wmo-teco, helsinki, 31 aug – 1 sept 2010 vuerich e. (italy)

32
THE WMO FIELD INTERCOMPARISON OF RAINFALL INTENSITY (RI) GAUGES in Vigna di Valle (ITALY), October 2007- April 2009: relevant aspects and results. WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010 Vuerich E. (ITALY) Sestola (Italy), 23 June 2009

Upload: maura

Post on 04-Feb-2016

41 views

Category:

Documents


0 download

DESCRIPTION

Sestola (Italy), 23 June 2009. THE WMO FIELD INTERCOMPARISON OF RAINFALL INTENSITY (RI) GAUGES in Vigna di Valle (ITALY), October 2007- April 2009: relevant aspects and results. WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010 Vuerich E. (ITALY). CONTENTS. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

THE WMO FIELD INTERCOMPARISON OF RAINFALL INTENSITY (RI) GAUGES in Vigna di

Valle (ITALY), October 2007- April 2009: relevant aspects and results.

WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010 Vuerich E. (ITALY)

Sestola (Italy), 23 June 2009

Page 2: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

CONTENTS

BACKGROUND AND

OBJECTIVES

PROCEDURES AND

METHODS

DATA ANALYSIS AND

RESULTS

CONCLUSIONS AND

DEVELOPMENTS

RECOMMENDATIONS

THE INTERCOMPARISON TEST SITE

ITALIAN MET SERVICE – AIR FORCE (IMS)

CENTRE OF MET EXPERIMENTATIONS (RESMA)

VIGNA DI VALLEITALY

Page 3: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

BACKGROUND

Following the increased recognition of scientific and practical issues related to the assessment of possible climatic trends, the mitigation of natural disasters (e.g. storms and floods), the hindering of desertification, the attention paid to accuracy and reliability in rainfall measurements is currently increasing

The WMO recognized these emerging needs and promoted a couple of mile-stone meetings:

- (1) Bratislava, Slovak Republic, 23-25 April 2001: calibration of rain gauges and general aspects of measurements (I phase: WMO Laboratory Intercomparison of RI gauges, 2004-2005)

- (2) Geneva, Switzerland, 5-9 December 2005: operational aspects of field measurements and achievable accuracy (II phase: WMO Field Intercomparison of RI gauges, 2007-2009)

Page 4: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

BACKGROUND Previous WMO intercomparisons: accumulated

amounts, on low RI or on snowfall, on qualitative information

The WMO FI-RI gauges started in Vigna di Valle (Italy) at the IMS Centre of Met Experimentations on the 1st of October 2007 and, after a 6-months extension, it was concluded in April 2009 (first intercomparison for quantitative 1MIN-RI)

Main objective: to assess and compare quantification and catching errors of both catching and non-catching type of RI gauges with the emphasis on high rainfall intensity in order to complain with the WIGOS basic concepts of improving standardization, data homogeneity, data quality, compatibility, interoperability.

Other tasks: providing draft recommendations for CIMO and guidance material for inclusion into the CIMO guide.

Page 5: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

PROCEDURES AND METHODS 30 RI gauges (4 working references) capable to measure

RI up to 200mm/h at 1min resolution were selected, randomly distributed (avoid clustering of large gauges) and installed on dedicated ground platforms at 1 m height (no windshields)

13 additional meteorological sensors were installed in later positions for monitoring environmental conditions.

Page 6: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

PROCEDURES AND METHODS

# RAIN GAUGES MEAS. PRINCIPLE

1 RIMCO 7499 Tipping bucket

2 Paar AP23 Tipping bucket

3 Precis-Mecanique Tipping bucket

4 Thies PT Tipping bucket

5/27 ETG R102 Tipping bucket

6 LSI-LASTEM DQ031

Tipping bucket

7 SIAP-MICROS UM7525/I

Tipping bucket

8/28 CAE PMB2 Tipping bucket

9 Davis Rain Collector II

Tipping bucket

10 Lambrecht 15188 Tipping bucket

11 MTX PP040 Tipping bucket

12 Env. Meas. Ltd ARG100

Tipping bucket

13/29 Meteoservis MRW500

Weighing gauge

# RAIN GAUGES MEAS. PRINCIPLE

14 Vaisala VRG101 Weighing gauge

15 OTT Pluvio Weighing gauge

16 EWS PG200 Weighing gauge

17/30 GEONOR T-200B Weighing gauge

18 MPS TRwS Weighing gauge

19 SA „MIRRAD“ MPA-1M

Not Participating

20 Vaisala PWD22 Optical Disdrometer

21 OTT Parsivel Optical Disdrometer

22 Thies LPT Optical Disdrometer

23 Vaisala WXT510 Acoustic impact

24 Eigenbrodt ANS 410

Water pressure

25 KNMI electric raingauge

Water level

26 PVK ATTEX “DROP”

Doppler Radar

List of participating rain gauges (26+4 working references)

Page 7: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

PROCEDURES AND METHODS 4 Reference gauges, the “Composite Working Reference

” (C.W.R.) were inserted in Reference Rain Gauge Pits (R.R.G.P.) at the centre of the Intercomparison site (collectors at ground level) – Minimization of weather related catching errors (e.g. Jevons effect, 1861)

Standard adopted: ISO/EN-13798: “Specification for a reference rain gauge pit”, recently revised in 2010CWR (Recom. 3 CIMO-XIV): corrected tipping bucket rain gauges (TBRG) and weighing gauges (WG) with the shortest step response and the lowest uncertainty according to Results of WMO Laboratory Intercomparison of RI 2004-2005.

ETG-R102

GEONORT200B

CAE-PMB2Meteoserv

isMRW500

Page 8: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

PROCEDURES AND METHODS 20 catching type gauges were calibrated in the

laboratory before the Field Intercomparison. The WMO recognized laboratory at the University of Genoa performed the calibration based on the generation of a constant water flow from a reference hydraulic device (tests according to Recom.2 CIMO-XIV + statistics based on 1 min evaluation of relative errors )

Objectives: to single out the quantification errors associated with each instrument; measurement uncertainty and understanding of field resultsReference hydraulic

device

The Qualification Module for RI Measurement Instruments

developed at the University of Genova

(Ur(95%)=0,45%)

Page 9: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

PROCEDURES AND METHODS Lab tests: reference flow rates at 2, 20, 50, 90, 130, 170,

200 mm/h at 1MIN resolution for a variable duration, arithmetic mean of relative errors and spreading of data around it, providing correction curves (to be used beyond the intercomp. time)

Results: the constant flow response plot, gauge relative error plotted versus Lab reference RI (superimposed box-plots)

Page 10: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

PROCEDURES AND METHODS Further lab tests: investigation of the dynamic

performance gauges at 1MIN resolution through the evaluation of the step response to a step input (time constant) which is a measure of the instrument stability and ability to detect rapid changes of the input signal.

Results: step response plot, ratio of measured RI / Lab Reference RI versus time.

Page 11: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

PROCEDURES AND METHODS

Quality assurance:

Raw data are processed by the Automatic Quality Control (AQC) to provide quality checked 1-minute data

Periodic field tests: Field Calibration Device (developed at DICAT Laboratory – Univ. of Genoa) – Metrological confirmation

Periodic maintenance, daily visual checks, cleaning of instruments, calibration status checks

Meeting of Participants and local staff (Vigna di Valle, 21-22 MAY 2008) – HMEI participated and reported

QA reports by the Site Manager to the ET.

Meeting of Participants

Field Tests-Calibration Device

Page 12: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS Analysis dataset: 1MIN-QC checked RI data obtained

through the reduction of the full FI dataset (162) by considering only synchronized events and events with 2 consecutive minutes and RI>12mm/h (totally 43 events, approx. 740 data).

No. 2 events with RI > 150mm/h; No. 5 events with RI>100mm/h; No. 15 events with RI>60mm/h …..

Total availability of 1-min data (rain/no rain)

1-min valid data (rain/no rain): percentage of

available 1min data that are valid

according to QC

Total numbers of

precipitation daily events

Hail and Mixed Rain/Hail events

T.A. = 95.4% 98.2% of T.A.162

(Full FI Dataset)

6 events: 13th Jan, 4th Feb, 7th May, 30th Oct 2008; 1st

Jan, 5th Mar 2009

Numbers of synchronized

events

Numbers of events for reference RI

calculations

Number of events for data analysis of rain

gauges

Rainfall accumulated over

the intercomparison

period85

(Reduced FI Dataset)

79 (28 000 1-min data)

43(740 1-min RI

data)1325 mm

Page 13: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS

Main objective: performing the best inter-comparison of rain gauges in high RI field conditions through the determination of a reference value (1MIN RIref)

RIref is the best estimation of the 1 min RI true value and it must be derived through a RI Composite Working Reference value.

4 Nov 2008

Page 14: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS

Method: weighted average of 1-min RIi measured by the 4 reference gauges

Weights (µi, working reference gauge i) were the most challenging issue, since they must take into account effects related both to dynamical characteristics and possible lack of synchronization on 1 min time base

Si are 3 statistical parameters calculated throughout the database of all precipitation events

F = 0,1: “gross” parameter, for each working reference and for each event (F=1 if the working ref is not affected by 1-min lack of synchronization or high dynamic oscillation)

ii

iii

ref

RIRI

iii

iii FS

FS1

1

ik

ikiS

N

RIRIN

j

kj

ij

ik

1

2

2

)(

(k≠i)

F=0,1

µR102 µPMB2 µMRW50

0

µT200B

0.27 0.25 0.23 0.25

Page 15: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS

CAE-PMB2

M-S MRW500

GEONOR- T200B

ETG-R102

Page 16: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS Synchronization: to compare the 1-min RI data of all

instruments, a synchronization of ±10 s was required, in other words the internal clock of the instrument should be within ±10 s compared to the DAQ system timestamp (nominal timestamp). If the difference/delay between the instrument’s data output time and the nominal timestamp exceeds the required ±10 s time interval, the result cannot be correctly compared to synchronized gauges.

0

10

20

30

40

50

60

70

12.30 12.34 12.38 12.42 12.46 12.50 12.54 12.58 13.02

Time

RI

[mm

/h]

gauge Agauge Bgauge C

Gauge A and C are synchronized with the DAQ system clock; gauge B has a delay exceeding 10 seconds. Arrows indicate sample points of B with large difference due to non-synchronized data points of gauge B.

Page 17: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS Uncertainty of the reference: assuming a normal distribution

of the deviations of pit gauges RI, the standard deviation of the distribution with respect to the reference intensity is calculated according (σ=[∑(RI-RIref)2/N]1/2); thus U95(RIref)=2 σ=4.3mm/h

Relative uncertainty: ur(RIref)= (U(RIref) / RIref )∙100 [%] (green in plot)

Page 18: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS

To show the general results of the intercomparison, the intensities measured by the rain gauges are plotted versus 1MIN Riref (CWR) and data are fitted with a power law trend curve RI=a x RIrefb (corresponding bet fit equation reported on graphs)

To assess the accuracy of field measurements compared to the reference, a tolerance region (composed of the WMO required 5% uncertainty and of the uncertainty of the reference, calculated as [urel(RIref)2+52]1/2 [%] ) is represented in dashed lines on each plot.

For easier comparison, the instruments have been divided into seven groups according to the measurement principle, as indicated in the title of each plot. WG instruments are split in two groups for easier presentation of results.

Specific results and plots: Data Sheets included in the Final Report

To assess the impact of wind losses (Jevons effect) on catching errors, the ratio RIiOUT/ RIi is plotted versus wind speed (being RIiOUT and RIi the measured intensities by two identical gauges, one in the pit and the other one in the corresponding open field platform)

Page 19: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS

yMcVan = 1.31x0.90 R2 = 0.68

yPAAR = 1.15x0.96 R2 = 0.85

yLASTEM = 1.06x0.96 R2 = 0.72

yDAVIS = 1.16x0.92 R2 = 0.73

yMTX = 0.96x1.0 R2 = 0.79

yEML = 1.21x0.92 R2 = 0.75

0

20

40

60

80

100

120

140

160

180

200

220

0 20 40 60 80 100 120 140 160 180 200 220

RI reference [mm/h]

RI [

mm

/h]

PP040-MTX

AP23-PAAR

DQA031-LSI LASTEM

ARG100-EML

RIM7499020-McVan

Rain Collector II-DAVIS

tolerance region

TBRG

Page 20: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTSTBRG-SC

y ETG= 1.01x0.99 R2 = 0.88

yCAE = 0.78x1.05 R2 = 0.87

ySIAP-MICROS = 0.92x1.02 R2 = 0.73

0

20

40

60

80

100

120

140

160

180

200

220

0 20 40 60 80 100 120 140 160 180 200 220

RI reference [mm/h]

RI [

mm

/h]

tolerance region

UMB7525/I/SIAP-MICROS

PMB2-CAE

R102-ETG

Page 21: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTSTBRG-PC

y LAMBRECHT = 1.21x0.96 R2 = 0.81

y THIES= 1.01x0.99 R2 = 0.85

0

20

40

60

80

100

120

140

160

180

200

220

0 20 40 60 80 100 120 140 160 180 200 220

RI reference [mm/h]

RI [

mm

/h]

tolerance region

PT 5.4032.35.008-THIES

LB-15188-LAMBRECHT

Page 22: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTSTBRG-MC &

LRG

y Precis Mecanique= 1.08x0.95 R2 = 0.77

y KNMI= 1.05x0.97 R2 = 0.82

0

20

40

60

80

100

120

140

160

180

200

220

0 20 40 60 80 100 120 140 160 180 200 220

RI reference [mm/h]

RI [

mm

/h]

tolerance region

R013070-PRECIS MECANIQUE

Electrical raingauge-KNMI

Page 23: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTSWG (1)

yPLUVIO-OTT = 0.98x1.00 R2 = 0.90

yEWS = 0.98x1.00 R2 = 0.81

y GEONOR= 0.96x1.00 R2 = 0.89

0

20

40

60

80

100

120

140

160

180

200

220

0 20 40 60 80 100 120 140 160 180 200 220

RI reference [mm/h]

RI [

mm

/h]

tolerance region PLUVIO-OTT

PG200-EWS

T200B-GEONOR

Page 24: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTSWG (2)

yMETEOSERVIS= 1.01x0.98 R2 = 0.74

y MPS= 1.09x0.95 R2 = 0.59

y VRG101-VAISALA= 1.12x0.75 R2 = 0.12

yEIGENBRODT = 1.09x0.96 R2 = 0.67

0

20

40

60

80

100

120

140

160

180

200

220

0 20 40 60 80 100 120 140 160 180 200 220

RI reference [mm/h]

RI [

mm

/h]

tolerance region

VRG101-VAISALA

TRwS-MPS

ANS 410/H-EIGENBRODT

MRW500 -METEOSERVIS

Page 25: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTSNON CATCHING TYPE RAIN

GAUGES

y WXT510-VAISALA= 1.72x0.91 R2 = 0.74

y PVK ATTEX= 1.43x0.82 R2 = 0.53

yPARSIVEL-OTT = 0.82x1.10 R2 = 0.77

yPWD22-VAISALA = 0.81x0.94 R2 = 0.51

yLPM-THIES = 0.93x1.07 R2 = 0.80

0

20

40

60

80

100

120

140

160

180

200

220

0 20 40 60 80 100 120 140 160 180 200 220RI reference [mm/h]

RI [

mm

/h]

tolerance region

PWD22-VAISALA

LCR-PVK ATTEX

WXT510-VAISALA

PARSIVEL-OTT

LPM-THIES

Page 26: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTSMeasurement accuracy of all gauges by relative deviations

from RIref

Page 27: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

DATA ANALYSIS AND RESULTS

A relevant effect of the wind did not appear in this intercomparison, we are able to affirm that the wind is not affecting in a significant way the outer instruments compared to those installed in the pit.

yR102 = -0.008x + 0.985

y T200B= -0.009x + 1.003

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Wind speed [m/s]

RIo

ut/R

Ipit

Gauge R102 - RIout/RI_pit

Gauge T200B - RIout/RI_pit

Linear (Gauge R102 - RIout/RIpit)

Linear (Gauge T200B - RIout/RIpit)

Page 28: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

CONCLUSIONS AND DEVELOPMENTS This comparison at 1MIN time scale in field conditions

demonstrates the possibility to evaluate the performance of RI gauges, identifying directions where research and technological development should be oriented.

The achievable accuracy of WG can be improved in field conditions by means of the reduction of the response time below 1-minute and by appropriate filtering methods

With regard to TBRG, the method applied by TBRG-SC confirms the possibility to improve the 1-min RI resolution and to provide accurate field measurements

With regard to TBRG, the method applied by TBRG-PC revealed the possibility to provide accurate field measurements at higher RI, even if the performance is limited by their resolution at lower RI.

With regard to non catching type rain gauges, despite the need of very low maintenance and the possibility to determine the type of precipitation (not for all), none of them agreed well with the reference. We are confident that their performance will be improved in the next future, considering the possibility of improving calibration methods to reduce systematic errors.

Page 29: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

CONCLUSIONS AND DEVELOPMENTS Disdrometers tended to overestimate the RI. Despite their

very different calibration procedures, they agreed better to each other than to the reference, as indication of a good degree of precision but they were not as accurate as conventional gauges

For the best quality instruments, the achievable measurement uncertainty in laboratory, under constant flow conditions, was found to be 5% above 2 mm/h and 2% above 10 mm/h.

One of the most challenging aspects of the Intercomparison was the definition of a 1-minute field reference intensity. The procedure adopted confirmed the suitability of R102-ETG, PMB2-CAE and T200B-GEONOR for the calculation of the reference.

The rainfall intensity is highly variable from minute to minute. Therefore, the time synchronization of the instruments is crucial to inter-compare their measurements and to design the measurement systems.

The results of the intercomparison confirmed the feasibility to measure 1MIN RI and provided information on the achievable measurement uncertainties, responding to the need of accuracy, homogeneity and standardization of RI measurements, as required by WIGOS.

Page 30: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

CONCLUSIONS AND DEVELOPMENTS In field conditions, the uncertainty of the RI CWR was

evaluated to be 4.3 mm/h. Consequently, the relative uncertainty of the reference was found to be below 5% only for intensities above 90 mm/h. Below 90 mm/h, the relative uncertainty of the reference values was higher than the 5% required measurement uncertainty provided in the CIMO Guide.

A number of standardization activities at national and European level (CEN- TC “Hydrometry”)) has been promoted by the Italian Met Service, such as a national standard on the classification of rain gauges according to accuracy, the revision of the standard “Specification for a reference rain gauge” (ISO/EN13798:2010), a pre-standard (CEN technical report) on RI measurements (a possible WMO/ISO common standard in future)

The IMS proposed the establishment of a CIMO Lead Centre on precipitation intensity (Univ. Genova, Vigna di Valle, M.Cimone).

Page 31: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

RECOMMENDATIONS (3 of 18) The 1-min RI is highly variable from minute to minute.

Therefore, it is recommended that 1-min RI should only be measured in a station and used for further analysis if the following conditions are met: (a) All 1-min data must be transmitted and used (a single 1-min RI value is not representative of a longer period of time; (b) A very good time synchronization, better than 10 s, must be achieved, both between the reference time and the different instruments of the observing station.

Changes be made to the CIMO Guide Table:Precipitation intensity (liquid):

Achievable measurement uncertainties: -Under constant flow conditions in laboratory, 5% above 2 mm/h, 2% above 10 mm/h. -In field conditions, 5 mm/h, and 5% above 100 mm/h

It is recommended that RI measurements be further standardized based on the advances in knowledge obtained from this intercomparison to allow the users to obtain homogeneous data sets (based on the achievable RI measurement performance (accuracy) rather than on measuring principle or gauge design/technical solutions)

Page 32: WMO-TECO, Helsinki, 31 Aug – 1 Sept 2010                             Vuerich E. (ITALY)

ACKNOLEDGEMENTS

The ITALIAN MET SERVICE – AIR FORCE

The WMO-CIMO (Dr Ondras, Dr Ruedi)

The colleagues, local staff of the intercomparison site (Vigna di Valle)

Dr Claudia Monesi – Univ. of Rome TRE (data analyst)

Dr Eckhard Lanzinger – DWD (The project Leader)

Dr Bruce Baker –USA (ET member)

Dr Michel Leroy – Meteo France (ET chair)

And

Prof. Luca Lanza and Eng. Luigi Stagi – Univ. of Genova