1. Observation Impact Workshop, Geneva, 19-21 May 2008 Intercomparison of sensitivity to observations in the context of THORPEX and the THORPEX Pacific-Asia regional campaign (T-PARC) Pierre Gauthier Department of Earth and Atmospheric Sciences Université du Québec à Montréal Contributions from Carla Cardinali (ECMWF), Ron Gelaro (GMAO), Rolf Langland (NRL), Pat Harr (NPS), Florence Rabier and Gérald Desroziers (Météo-France) Stéphane Laroche and Simon Pellerin (Environment Canada)

2. Observation Impact Workshop, Geneva, 19-21 May 2008 Introduction The THORPEX data assimilation and observation strategies working group (DAOS-WG) Intercomparison experiment on observation impact The Pacific-Asia Regional Campaign (T-PARC) Evaluating the impact of observations collected during the T-PARC –The value of targeted data Perspectives

3. Observation Impact Workshop, Geneva, 19-21 May 2008 Objectives of THORPEX DAOS-WG Impact of observations –Guidance for observation campaigns and the configuration of the Global Observing system –Evaluation of observation impact with different systems –Assessment of the value of targeted observations –Intercomparison experiment in the context of the T- PARC campaign Improving the use of satellite data –Use of sensitivity information to do adaptive data thinning –Related to the use of flow dependent background error covariances –OSSEs

4. Observation Impact Workshop, Geneva, 19-21 May 2008 The observation impact intercomparison experiment Baseline experiment –Common set of observations assimilated by all centres –Assimilation and model configurations –Metrics to measure the impact of observations Selection of period –Winter phase of the T-PARC: December 2008 to February 2009 –Period selected: January 2007 observations available were closer to what would be available during T-PARC

5. Observation Impact Workshop, Geneva, 19-21 May 2008 Observations assimilated by NRL, GMAO and ECMWF (also at Météo-France and Environment Canada) Radiosondes Dropsondes Land surface stations (all data except winds and humidity) Ship surface (winds and p s ) Aircraft (all data except humidity) AMV from geostationary satellites (no rapid-scan winds) MODIS winds AMSU-A radiances QuikScat

6. Observation Impact Workshop, Geneva, 19-21 May 2008 Comparison of the characteristics of the systems NRLGMAOECMWF Analysis T239L30 3D-Var 0.5ºx0.67ºL72 3D-Var T255L60 12-h 4D-Var ForecastsT239L30 0.5ºx0.67º L72 (Finite Volume model) T255L60

7. Observation Impact Workshop, Geneva, 19-21 May 2008 7 Sensitivity to Observations: Sensitivity to Background: Adjoint of Assimilation Equation Adjoint of forecast model produces sensitivity to Baker and Daley 2000 (QJRMS)

8. Observation Impact Workshop, Geneva, 19-21 May 2008 8 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) OBSERVATIONS ASSIMILATED 00UTC + 24h

9. Observation Impact Workshop, Geneva, 19-21 May 2008 Evaluation of the impact of observations Measure of the reduction in forecast error Evaluation at the initial time

10. Observation Impact Workshop, Geneva, 19-21 May 2008 Sensitivity with respect to analysis Configuration of the measure of forecast error –Departure with respect to a verifying analysis (each centre uses its own) –Dry adjoint model –24h (third order) sensitivity gradient (LB04), dry forecast error norm, from surface to 150hPa Forecast Sensitivity to Observation – impact at 0,6,12,18 (3D-Var or 4D-Var 6h) or 00, 12 (4D-Var 12 h)

11. Total observation impact at 00 UTC

12. Total observation impact at 12 UTC

13. Observation count x10 7

14. Observation Impact Workshop, Geneva, 19-21 May 2008 Impact per observation X10 -6

15. NAVDAS-NOGAPS Percent of observations that produce forecast error reduction (e 24 – e 30 < 0)

16. Observation Impact Workshop, Geneva, 19-21 May 2008 Other approaches to evaluate the impact of observations OSEs Information content and the degrees of freedom per signal (DFS) –DFS = tr (AB -1 ) –Reduction of analysis error

17. ECMWF WMO Observation Impact Geneva May 2008 slide 17 Observation Forecast Sensitivity Intercomparison (J/kg Dry norm - LB4 SG 0-150 hPa) and Observation Analysis Sensitivity (10 -6 )

18. Observation Impact Workshop, Geneva, 19-21 May 2008 Preliminary conclusions Numerous differences between the systems remain –Baseline experiment provided a common context against which three different systems evaluated the impact of observations with the same method –Differences persist in terms of assimilation methodologies and models (e.g., 3D-Var and 4D-Var) –The impact of observations differs from one system to another –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 (Météo-France) –Encourage other centres to participate

19. North American Region THORPEX Pacific Asian Regional Campaign (T-PARC) David Parsons Co-chair,North American THORPEX Regional Committee Contributions from Tetsuo Nakazawa, Dehui Chen, Pat Harr, Istvan Szunyogh, Anna Agusti-Panareda, Sarah Jones, Martin Weismann, Carla Cardinali, etc…

20. North American Region Western WA Flood (Seattle 1-day record) CA Wild Fires (downslope winds) BC’s flood of the Century (18.5”) What is happening in this region?

21. Major scientific issues for T-PARC Tropical cyclogenesis –Better understand the large-scale influences on cyclogenesis and their relation to cyclone structure –To examine the predictability of cyclogenesis and develop strategies to improve forecast skill –To examine the evolution and the role of convection during cyclogenesis Recurvature –To understand dynamic/thermodynamic environmental fields which affect TC recurvature –To better understand ensemble spread and improve the utilization of ensemble information in disaster mitigation –To develop, refine typhoon targeting capabilities with the goal of improving regional and downstream predictions

22. Major scientific issues for T-PARC Extra-Tropical transition –Factors limiting the regional and global predictability of the interaction between the tropical cyclone and the mid-latitude flow –Structural changes in the tropical cyclone core during the ET process and how these changes are related to the evolution of the distribution of precipitation –to develop and test observational, assimilation and modeling strategies to improve local and downstream predictive skill for ET events Winter storms –to develop and test new adaptive observation strategies for winter systems that overcome the current limitations of aircraft targeting –to better understand and predict Rossby wave triggering and enhancement in the Pacific wave guides –to extend the adaptive use of in-situ and satellite observations to medium range prediction

27. Proposing Institutions North America –US Academic Community: SUNY at Stony Brook, U. of Hawaii, Naval Post Graduate School, U. of North Carolina Charlotte, Pen. State, U. of Washington, U of Maryland, SUNY Albany, U of Miami, U of Wisconsin, Florida State U –US Research Institutions: NCAR, NOAA/NCEP, NOAA/NWS, Naval Research Lab, NASA/Goddard –Canada: Environment Canada Asia –China: Chinese Academy of Meteorological Sciences, Chinese Meteorological Administration plus members of the Academic Community in China –Japan: Japan Meteorological Agency, Japan Marine Science and Technology Center (JAMSTEC), Kyoto U, Nagoya U, Tohoku U, Tsukuba U, U of Tokyo –Korea: Korean Meteorological Administration, Cheju National U, Ehwa Womans U, Kongju National U, Kyungpook National U, Seoul National U,Yonsei –Collaboration with an expanded DOTSTAR program Europe –Germany: U of Karlrsuhe, Institut für Physik der Atmosphäre, DLR, –Others (ECMWF and National Met Centers)

28. T-PARC and Observing System/Observing Strategies Research Typhoon genesis –Relevant science Impact of assimilating new type of measurements on typhoon genesis (radar reflectivity, winds in clear air and clouds, synoptic style in-situ obs vs dropsondes, rapid scan satellite obs) Evaluation of initial condition and model error in genesis regions Advancing knowledge of the genesis process and the factors limiting predictive skill –Instruments NRL P-3 with ELDORA Doppler radar, dropsondes, ocean SST and (perhaps) a Doppler lidar for mesoscale (US) Driftsonde and tropical island radiosonde sites for large-scale for 2007 and 2008 (proposed China, France, and US)

29. T-PARC and Observing System/Observing Strategies Research Typhoon landfall and recurvature –Relevant science Advance understanding and test the impact of typhoon in-situ and satellite targeting strategies Improve typhoon prediction and society’s utilization of typhoon forecast products Impacts of future remote sensing strategies from space (winds, water vapor, radars with frequent updates)

30. T-PARC and Observing System/Observing Strategies Research Typhoon landfall and recurvature –Instruments NRL P-3 with ELDORA Doppler radar, dropsondes, ocean SST and (perhaps) a Doppler lidar for mesoscale (US) DLR Falcon with Doppler lidar, water vapor lidar and dropsondes Dropsonde aircraft (China, DOTSTAR, Japan, Korea?) Driftsonde for 2007 and 2008 (proposed China, France, US) Coordination of Chinese IOPs over land: SCHeREX-“973” basic research project; Tibet-Est surrounding Exp., 4 MeTebs of LaSW (Guangzhou, Wuhan, Anhui and Shanghai) Rapid scan MTSAT satellite observations Collaboration with extended DOTSTAR program (dropsonde aircraft and driftsonde)

31. T-PARC and Observing System/Observing Strategies Research Extra-Tropical transition –Relevant science Advance understanding and test the regional and downstream impacts of targeted measurements (typhoon vs middle latitude) by in-situ and satellite measurements Impacts of future remote sensing strategies from space (winds, water vapor, radars with frequent updates) Understanding the factors that limit predictability

32. Forecast Skill Bifurcation ET Tracks From Jones et al., 2003: Wea. And Forecasting General Decrease in Forecast Skill for ET Storms

33. T-PARC and Observing System/Observing Strategies Research Extra-Tropical transition –Instruments NRL P-3 with ELDORA Doppler radar, dropsondes, ocean SST and (perhaps) a Doppler lidar for mesoscale (US) DLR Falcon with Doppler lidar, water vapor lidar (US, Germany, Japan?, Korea?, Canada?) and dropsondes Dropsonde aircraft (China, DOTSTAR, Japan, Korea?) Driftsonde for 2007 and 2008 (proposed China, France, US) Coordination of Chinese IOPs over land: SCHeREX-“973” basic research project; Tibet-Est surrounding Exp., 4 MeTebs of LaSW (Guangzhou, Wuhan, Anhui and Shanghai) plus other Asian sounding networks Rapid scan MTSAT satellite observations Collaboration with extended DOTSTAR program (dropsonde aircraft and driftsonde

34. T-PARC and Observing System/Observing Strategies Research Winter phase –Instruments Upgraded and enhanced Russian radiosonde network and continuation of some Chinese land-based sounding enhancements (Tibetan Plateau) US NOAA G-IV with dropsondes (western Pacific) Air Force C-130’s with dropsondes (central Pacific) NOAA P-3 or other assets in the eastern Pacific NOAA and NASA satellites Relevant science Value of sensitivity information for targeting and adaptive data selection strategies Led by Zoltan Toth

35. Summary and conclusion (1) New approaches are being investigated to evaluate the impact of observations on the quality of forecasts –Forecast sensitivity to observations Adjoint based approaches Ensemble methods (e.g., ETKF) DFS and information content –Objective is to obtain robust and reliable methods to evaluate the impact of observations on the quality of weather forecasts Intercomparison experiment –Numerous components are involved associated with model, observations, assimilation methods and flow regimes –Intercomparison experiment has value in that it reduces several of the differences to bring the systems on a common ground (e.g., observations used, flow regimes, resolution) –Calibration of assimilation systems raises some questions about the value of ‘degrading’ a system in that context

36. Summary and conclusion (2) Value of data deployed during T-PARC –Experiment aims at capturing the different stages of Tropical cyclones from their genesis to their migration into northern latitudes –Value of data over the Pacific for the short to medium-range forecasts over Asia and North America –Meteorological high-impact events in Asia and North America Data assimilation objectives –Assess the impact of observations on deterministic and probabilistic forecasts –Targeting techniques and adaptive satellite data assimilation –Large sets of data will be made available that could be used to better use satellite data in those situations

37. Thank you

38. Observation Impact Workshop, Geneva, 19-21 May 2008 Other objectives Research on model error modeling and estimation –Considered to be a necessity for model of increasing resolution, convection, cloud representation ECMWF: weak-constraint 4D-Var with long assimilation windows –Biases need to be addressed too –Explore possibilities of using TIGGE framework to estimate model and background error characteristics Observation error correlation –Design of observation campaign to estimate observation error statistics –Identify existing Cal/Val campaigns with similar objectives (in collaboration with the Obs WG) –Make it known what exactly the assimilation needs in terms of observation error characteristization Data assimilation in the Tropics –THORPEX and AMMA