daos-wg: data assimilation and observing systems working group pierre gauthier department of earth...

19
DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Upload: alex-gardner

Post on 27-Mar-2015

218 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

DAOS-WG:Data Assimilation and Observing Systems Working Group

Pierre GauthierDepartment of Earth and Atmospheric Sciences

Université du Québec à Montréal

Page 2: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Merge of OS with DAOS

• Work of the Observing Systems working groups already covered through OPAG-IOS

• Best to combine with DAOS (Observing Systems)• One co-chair from OS group (R. Saunders) and F. Rabier resigns as co-

chair of the DAOS-WG.

• Composition of the working group (16 members)

Pierre Gauthier (UQAM, Canada) and Roger Saunders (UK Met Office), co-chairs

* Bertrand Calpini (MeteoSwiss, Switzerland), Carla Cardinali (ECMWF)

* Jochen Dibbern (EUMETNET/DWD), Ron Gelaro (NASA, US)

* Tom Hamill (NOAA/ESRL), Tom Keenan (CAWCR , Australia)

* Ko Koizumi (JMA), Rolf Langland (NRL) , Andrew Lorenc (UK Met Office)

* Tetsuo Nakazawa (JMA, Japan), Florence Rabier (Météo-France)

* Peter Steinle (CAWCR, Australia)

* Michael Tsyroulnikov (Hydromet Research Centre, Russia)

* Chris Velden (University of Wisconsin, US),, Germany),

Page 3: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Data Assimilation and Observing Systems mission

• The DAOS WG ensures that THORPEX contributes to the optimisation of the use of the current WMO Global Observing System.

• It contributes to the development of a well-founded strategy for the evolution of the Global Observing System to support NWP (primarily 1-14 days).

Page 4: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Data Assimilation and Observing Systems WG strategy

• It addresses issues in DA and improved understanding of the sources and growth of errors in analyses and forecasts

• It promotes research activities that lead to a better use of observations and the understanding of their value

• It provides input and guidance for THORPEX regional campaigns for the deployment of observations to achieve their objectives

• This will be done in collaboration with the CBS OPAG-IOS

Page 5: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Approaches to measuring the impact of assimilated observations

Information content

* based on the relative accuracy of observations and the background state

Observing System Experiments

* Data denials

* Global view of the impact of observations on the quality of the forecasts

Observation impact on the quality of the forecasts

* Sensitivities with respect to observations based on adjoint methods (Baker and Daley, 2000; Langland and Baker, 2004)

* Ensemble methods

Page 6: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Information content

• Ratio of the analysis error covariance to B

The information gained from assimilating a given set of observations is represented by the second term, where N is the dimension of the model space

• … and in observation space

with M being the number of observations

KHKHIBP trNtrtrtr a 1

HKHBHHHP

HBHBHHPP

trMtr TTa

TTaa

1

Page 7: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Globeobs_Total

gionRetype_ObsgionRe

type_Obs DFS

DFS100(%)DFS

We assumed that the complete set of observations can be split in observation subsets with independent errors (R is block-diagonal);

Regions : HN, HS, TROPICS;Obs_types : AI, GO, PR, SF, SW, AMSU-A, AMSU-B, RAOB;

Computation of DFS in MSC’s 3D-Var and 4D-Var systems(Lupu, 2009)

DFS for each type of observations

Page 8: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Observation impact per observation in each region

k

gionRek

p

DFS100(%)IC

Lupu, 2009

Page 9: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

OSEs experiments: 3D-Var and 4D-Var, North America

k

NAk

p

DFSDFS values per obstype normalized by the number of observations.

NO_RAOB: DFS per single observation notably increases, especially for AMSU-B and GO;

NO_AIRCRAFT: DFS per single observation notably increases, especially for RAOB, SF and PR; For other observations (GO, SW and AMSU-B) DFS per obs also increases slightly.

Page 10: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Observations move the model state from the “background” trajectory to the new “analysis” trajectory

The difference in forecast error norms, , is due to the combined impact of all observations assimilated at 00UTC

Observation Impact Methodology(Langland and Baker, 2004)

24 30e e

OBSERVATIONS ASSIMILATED

00UTC + 24h

24 30e e

24e30e

Page 11: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Evaluation of the impact of observations

At initial time

b

bTb

a

aTa

TTb

ba

JJe

xL

xLKxHy

where : observation departure from the background state

Computation of the Observation Impact:

One can obtain w by slightly adapting the assimilation to solve

HwRxx

K 1

0

ttb

b

a

aT JJ

bxHy

0

11

2

1

2

1

ttba

TTT JJF

xx

wHwRHwwBww

0ttba

JJ

xx

Page 12: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Removal of AMSUA results in large increase in AIRS (and other) impacts

Removal of AIRS results in significant increase in AMSUA impact

Removal of raobs results in significant increase in AMSUA, aircraft and other impacts (but not AIRS)

Combined Use of ADJ and OSEs (Gelaro et al., 2008)

…ADJ applied to various OSE members to examine how the mix of observations influences their impacts

Page 13: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

WMO Workshop on impact studiesGeneva, 19-21 May 2008

• Organized by the OPAG-IOS• Participation of the DAOS

* Cardinali, Gauthier, Gelaro, Koizumi, Langland, Rabier, Steinle

* Preliminary results from the intercomparison experiment were presented by Cardinali, Gelaro and Langland

• DAOS-WG to provide input on the design of the global observing system

* Recommendation at the workshop to use the adjoint based method to get a more detailed assessment of the impact of observations

* Nice complement to OSEs

Page 14: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Intercomparison experiment

• Numerous differences exist between assimilation systems and influence the evaluation of the impact of observations

* Observation coverage

* Assimilation methodology

* Forecast model

* Metrics used to compute the forecast sensitivity Possibility to use more appropriate metrics for socio-economic

applications

* Baseline experiment provided a common context against which three different systems evaluated the impact of observations with the same tool

* Differences nevertheless persist in terms of assimilation methodologies and models (e.g., 3D-Var and 4D-Var)

* For each system, the total impact of observations evaluated with the LB04 method is consistent with results from OSEs.

• Further experimentation with different approaches* Ensemble methods

Page 15: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Value of targeted data (1)

• Value of extra-tropical targeted data has been found to be positive but small, on average* Observations taken in sensitive areas have more value than

observations deployed randomly

* Past experiments do not provide evidence of big impact obtained from just a few observations (when averaged over a large sample of cases)

* There are limitations due to the current assimilation methodologies (not yet fully flow-dependent)

* Sensitive areas characterization does not appear to be the first order problem

• Additional observations for tropical cyclones have proven to be useful.

Page 16: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Value of targeted data (3): Recommendations

• Additional benefit may be obtained from :

* Optimization of existing operational resources

* Adaptive processing and data selection of satellite data

• OSSEs would be useful to evaluate the impact of instruments and targeting strategies

* Collaborative effort on OSSEs based on nature run from ECMWF

* Calibration with respect to the impact of synthetic data Using OI adjoint based methods, synthetic data show a

similar quantitative impact to real observations

* Evaluation of the anticipated impact of future instruments need to be made in the context of the future observing network and future modeling and assimilation systems

Page 17: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

REAL OBS OSSE

impact

count

J/kg J/kg

Impact of various observing systems on GEOS-5 24h forecast error

OSSE calibration for Jan 2006 vs. Jan 2007 reference *First Results*

Page 18: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Value of targeted data (4)

• Targeting for longer range forecasts

* Issues for targeting at shorter range remain and should be addressed before getting into longer range forecasts

* Which flow regimes show lower predictability and what impact additional data may have.

Cardinali et al. (2007) and Langland et al. (2008) have presented results on that. Can we make this distinction?

* Evidence shows that removing or adding data does not lead to significant impact in the longer range

Experiments from Kelly et al. (2007) show that removing data from the North Pacific does not have any impact on Europe at day-6

Page 19: DAOS-WG: Data Assimilation and Observing Systems Working Group Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal

Conclusions

• Diagnosing the by-products of the assimilation* allows to estimate the information content brought by any given type of

observations on the analysis

* Same diagnostics provide information to recalibrate the error statistics used in the assimilation

* Todling (2009): observation impact is obtained from lagged forecasts• Impact of observations depend on the observing environment

* When removing some observation type, remaining observations may compensate by having more impact.

• Value of targeted data* Preparation of a review paper to summarize the results from several studies

* Optimization of existing operational resources

* Experimentation on adaptive processing and selection of satellite data• Observations

* Quality of Indian radiosondes has significantly improved

* Russian radiosondes network

* Scatterometer data and MODIS winds

* GPS radiooccultation