N); this should result in a total of N×22 flights through those 7 months.
For sites performing an average of N flights per month i.e. N×12 flights per year: Where the seasonal cycle in natural variability is not yet known, intervals between flights should be constrained by (20/N) < t < (40/N) where t is the interval in days. Once a climatology of the seasonal cycle in natural variability has been determined, during the 4 months of the year exhibiting highest natural variability, intervals between flights should be constrained by (15/N) < t < (25/N) where t is the interval in days; this should result in a total of N×6 flights through those 4 months. For the remaining 8 months of the year, intervals between flights should be constrained by (35/N) < t < (45/N); this should result in a total of N×6 flights through those 8 months.
Ozonesondes For sites performing an average of N flights per week i.e. N×52 flights per year: Where
the seasonal cycle in natural variability is not yet known, intervals between flights should be constrained by (4/N) < t < (10/N) where t is the interval in days. Once a climatology of the seasonal cycle in natural variability has been determined, during the 5 months of the year exhibiting highest natural variability, intervals between flights should be constrained by (3.5/N) < t < (6.5/N) where t is the interval in days; this should result in a total of N×30 flights through those 5 months. For the remaining 7 months of the year intervals between flights should be constrained by (7/N) < t < (13/N); this should result in a total of N×22 flights through those 7 months.
For sites performing an average of N flights per month i.e. N×12 flights per year: Where the seasonal cycle in natural variability is not yet known, intervals between flights should be constrained by (20/N) < t < (40/N) where t is the interval in days. Once a climatology of the seasonal cycle in natural variability has been determined, during the 4 months of the year exhibiting highest natural variability, intervals between flights should be constrained by (15/N) < t < (25/N) where t is the interval in days; this should result in a total of N×6 flights through those 4 months. For the remaining 8 months of the year, intervals between flights should be constrained by (35/N) < t < (45/N); this should result in a total of N×6 flights through those 8 months.
Raman lidarMaterial still to come here from experts on this measurement system.
GPS integrated precipitable waterMaterial still to come here from experts on this measurement system.
66
25632564
2565
256625672568256925702571257225732574
257525762577257825792580258125822583
2584
258525862587258825892590259125922593
259425952596259725982599260026012602
2603
2604
2605
2606
135136
Microwave profilerMaterial still to come here from experts on this measurement system.
Microwave radiometerMaterial still to come here from experts on this measurement system.
Fourier Transform SpectrometerMaterial still to come here from experts on this measurement system.
CeilometerMaterial still to come here from experts on this measurement system.
Cloud radarMaterial still to come here from experts on this measurement system.
Wind profilerMaterial still to come here from experts on this measurement system.
SODAR/RASSMaterial still to come here from experts on this measurement system.
Doppler lidarMaterial still to come here from experts on this measurement system.
Aerosol or ozone lidarMaterial still to come here from experts on this measurement system.
RadarMaterial still to come here from experts on this measurement system.
7.3 Operation and maintenance, quality standardsThe more traditional approach of setting a quality standard and then assessing whether each meas-urement meets that standard is less applicable in GRUAN where the emphasis is more on describ-ing, quantifying and verifying measurement uncertainty estimates and then communicating the quality of the measurement through that uncertainty estimate. That said, sStandards of operation and maintenance for each instrument used in GRUAN should be developed to ensure that min-imum quality standards are achieved. This will be necessary to minimize sources of error when measurements are being made using sophisticated instruments that may not always be completely familiar to the operator. This will be more likely the case when measurements are being made un-der operational conditions. Operation and maintenance protocols should be such that collection of detailed meta-data is mandatory as these meta-data will be vital to establishing measurement un-certainties.
67
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
26282629263026312632263326342635263626372638
137138
8 DATA MANAGEMENT
8.1 Overview of GRUAN data flowA schematic representation of the flow of data within GRUAN and from GRUAN to the user community is shown in Figure 3.
The GRUAN Data Management Manual3 defines four data levels, viz.:Level 0 (L0): Original raw data. This is the ‘rawest’ form of data available e.g. measured voltages before any processing has been applied. Even for the same instrument, formats of L0 data files are likely to differ between sites.Level 1 (L1): Converted raw data. These data are stored in a common well-described file format intended for long-term storage.Level 2 (L2): This is a standard GRUAN data product that does not require incorporation of independent measurements.
3 Available from http://www.dwd.de/bvbw/generator/DWDWWW/Content/Projekte/Gruan/Downloads/documents/gruan-td-1__data__management__v04__doc,templateId=raw,property=publicationFile.pdf/gruan-td-1_data_manage-ment_v04_doc.pdf
68
Figure 3:A schematic representation of the flow of data in GRUAN. Blue arrows show the standard flow of data. The red arrows show the flow of near-real time data. Data provided to end-users via red routes are not ‘GRUAN data’. Different data exchange protocols should operate for exchange of data within GRUAN (shaded green region) and from the GRUAN external data archive to end-users.
26392640
2641
26422643
26442645264626472648264926502651
139140141
142143
Level 3 (L3): This is also a standard GRUAN data product but an integrated product that does require incorporation of independent measurements.
Measurements and meta-data are bound together in each of these 4 levels of data. L0 data are in-gested from all GRUAN sites into the internal GRUAN data archive hosted at the Lead Centre (see Section 8.5). Direct exchange of L0 data between sites is discouraged since this circumvents the data versioning protocols and reduction of the raw data to a the common L1 file format. Simil-arly, direct exchange of L1 data between sites is discouraged since this circumvents network wide application of calibration techniques, and other algorithms applied to convert L0 to L1 data that would be implemented either at the Lead Centre or at a centralized GRUAN data processing site (see below). L0 data are expected to be archived in perpetuity at the measurement site to provide a failsafe backup while L1 data are expected to be archived in the internal GRUAN data archive at the Lead Centre.
Where GRUAN sites have agreed to the near-real time release of their data, these data will be made immediately available via the WIS. This will require some local site-based processing of the L0 data to create data suitable for submission to the WIS.
Processing of the L1 data held in the GRUAN internal data archive to produce L2 and L3 GRUAN data products will occur either at the Lead Centre or at a GRUAN station that specializes in processing data for a particular instrument. This processing would include applying the neces-sary recalibrations, corrections, and the uncertainty analysis in a consistent and traceable manner across identical instruments from different sites. The L2 and L3 data, including its meta-data and documentation, are provided to the user community through the external GRUAN data archive hosted at NCDC. A performance monitoring process (see Section 9), implemented at the Lead Centre, will provide feedback on performance to individual sites.
8.2 GRUAN data policyThis section summarizes and expands on the GRUAN data policy document prepared by the GCOS secretariat4. Since GRUAN is co-sponsored by WMO it is appropriate that any policy for release and dissemination of GRUAN data complies with WMO policy, practice and guidelines for the exchange of meteorological and related data and products. Specifically GRUAN data dis-semination and use should comply with WMO Resolution 40 (Cg-XII) which calls for free and unrestricted international exchange of meteorological data and related data and products. Because most GRUAN measurements are considered ‘essential’ in the context of Resolution 40, they are required to be exchanged without charge and with no conditions on their use. GRUAN stations are likely to be providing data to other networks which may have policies in place to protect the rights of the data providers to their own data. No conflict arises here since the data being provided through other networks are not ‘GRUAN’ data and are therefore not subject to the requirements of Resolution 40.
Three levels of exchange of GRUAN data should be recognised:i) Exchange of data within the GRUAN community. This should always occur through the
GRUAN Lead Centre so that the exchange can be controlled by data policies developed spe-cifically for internal exchange of GRUAN data.
ii) Dissemination of GRUAN products to end-users. This should always occur through the offi-cial GRUAN data centre (see Section 8.6). A different policy should be implemented to con-trol the dissemination of GRUAN data at this level. The only exception to disseminating GRUAN data through the lead centre would be when ‘pre-GRUAN’ data (see below) are sub-mitted in near real-time to centres such as ECMWF.
4 Available from http://www.dwd.de/bvbw/generator/DWDWWW/Content/Projekte/Gruan/Downloads/GRUAN__LC/gruan__data__policy,templateId=raw,property=publicationFile.pdf/gruan_data_policy.pdf
69
265226532654265526562657265826592660266126622663
266426652666
26672668266926702671267226732674
2675
267626772678267926802681268226832684268526862687
268826892690269126922693269426952696
144145
146147
iii) Dissemination of pre-GRUAN data on the WMO GTS/WIS for assimilation in NWP simula-tions.
A distinction should be made between 'standard data' and 'enhanced or experimental data' ob-tained at GRUAN sites: Standard data (e.g., near surface synoptic observations, radiosonde observations) have general
exploitation value, common measurement technology, generally well understood, and few problems with data interpretation.
Enhanced or experimental data (e.g., Raman LIDAR, microwave radiometer, surface radi-ation, GPS precipitable water) have high exploitation value, sophisticated measurement tech-nology and/or of experimental nature, would recommend contact to site scientist for correct interpretation of data, and would require considerable efforts to maintain continuous measure-ments and high quality of the data.
Enhanced or experimental data are more likely to be subject to limitations on dissemination than standard data.
The primary goals of GRUAN (see Section 1.2) are not consistent with near real-time dissemina-tion of measurements made at GRUAN sites. Generating high precision, high quality measure-ments with well characterized uncertainties takes a significant investment of time and effort. In GRUAN the emphasis is clearly on providing reference quality measurements rather than provid-ing near real-time measurements. However, it is recognized that GRUAN measurements are likely to be very useful to a number of users requiring data in near real-time e.g. for initializing NWP models. Therefore, where possible, and where it does not detract from achieving the primary goals of GRUAN, GRUAN sites should submit real-time data to end-users via the GTS/WIS. These, however, are not termed ‘GRUAN’ data since they would not necessarily have been subjected to the stringent QA/QC procedures that are core to GRUAN’s operation. Rather they are what might be termed ‘pre-GRUAN’ data. The measurements for which real-time submission may be valu-able are also more likely to be 'standard data' as described above. The WIS requirements, e.g. on meta-data, and the transmission of near-real time data via the GTS is strongly encouraged but is not considered a mandatory requirement for GRUAN sites (see Section 5.2). This decision to ex-clude near real-time submission of GRUAN data from the list of mandatory requirements for a GRUAN site is consistent with the recommendation of the AOPC who at their XIVth session stated in recommendation #29 ‘AOPC recommended that GRUAN data policy should request sites to provide all data in a free and unrestricted manner (in accordance with WMO Resolution 40 (Cg-XII)), and if possible [our emphasis] in real time, in order to be of maximum value for all applications’. Where sites do not currently have the infrastructure or expertise in making such submissions, assistance from WMO should be obtained in the form of hardware and/or training. There may be advantages to submitting data in near real-time since data assimilation algorithms are able to flag data that appear to be statistically anomalous. If such two way communication can be established between GRUAN and the NWP/data assimilation community, such information could form an important part of the measurement meta-data (Section 10). Near real-time release of standard GRUAN data will also facilitate the quality control link between GRUAN and GUAN.
When GRUAN data are used in a scientific publication, the origin of the data must be acknow-ledged and referenced. A minimum requirement is to reference GRUAN as a reference network of GCOS and to acknowledge the GRUAN data archive at NCDC as the source. If data from only one GRUAN site (or a limited number of sites) have been used, additional acknowledgement of those site(s) and their sponsoring institutions or organizations must be given, as specified in the meta-data associated with the data files.
70
26972698
2699270027012702270327042705270627072708
27092710
271127122713271427152716271727182719272027212722272327242725272627272728272927302731273227332734273527362737
273827392740274127422743
148149
Inclusion of GRUAN scientists as co-authors on papers making extensive use of GRUAN data (and in particular enhanced or experimental data) is justifiable and highly recommended, in partic-ular if a site scientist has responded to questions raised about data quality and/or suitability for the specific study in question, or has been directly involved in contributing to the paper in other ways. Co-authorship should not be a pre-condition for release of GRUAN data. However, for enhanced or experimental data it is highly recommended that data users invite site scientists to become co-authors on resultant publications, or determine whether an acknowledgement would be sufficient. Users of enhanced or experimental GRUAN data should be encouraged to establish direct contact with site scientists for the purpose of complete interpretation and analysis of data for publication purposes. GRUAN meta-data should include all information related to acknowledgements and/or co-authorship on publications making use of the data.
8.3 Collation of meta-data Two different types of meta-data need to be accommodated within the GRUAN data management facilities, viz.:i) Meta-data describing the context in which the measurement was made i.e. the calibration pro-
cedures, data processing algorithms employed, traceability to SI standards, log books, etc.. This information will be relevant to a set of data and not specific to any particular datum.
ii) Meta-data associated with each datum. For example, for point source measurements, as op-posed to column or partial column measurements, in addition to the measurement uncertainty associated with that datum, meta-data such as the exact date and time associated with the datum, as well as the exact altitude, latitude and longitude must be directly available or easily derivable from other meta-data. The provision of such meta-data recognises the fact that e.g. balloon-borne instruments drift in latitude and longitude during a flight. These data can only be used in 4D-Var assimilation of they are tagged with their 4D (time, latitude, longitude, altitude) coordinates.
8.4 Data formatIn the same way that a distinction should be made between the distribution of data within the GRUAN community and the dissemination of GRUAN data to end-users, a distinction should be made with regard to prescribed data formats for these two different aspects of data distribution, viz.:i) For distribution of data within GRUAN the emphasis should be on expediency. Different data
formats for different instruments should be permitted and not discouraged. Whatever format facilitates quick and automated processing of data and its associated meta-data should be used.
ii) For dissemination of GRUAN data to clients, a format should be selected that is flexible enough to allow a common format across all GRUAN products, should have an existing large user-base in the client community, should easily allow the inclusion of meta-data in each data file, should be an open format/standard that requires no licensing, should be self-describing, and should have a large suite of readily available tools for manipulating the data files. Per-haps the most suitable format would be NetCDF and better still CF (Climate and Forecast) compliant NetCDF. Tools such as NCO5 (NetCDF operator) should then be made available for manipulating these files.
5 http://nco.sourceforge.net/
71
27442745274627472748274927502751275227532754
2755
2756275727582759276027612762276327642765276627672768
2769
2770277127722773277427752776277727782779278027812782278327842785
150
151152
8.5 Data submissionIf sites elect to submit near real-time data to end-users, this should be done directly through the WIS or through their own portals but without a GRUAN label attached as the product does not have a robustly quantified uncertainty estimate. Otherwise all data from GRUAN sites should flow through the Lead Centre. The expectation might be that GRUAN sites submit their raw data to the GRUAN Lead Centre as soon as possible after the measurement but with the policy in place that these data will not be made available outside of the GRUAN community at this time. A facil -ity for imposing time limits on making the data available to the end-user community for different stations should be implemented as this does not contravene WMO Resolution 40 (Cg-XII). In this way stations are more likely to be willing to make their raw data immediately available within the GRUAN community without compromising their rights to first publication of the data (some funding agencies may even insist that such a data policy is in place).
Procedures for submitting data and meta-data from GRUAN sites to the GRUAN archive should be developed in such a way as to minimize the effort required at the GRUAN sites and to harmon-ize the process of data collection and data quality control across the network as a whole. For ex-ample, submission of data to the GRUAN archives can be easily automated if the mode of sub-mission is through FTP to a server based at the Lead Centre, whereas if submission must be done through a web portal this cannot be easily automated and is likely to be very time consuming for individual GRUAN sites.
Where data submission tools can be developed centrally (e.g. at the Lead Centre) and distributed for use to GRUAN sites to facilitate data submission to the GRUAN archives, this is preferable to each site independently developing such tools. The ability for sites to jointly contribute to sup-porting such network wide activities would be desirable.
8.6 Data disseminationDissemination of GRUAN data products to end-users/customers shall occur through an official GRUAN data Centre hosted at NCDC. Access to GRUAN data through a single source will rein-force the model that GRUAN data are homogeneous both in time and across GRUAN stations.
For climate research in particular it is important that users of climate data can, if required, obtain complete information on how the data they are using were acquired. Therefore, users of GRAUN data shall have access not only to the measurements and their uncertainties, but also to the instru-ment, operating procedures, data reduction algorithms used, and to when changes to any of these occurred through the complete time period of the data set.
A facility should be established whereby users of GRUAN data products can voluntarily register their use of the data. This would: Allow the Lead Centre to maintain statistics on data usage. This would be useful when apply-
ing for funding to support GRUAN operations. Allow users of data to be informed if and when newer versions of the data become available. Facilitate reporting of potential errors/anomalies in the data by end-users.
Such a facility might need to exist independently of the GRUAN NCDC archives to avoid legal issues related to data retention by US government agencies.
As discussed above, GRUAN sites are likely to also be members of other networks and are likely to submit data to end-users through other network's archives. Data submitted through a non-GRUAN networks may be subject to different data processing, different QA/QC procedures, and different calibrations resulting in a data product that is different to the GRUAN product. This is
72
2786
27872788278927902791279227932794279527962797
2798279928002801280228032804
2805280628072808
2809
281028112812
28132814281528162817
281828192820282128222823
28242825
2826282728282829
153154
not seen as a problem since the product delivered through other networks is not identified as ‘GRUAN’ data.
Users of GRUAN data need to know the version of any dataset they are using and whether newer versions might be available. The names of data files must therefore include the data version iden-tifier to facilitate easy identification of the data version. An application to periodically check for updates of GRUAN data files found on a client computer with the database at NCDC needs to be developed.
8.7 Data archivingGRUAN does not necessarily need to build its own data archive and user interface. This is a rather costly operation for any large network and partnering with an established data archive such as NCDC with a user-friendly interface should be preferred. Because data cannot be quality assured or corrected in near real-time, additional processing steps and uncertainty estimate assignment will be required. This key processing will be allowed to grow, and thus, data versioning will be re-quired. It is important that the GRUAN archive includes all previous versions of any given data set so that analyses using previous versions of data can be repeated if required.
8.8 Quality control at the instrument/site levelPart of the data management within GRUAN includes feedback to the sites in the form of reports on data submission, data quality, and comprehensiveness of meta-data submitted. Existing al-gorithms, potentially supplemented by future algorithms to be developed, shall be used operation-ally to identify systematic errors, anomalies or instrumental issues. Results of such tests shall be communicated back to GRUAN sites on short timescales so that remedial action can be taken if required. Following the example of the ARM Data Quality Office6, communicating quality con-trol results to GRUAN site operators and engineers will facilitate improved instrument perform-ance and thereby minimize the amount of unacceptable data collected.
6 http://dq.arm.gov/
73
28302831
28322833283428352836
2837
2838283928402841284228432844
2845
28462847284828492850285128522853
155
156157
9 POST-PROCESSING ANALYSIS AND FEEDBACK
This section is still under development and far from complete.
Analysis of GRUAN data products by end-users will need to be sensitive to data versioning. As new knowledge becomes available and data are reprocessed as a result, newer versions of data sets will be provided through the GRUAN archives and end-users need to be aware of such up-dates and, if necessary, repeat their own analyses. Users of GRUAN data must always document the version of data used to ensure that the analyses can be independently replicated. Key to this process will be the ability to make users aware of updated versions of data sets that they previ-ously accessed, now becoming available. The data processing centre, either the Lead Centre or the designated GRUAN site specializing in processing of that particular data set, should be tasked with data version control and ensuring that the necessary meta-data on data versions are made available to end-users.
Inevitably, algorithms change and errors in data processing occur that are not necessarily apparent until the data are used. Therefore, a facility that allows data users to report potential bugs or an-omalies found in data during analyses of the data needs to be designed and implemented. This might be modelled on the ARM Program Climate Research Facility bug reporting system.
A quality system should include procedures for feeding back into the measurement and quality control process to prevent the errors from recurring. Quality assurance can be applied in real-time post measurement, and can feed into the quality control process for the next process of a quality system, but in general it tends to operate in non-real time.
Additional points to be covered in this section include
Automation and fast feedback.
Quality management is a large part of post-processing (see Section 10)
74
28542855
2856
2857285828592860286128622863286428652866
2867286828692870
2871287228732874
2875
2876
2877
158159
10 QUALITY MANAGEMENT
This section is still under development and far from complete.
This chapter defines the principles and the methodological framework for GRUAN operations, and details how activities will be coordinated to manage and control data quality within GRUAN. Quality management within GRUAN consists of quality assurance and quality control. Quality as-surance focuses on providing confidence that quality requirements will be fulfilled and includes all the planned and systematic activities implemented in a quality system so that quality require-ments for a product or service will be fulfilled. Quality control is associated with those compon-ents used to ensure that the quality requirements are fulfilled and includes all the operational tech-niques and activities used to fulfil quality requirements. The GRUAN quality management policy is to achieve a level of data quality that allows the primary goals of GRUAN (see Section 1.2) to be met for all potential users of GRUAN data products.
Assuring the quality of the GRUAN data begins with a robust process of describing, quantifying and validating all sources of uncertainty in all GRUAN measurements and by providing rich metadata that describe all facets of the measurement process. Where total measurement uncertain-ties lie below some prescribed threshold this increases confidence in the quality of the GRUAN data. The use of redundant measurements, as described in Section 3.1.3, also serves to assure the quality of the GRUAN data products. Agreement of two independent measurements, preferably based in different measurement principles, provides a high degree of confidence that no signific-ant systematic effect was disregarded and uncertainties were not under-estimated. Laboratory tests and intercomparisons are fundamental methods for establishing and confirming uncertainty estim-ates for GRUAN data products. Laboratory tests provide an opportunity to investigate in detail the performance of instruments under controlled conditions and to measure differences against certi-fied references or other standards. Data from these experiments can be used to detect biases that may be corrected for and to determine calibration uncertainties. Field intercomparisons allow multiple in situ sensors and remote sensing data to be directly compared under the actual atmo-spheric conditions of the required measurement, including the complex environmental conditions (temperature, humidity, pressure, wind/flow rate, radiation, and chemical composition) that can-not be fully reproduced in the laboratory. These complementary activities increase confidence that measurements are subject to neither unanticipated effects nor undiscovered systematic uncertain-ties. Therefore field experiments are particularly useful for assuring the quality of GRUAN data products. The use of GRUAN data in data assimilation also adds to the assurance of GRUAN data quality since the measurements, with their uncertainties, can be tested for comparability with the data assimilation model values in a 4D-Var assimilation setting within the known internal variab-ility of the system.
Quality control will be achieved through the application of the various measurement protocols defined in this manual and in related measurement system guides. Visual inspection of all data by science/instrument experts will be required for all instruments to minimize issues that slip through automated routines. The Lead Centre shall coordinate this effort, which shall be distributed across different GRUAN sites. As outlined in Section 3.1.3, vertically resolved uncertainty estimates, prepared independently for each site, will be used as a metric to compare the site-to-site quality of the observations.
Section 4.2 of this manual provides explicit requirements regarding random errors, bias, stability, resolution and representativeness for measurements made within GRUAN. Minimizing cost without compromising quality is also an implied or explicit requirement for measurements made within GRUAN. The purpose of quality management is to ensure that GRUAN data meet the re-
75
28782879
2880
2881288228832884288528862887288828892890
28912892289328942895289628972898289929002901290229032904290529062907290829092910291129122913
2914291529162917291829192920
2921292229232924
160161
quirements in terms of uncertainty, resolution, continuity, homogeneity, representativeness, timeliness, format etc. for their intended use, at a minimum practicable cost. GRUAN recognizes that all measurements are imperfect, but, if their quality is known and demonstrable, they can be used appropriately.
Quality management is required at all points in the measurement process from network planning and training, through installation and station operations to data transmission and archiving. This quality management must include feedback and follow-up provisions across a range of timescales from near real-time to annual reviews. Because of the emphasis on the provision of robust meas-urement uncertainties and the associated requirement for in-depth quality management, the re-sources required with GRUAN will likely be a significant cost of operating the network, and very likely more that the few percent overall costs typical of many observational networks. However, without this expenditure, the quality of the data will be unknown, and their usefulness diminished.
A key aspect of quality management within GRUAN will be fulfilling customer requirements. To this end systems shall be developed to:
1. Inform users of GRUAN products of changes in measurements systems at specific sta-tions.
2. An incident reporting system that can flag data anomalies to users.3. Inform users of the availability of updates to previously accessed data products.4. Provide ‘help desk’ support to users of GRUAN data products.
Establishing close working relationship with instrument manufacturers will also be central to quality assurance within GRUAN.
A common component of quality assurance is quality monitoring or performance monitoring, a non-real-time activity in which the performance of the network or observing system is examined for trends and systematic deficiencies. Performance monitoring within GRUAN will primarily be the responsibility of the Lead Centre, but where other specialists may be co-opted to assist in per-formance assessments. The outcomes of recertification of GRUAN sites (see Section 5.4) and GRUAN site audits (see Section 5.5) will be an essential component of performance monitoring. Requests for external, independent assessments of GRUAN performance from key user groups of GRUAN data products might also serve a useful performance monitoring function. The develop-ment of quantitative performance indicators such as:
1. Data downloads,2. The number of peer reviewed publications in which GRUAN data have been used,3. Scientific case studies of the added value resulting from the use of GRUAN data products,
may serve to provide year-to-year traceability of GRUAN’s impact on the climate community.
76
2925292629272928
29292930293129322933293429352936
2937293829392940294129422943
29442945
2946294729482949295029512952295329542955295629572958
162163
ACRONYMS
ARM: Atmospheric Radiation Measurement programme
ACRF: ARM Program Climate Research Facility
AOD: Aerosol Optical Depth
AOPC: Atmospheric Observation Panel for Climate
CBS: WMO Commission for Basic Systems
CDR: Climate Data Record
CIMO: WMO Commission for Instruments and Methods of Observation
GATNDOR: GRUAN Analysis Team for Network Design and Operations Research
GCOS: Global Climate Observing System
GHG: Well-mixed greenhouse gas (CO2, CH4, N2O, CFCs, HFCs, PFCs, SF6, etc.)
GNSS: Global Navigation Satellite System
GOS: Global Observing System
GRUAN: GCOS reference upper air network
GSICS: Global Space-Based Intercalibration System
GTS: Global Telecommunication System
GUAN: GCOS upper air network
ICM: Implementation - Coordination Meeting (GRUAN)
ISCCP: International Satellite Cloud Climatology Project
NCDC: NOAA National Climate Data Centre
NMS: National Meteorological Service
NOAA: National Oceanic and Atmospheric Administration
NWP: Numerical Weather Prediction
PDF: Probability Distribution Function
RMS: Root Mean Square
TCCON: Total Carbon Column Observing Network
UT/LS: Upper troposphere/lower stratosphere
WIS: WMO Information System
WWW: World Weather Watch
77
29592960
2961
2962
2963
2964
2965
2966
2967
2968
2969
2970
2971
2972
2973
2974
2975
2976
2977
2978
2979
2980
2981
2982
2983
2984
2985
2986
2987
2988
164165
Appendix A – Expanded details on additional GRUAN Essential Cli-mate Variables
A.1. Wind speed (priority 2)
The high accuracy of 0.5 m/s prescribed for wind speed is needed to delineate calm conditions from light winds.
A.2. Wind direction (priority 2)
No supplementary comments yet.
A.3. Ozone (priority 2)
During a discussion at the ICM-2 meeting, it was suggested that ozone should develop into a pri-ority 1 variable for GRUAN. The consensus appears to be that it remains a priority 2 variable.
A.4. Methane (priority 2)
No supplementary comments yet.
A.5. Net radiation (priority 2)
The prescribed precision and accuracy values of 5 W/m2 match the requirements for the BSRN network.
A.6. Incoming short-wave radiation (priority 2)
The stated measurement range of 0 to 2000 W/m2 exceeds the solar constant (1366 W/m2) but is required since in the presence of partly cloudy skies and when the sub is not obscured by cloud, reflections off clouds can enhance surface short-wave radiation significantly. The prescribed pre-cision and accuracy values of 3 and 5 W/m2 respectively, match the requirements for the BSRN network.
A.7. Outgoing short-wave radiation (priority 2)
The prescribed precision of 2 W/m2 and accuracy of 3% match the requirements for the BSRN network.
A.8. Incoming long-wave radiation (priority 2)
The prescribed precision and accuracy values of 1 and 3 W/m2 respectively, match the require-ments for the BSRN network.
A.9. Outgoing long-wave radiation (priority 2)
The prescribed precision and accuracy values of 1 and 3 W/m2 respectively, match the require-ments for the BSRN network.
A.10. Radiances (priority 2)
The stated stability requirement of 0.03%/decade is achievable through SI traceability. The preci-sion and accuracy requirements of 0.01% and 0.15% respectively are applicable for mean sea-sonal radiances at ~1000 km spatial scale.
78
2989
29902991
2992
29932994
2995
2996
2997
29982999
3000
3001
3002
30033004
3005
30063007300830093010
3011
30123013
3014
30153016
3017
30183019
3020
302130223023
166167
A.11. Aerosol optical depth (priority 2)
Measurements of all aerosol parameters should be spectrally resolved. The aerosol optical depth is the most important of the aerosol parameters. While the other aerosol parameters will scientific-ally useful if the aerosol optical depth is large, when the aerosol optical depth is small, measure-ments of other aerosol parameters become less valuable.
A.12. Aerosol total mass concentration (priority 2)
Size-fractionated measurements are required.
A.13. Aerosol chemical mass concentration (priority 2)
Size-fractionated measurements are required.
A.14. Aerosol light scattering (priority 2)
Size-fractionated and spectral measurements are required.
A.15. Aerosol light absorption (priority 2)
Size-fractionated and spectral measurements are required.
A.16. Cloud amount/frequency (priority 2)
The prescribed precision and accuracy ranges of 0.1%-0.3% result from cloud variations of 1-3% found in the ISCCP database. The prescribed long-term stability requirement of 0.1%-0.2% res-ults from the 1-2%/decade trends found by Norris (2005).
A.17. Cloud base height (priority 2)
The prescribed measurement range of 0-20 km (1000-50 hPa) is consistent with the vertical cloud range found in Rossow and Schiffer (1999). The prescribed precision and accuracy of 100 m (10-40 hPa) is consistent with variations derived from the ISCCP database. The long-term stability re-quirement of 20m/decade is what would be required to detect the trend in global mean cloud base height of 44 m/decade reported by Chernykh et al. (2001)7.
A.18. Cloud layer heights and thicknesses (priority 2)
The prescribed vertical resolution of 50 m is required to resolve cloud layer thickness of ~30 m for cirrus clouds and is easily achievable with a lidar based system (Winker and Vaughan, 1994).
A.19. Carbon Dioxide (priority 3)
This ECV was not included in Appendix 1 of GCOS-112 but is key to understanding trends in tro-pospheric stratospheric temperatures and so is included here.
A.20. Cloud top height (priority 3)
Cloud top height measurements are also important for radiosonde temperature uncertainty ana-lysis. When a radiosonde emerges into dryer air above a cloud ,evaporation of the condensed wa-ter cools the sensor and creates a cool bias in this region. This effect can lead to deviations up to 1K above a cloud and the data need to be flagged appropriately, e.g., by assigning a correspond-inglyincreased uncertainty to data in such regions.
7 Trends reported in Chernykh have been questioned by Seidel and Durre (2003)
79
3024
3025302630273028
3029
3030
3031
3032
3033
3034
3035
3036
3037
303830393040
3041
30423043304430453046
3047
30483049
3050
30513052
3053
305430553056305730583059
168
169170
A.21. Cloud top pressure (priority 3)
No supplementary comments yet.
A.22. Cloud top temperature (priority 3)
No supplementary comments yet.
A.23. Cloud particle size (priority 4)
No supplementary comments yet.
A.24. Cloud optical depth (priority 4)
No supplementary comments yet.
A.25. Cloud liquid water/ice (priority 4)
No supplementary comments yet.
80
3060
3061
3062
3063
3064
3065
3066
3067
3068
3069
171172
REFERENCES
Alexandersson H. and A. Moberg, Homogenization of Swedish temperature data. Part I: homo-geneity test for linear trends, Int. J. Climatol., 17, 25-34, 1997.
Austin J., J. Wilson, F. Li and H. Vömel, Evolution of Water Vapor Concentrations and Strato-spheric Age of Air in Coupled Chemistry-Climate Model Simulations, J. Atmos. Sci., 64, 905-921, 2007.
Bodeker G.E., I.S. Boyd, and W.A. Matthews, Trends and variability in vertical ozone and tem-perature profiles measured by ozonesondes at Lauder, New Zealand: 1986-1996. J. Geophys. Res., 103 (D22), 28661-28681, 1998
Bodeker G.E., J.C. Scott, K. Kreher, and R.L. McKenzie, Global ozone trends in potential vorti-city coordinates using TOMS and GOME intercompared against the Dobson network: 1978-1998, J. Geophys. Res., 106 (D19), 23029-23042, 2001.
Boyd I.S., G.E. Bodeker, B.J. Connor, D.P.J. Swart, and E.J. Brinksma, An assessment of ECC ozonesondes operated using 1% and 0.5% KI cathode solutions at Lauder, New Zeal-and. Geophys.Res. Lett., 25 (13), 2409-2412, 1998.
Brinksma E.J., J.B. Bergwerff, G.E. Bodeker, K.F. Boersma, I.S. Boyd, B.J. Connor, J.F. de Haan, W. Hogervorst, J.W. Hovenier, A. Parrish, J.J. Tsou, J.M. Zawodny, and D.P.J. Swart, Validation of 3 years of ozone measurements over Network for the Detection of Stratospheric Change station Lauder, New Zealand, J. Geophys. Res., 105 (D13), 17291-17306, 2000.
Chernykh I.V., O.A. Alduchov, and R.E. Eskridge, Trends in low and high cloud boundaries and errors in height determination of cloud boundaries, Bull. Amer. Meteor. Soc., 82(9), 1941-1947, 2001.
Daumont D., J. Brion, J. Charbonnier and J. Malicet, Ozone UV Spectroscopy I: Absorption Cross-Sections at Room Temperature, J. Atmos. Chem., 15, 145-155, 1992.
Forster P.M., G.E. Bodeker, R. Schofield, S. Solomon, and D.W.J. Thompson, Effects of ozone cooling in the tropical lower stratosphere and upper troposphere, Geophys. Res. Lett., 34, L23813, doi:23810.21029/22007GL031994, 2007.
Fueglistaler, S. and Haynes, P.H., Control of interannual and longer-term variability of strato-spheric water vapor, J. Geophys. Res., 110, D24108, doi:24110.21029/22005JD006019, 2005.
GCOS-92, Implementation plan for the global observing system for climate in support of the UN-FCCC.WMO/TD No. 1219, 2004.
GCOS-112, GCOS Reference Upper-Air Network (GRUAN): Justification, requirements, siting and instrumentation options. WMO/TD-No.1379, 2007.
GCOS-121, GCOS Reference Upper Air Network (GRUAN): Report of the GRUAN Implement-ation Meeting, Lindenberg, Germany, 26-28 February 2008.
GCOS-122, Fourteenth Session of the GCOS/WCRP Atmospheric Observation Panel for Climate (AOPC-XIV), Geneva, 12-25 April 2008, WMO/TD No. 1436, 2008.
GCOS-131, Report of the first GCOS reference upper air network implementation and coordina-tion meeting (GRUAN ICM-1). WMO/TD No. 1492, 2009.
GCOS-134, GRUAN Implementation Plan 2009-2013. WMO/TD No. 1506, 2009.GCOS-138, Implementation Plan for the Global Observing System for Climate in support of the
UNFCCC (2010 Update). WMO-TD/No. 1523.GCOS-140, Report of the Second GCOS Reference Upper Air Network Implementation and Co-
ordination Meeting (GRUAN ICM-2), WMO/TD No. 1526, 2009.Gettelman A., W.J. Randel, F. Wu and S.T. Massie, Transport of water vapor in the tropical tro-
popause layer, Geophys. Res. Lett., 29, 10.1029/2001GL013818, 2002.
81
3070
3071
3072307330743075307630773078307930803081308230833084308530863087308830893090309130923093309430953096309730983099310031013102310331043105310631073108310931103111311231133114311531163117
173174
Haimberger, L., Homogenization of Radiosonde Temperature Time Series Using Innovation Stat-istics, J. Climate, 20, 1377-1403, 2007.
Immler F.J., J. Dykema, T. Gardiner, D.N. Whiteman, P.W. Thorne, and H. Vömel, Reference Quality Upper-Air Measurements: guidance for developing GRUAN data products. Atmo-spheric Measurement Techniques, 3, 1217–1231, 2010.
Khaliq M.N. and T.B.M.J. Ouarda, On the critical values of the standard normal homogeneity test (SNHT), Int. J. Climatol., 27, 681-687, 2007.
Leiterer, U., H. Dier and T. Naebert, Improvements in radiosonde humidity profiles using RS80/RS90 radiosondes of Vaisala,Contrib. Atmos. Phys., 70, 319–336, 1997.
McDermid I.S., J.B. Bergwerff, G.E. Bodeker, I.S. Boyd, E.J. Brinksma, B.J. Connor, R. Farmer, M.R. Gross, P. Kimvilakani, W.A. Matthews, T.J. McGee, F.T. Ormel, A. Parrish, U. Singh, D.P.J. Swart, and J.J. Tsou, OPAL: Network for the detection of stratospheric change ozone profiler assessment at Lauder, New Zealand 2. intercomparison of revised results. J. Geo-phys. Res., 103 (D22), 28693-28699, 1998.
McDermid I.S., J.B. Bergwerff, G.E. Bodeker, I.S. Boyd, E.J. Brinksma, B.J. Connor, R. Farmer, M.R. Gross, P. Kimvilakani, W.A. Matthews, T.J. McGee, F.T. Ormel, A. Parrish, U. Singh, D.P.J. Swart, J.J. Tsou, P.H. Wang, and J. Zawodny, OPAL: Network for the detection of stratospheric change ozone profiler assessment at Lauder, New Zealand 1. blind intercompar-ison. J. Geophys. Res., 103 (D22), 28683-28692, 1998.
Norris J.R., Multidecadal changes in near-global cloud cover and estimated cloud cover radiative forcing, J. Geophys. Res., 110, D08206, doi:08210.01029/02004JD005600, 2005.
Ohring G., B. Wielicki, R. Spencer, B. Emery, and R. Datla, Satellite Instrument Calibration for Measuring Global Climate Change: Report of a Workshop. Bull. Am. Meteorol. Soc., 86, 1303-1313, 2005.
Peter T.,C. Marcolli,P. Spichtinger, T. Corti, M.B. Baker, and T. Koop, When Dry Air Is Too Hu-mid. Science, 314, 1399–1402, DOI: 10.1126/science.1135199, 2006.
Randel, W.J.; Shine, K.P.; Austin, J.; Barnett, J.; Claud, C.; Gillett, N.P.; Keckhut, P.; Langematz, U.; Lin, R.; Long, C.; Mears, C.; Miller, A.; Nash, J.; Seidel, D.J.; Thompson, D.W.J.; Wu, F. and Yoden, S., An update of observed stratospheric temperature trends, J. Geophys. Res., 114, D02107, doi:02110.01029/02008JD010421, 2009.
Rodgers, C. D.: Inverse Methods For Atmospheric Sounding: Theory and Practice, World Sci-entific Publishing Co. Pte. Ltd., 2, 240, 2000.
Rosenlof K.H., S.J. Oltmans, D. Kley, J.M. Russell III, E.-W. Chiou, W.P. Chu, D.G. Johnson, K. K. Kelly, H.A. Michelsen, G.E. Nedoluha, E.E. Remsberg, G.C. Toon, and M.P. McCormick, Stratospheric water vapour increases over the past half-century, Geophys. Res. Lett., 28 (7), 1195-1198, 2001.
Rossow W. B., and R. A. Schiffer, Advances in Understanding Clouds from ISCCP, Bull. Amer. Meteor. Soc., 80(11), 2261-2287, 1999.
Schmidlin, F.J., Derivation and application of temperature corrections for the United States radio-sonde, in Symposium on Meteorological Observations and Instrumentations, 7th, New Or-leans,LA, 14–18 January 1991, Preprints (A92-32051 12-47), Boston, MA, American Met-eorological Society, 227–231, 1991.
Seidel D.J. and M. Free, Measurement Requirements for Climate Monitoring of Upper-Air Tem-perature Derived from Reanalysis Data, J. Climate, 19, 854-871, 2006.
Schubert S., D. Dee, S. Uppala, J. Woollen, J. Bates, and S. Worley, Report of the Workshop on “The Development of Improved Observational Data Sets for Reanalysis: Lessons learned and Future Directions”, 31 pp., http://gmao.gsfc.nasa.gov/pubs/docs/Schubert273.doc, 2006.
Seidel D.J., and I. Durre, Comments on “Trends in low and high cloud boundaries and errors in height determination of cloud boundaries”, Bull. Amer. Meteor. Soc., 84(2), 237-240, 2003.
82
3118311931203121312231233124312531263127312831293130313131323133313431353136313731383139314031413142314331443145314631473148314931503151315231533154315531563157315831593160316131623163316431653166
175176
Seidel D. J., F. H. Berger, H. J. Diamond, J. Dykema, D. Goodrich, F. Immler, W. Murray, T. Peterson, D. Sisterson, M. Sommer, P. Thorne, H. Vömel, and J. Wang, Reference upper-air observations for climate: rationale, progress, and plans, Bull. Amer. Meteor. Soc., 3, 361-369, 2009.
Shine K.P., M.S. Bourqui, P.M.D. Forster, S.H.E. Hare, U. Langematz, P. Braesicke, V. Grewe, M. Ponater, C. Schnadt, C.A. Smith, J.D. Haigh, J. Austin, N. Butchart, D.T. Shindell, W.J.Randel, T. Nagashima, R.W. Portmann, S. Solomon, D.J. Seidel, J. Lanzante, S. Klein, V. Ramaswamy, and M.D. Schwarzkopf,A comparison of model-simulated trends in strato-spheric temperatures. Quarterly Journal of the Royal Meteorological Society, 129 (590), 1565–1588, 2003.
Smit H.G.J., W. Straeter, B.J. Johnson, S.J. Oltmans, J. Davies, D.W. Tarasick, B. Hoegger, R. Stubi, F.J. Schmidlin, T. Northam, A.M. Thompson, J.C. Witte, I. Boyd, and F. Posny, As-sessment of the performance of ECC-ozonesondes under quasi-flight conditions in the envir-onmental simulation chamber: Insights from the Juelich Ozone Sonde Intercomparison Ex-periment (JOSIE). J. Geophys. Res., 112, D19306, doi:10.1029/2006JD007308, 2007.
Soden B.J., D.D. Turner, B.M. Lesht, and L.M. Miloshevich, An analysis of satellite, radiosonde, and lidar observations of upper tropospheric water vapor from the Atmospheric Radiation Measurement Program. J. Geophys. Res., 109, D04105, doi:10.1029/2003JD003828, 2004.
Solomon S., K.H. Rosenlof, R.W. Portmann, J.S. Daniel, S.M. Davis, T.J. Sanford and G.-K.-Plattner, Contributions of Stratospheric Water Vapor to Decadal Changes in the Rate of Global Warming, Science, 327, 1219-1223, 2010.
Steiner, A.K.; Kirchengast, G.; Lackner, B.C.; Pirscher, B.; Borsche, M. and Foelsche, U., Atmo-spheric temperature change detection with GPS radio occultation 1995 to 2008, Geophys. Res. Lett., 36, L18702, doi:18710.11029/12009GL039777, 2009.
Suortti T.M., A. Kats, R. Kivi, N. Kämpfer, U. Leiterer, L.M. Miloshevich, R. Neuber, A. Paukkunen, P. Ruppert, H. Vömel and V. Yushkov, Tropospheric Comparisons of Vaisala Radiosondes and Balloon-Borne Frost-Point and Lyman-α Hygrometers during the LAUTLOS-WAVVAP Experiment, J. Atmosph. Oceanic Tech., 25, 149-166, 2008.
Thompson A.M., J.C. Witte, H.G.J. Smit, S.J. Oltmans, B.J. Johnson, V.W.J.H. Kirchhoff and F.J. Schmidlin, Southern Hemisphere Additional Ozonesondes (SHADOZ) 1998–2004 tropical ozone climatology: 3. Instrumentation, station-to-station variability, and evaluation with sim-ulated flight profiles, J. Geophys. Res., 112, D03304, doi:10.1029/2005JD007042, 2007.
Tiao, G. C., G. C. Reinsel, D. Xu, J. H. Pedrick, X. Zhu, A. J. Miller, J. J. DeLuisi, C. L. Mateer, and D. J. Wuebbles (1990), Effects of autocorrelation and temporal sampling schemes on es-timates of trend and spatial correlation, J. Geophys. Res., 95(D12), 20507-20517.
Trenberth, The need for a new upper air baseline network, [details from Kevin to come in here], 2003.
Vömel H., J.E. Barnes, R.N. Forno, M. Fujiwara, F. Hasebe, S. Iwasaki, R. Kivi, N. Komala, E. Kyrö, T. Leblanc, B. Morel, S.-Y. Ogino, W.G. Read, S.C. Ryan, S. Saraspriya, H. Selkirk, M. Shiotani, J.V. Canossa and D.N. Whiteman, Validation of Aura Microwave Limb Sounder water vapor by balloonborne Cryogenic Frost point Hygrometer measurements, J. Geophys. Res., 112, D24S37, doi:10.1029/2007JD008698, 2007a.
Vömel H., D.E. David and K. Smith, Accuracy of tropospheric and stratospheric water vapor measurements by the cryogenic frost point hygrometer: Instrumental details and observa-tions, J. Geophys. Res., 112, D08305, doi:10.1029/2006JD007224, 2007b.
Wang, J., D.J. Carlson, D.B. Parsons, T.F. Hock, D. Lauritsen, H.L. Cole, K. Beierle, and E. Chamberlain, Performance of operational radiosonde humidity sensors in direct comparison with a chilled mirror dew-point hygrometer and its climate implication. Geophys. Res. Lett., 30, 1860, doi:10.1029/2003GL016985, 2003.
83
3167316831693170317131723173317431753176317731783179318031813182318331843185318631873188318931903191319231933194319531963197319831993200320132023203320432053206320732083209321032113212321332143215
177178
Weatherhead, E. C., G. C. Reinsel, G. C. Tiao, X.-L. Meng, D. Choi, W.-K. Cheang, T. Keller, J. DeLuisi, D. J. Wuebbles, J. B. Kerr, A. J. Miller, S. J. Oltmans, and J. E. Frederick (1998), Factors affecting the detection of trends: Statistical considerations and applications to envi-ronmental data, J. Geophys. Res., 103(D14), 17149-17161.
Winker, D.M., and M.A. Vaughan, Vertical distribution of clouds over Hampton, Virginia ob-served by lidar under the ECLIPS and FIRE ETO programs, Atmos. Res., 34, 117-133, 1994.
WMO-No. 8, Guide to Meteorological Instruments and Methods of Observation, 7th edition, 2008.
WMO-No. 488, Guide to the Global Observing System, 2010.World Meteorological Organization, 2011: WMO Intercomparison of High Quality Radiosonde Systems, Yangjiang, China (J. Nash, T. Oakley, H. Vömel and Li Wei) Instruments and Ob-serving Methods Report No. 107, WMO/TD-No 1580, Geneva
Zhou S., M.A. Geller and M. Zhang, Cooling trend of the tropical cold point tropopause temperat-ures and its implications, J. Geophys.Res., 106, 1511-1522, 2001.
84
321632173218321932203221322232233224322532263227
322832293230
179180