JCSDA Observation System Simulation Experiments (OSSE) Plan for GOES-R Series Sounding Mission Fuzhong Weng, STAR/SMCD Stephen Lord, NCEP/EMC Lars Peter Riishojgaard, NASA/GMAO July***, 2007 With Suggestion by Michiko 1
Outlines Project Objectives Methodology and System Design Work Plan Project Schedule Project Milestones and Deliverables Expected Outcomes Joint Center for Satellite Data Assimilation 2
Project Objectives: Use Full OSSEs as a Quantitative Tool for Observing System design and planning OSSEs integrate new instruments into Data Assimilation System, thereby providing complete preparation for their use (JCSDA goal) Apply OSSE system to GOES-R and other advanced instrument candidates to demonstrate the potential added value of High resolution of vertical temperature and moisture structures from GOES-R hyperspectral infrared soundings for regional forecasting (e. g. flash flood and convection) Atmospheric wind products (e. g. GOES-R infrared cloud and water vapor winds) versus a direct wind measurement Geostationary microwave sounder Joint Center for Satellite Data Assimilation 3
Project Objectives: Provide quantitative evidence to guide future instrument initiatives for both atmosphere and oceans There are many simulation experiments. Only full OSSEs are able to provide a full scenario of data impact. Ø Forecast impact as well as analysis impact Ø Testing configuration of observing systems Joint Center for Satellite Data Assimilation 4
Methodology and System Design Introduction to OSSE Basic Concepts Real /OSSE Data Assimilation System Existing Real Observations With & Without Truth Nature Observations Nature Run New & Existing Observations With & Without Data Assimilation Analysis Verification Forecast Model Joint Center for Satellite Data Assimilation 5
JCSDA OSSE Partnership Interagency NCEP, STAR, NASA (GMAO, GLA, SIVO), ESRL collaboration 10 years of experience International Effort ECMWF, KNMI joined Joint OSSE studies last 2 years European and Asian community interests growing Universities and NESDIS Corporative Institutes UWISC/CIMSS (mesoscale OSSE) CSU/CIRA (mesocale OSSE) MSU/GRI Univ. of Utah Initial Focus Global forecast impacts of HES regional impacts on hurricane and high-impact weather events Joint Center for Satellite Data Assimilation 6
Methodology and System Design Nature Run: provides observations and truth for OSSE Global: Joint OSSE Nature run with. T 511 and T 799 by ECMWF (Equivalent to approximately 25 km and 15 km grid models) High resolution Nature Run with less than 5 km grid model later Assimilating Forecast Models: Global Forecast System – GFS GOES-5 at NASA/GMAO Weather Research and Forecasting Model – WRF Data Assimilation Systems Gridded Statistical Interpolation (GSI) for GFS, GEOS-5, WRF Advanced 4 D-Var techniques in GSI (4 DSV) Case Selection Hurricane and severe weather events in ECMWF Nature Run Downscaled cases (globally) using Regional Models Joint Center for Satellite Data Assimilation 7
Remark Severe weather cases don’t have to be local to US in OSSEs. We can make GOES-R sounder data anywhere on the planet (one of the nice things about OSSEs). While the hypothesis is that Geo IR soundings can improve prediction of severe weather events, one must not prejudge the result. Many have doubts as you might know. Better stated as a hypothesis: “what type of observation(s) will improve the prediction of severe wx events? Much fine resolution data will produce a larger scale impact by Super-Obbing effect. Joint Center for Satellite Data Assimilation 8
Specific Events for GOES-R OSSE GOES-R Series Sounder: Improve prediction of severe storms GOES-R OSSE must realistically simulate preconvection environments (moisture features) resolve cloud and precip structures produce adequate temporal/spatial sampling Joint Center for Satellite Data Assimilation 9
Special Requirements for GOES-R OSSE Design Requirements for Nature Run Systems High spatial/temporal resolution (realistic moisture features) Advanced physics Supercomputing environments Synthetic radiance simulations and validation Advanced radiative transfer models Error models for sampled observations from Nature Run Instrument noise properties Data Assimilation Advanced (4 dvar) techniques to use more temporal information and adjoint NWP physics Control run in which all the simulated data paralleling the current operational observational data stream are included Perturbation run in which the simulated candidate observations under evaluation are added Impacts Analysis Comparison of forecast skill between the control and perturbation runs Standard performance scores (focused on applications such as hurricane track and intensity, Aviation weather…. ) Subjective evaluation Joint Center for Satellite Data Assimilation 10
Work Plan: Phase I Preparation of Nature Runs Low Res Joint OSSE Global NR: 13 month T 511 (ECMWF) High Res Joint OSSE Global NR: two 5 week T 799 (ECMWF) Regional: CSU/RAMS (Candidate) Super high res Global NR Validation of Nature Runs Temperature/moisture/wind Cloud coverage: GOES imagery vs. simulations Cloud liquid/ice statistics: Cloudsat/Calipso vs. Simulations Initial OSSE Experiment demonstrating the assimilation of modeled temperature and humidity fields extracted from the Nature Run Joint Center for Satellite Data Assimilation 11
Work Plan: Joint OSSE Nature Runs Joint OSSE T 511 Nature Run Produced by ECMWF Equivalent approximately 25 km grid point model 40 km resolution in physics 91 vertical layers 3 hourly output from May 2005 to June 2006 integration Realistic extratropical storm frequency and statistics, hurricane and tropical waves Improved cloud Suitable for global OSSEs Joint OSSE T 799 Nature Run Produced by ECMWF Equivalent approximately 15 km grid point model 25 km resolution in physics 91 vertical layers Hourly outputs for two 35 day periods Better hurricane and severe storm seasons Suitable for most mesoscale OSSEs and to test synoptic and mesoscale impacts of GOES-R Joint Center for Satellite Data Assimilation Lifespan distribution of extratropical cyclones during February 2006 in Northern Hemisphere. Red bars are for NR. Green bars are for NCEP analysis 12
Need for higher resolution Nature Run Need for a Nature Run with higher resolution mesoscale OSSEs Hurricanes, lake snow effects, severe storms Less than 5 km model (without cloud parameterization) Frequency of output : 5 min Candidates Global cloud resolving model GFDL-ESRL (Planned delivery time 2012) NICAM Local high resolution global model Using Fibonacci grid Nested regional model CSU RAMS (regional atmospheric modeling system) RUC WRF Joint Center for Satellite Data Assimilation 13
Work Plan: Possibility for Regional OSSEs Possibilities for Regional Nature run Performance of high resolution regional models needs to be evaluated Noise from boundary conditions must be evaluated A 5 day Nature Run with resolution of 1 -4 km and 500 x 35 Possibly using the RAMS model (Other candidates: RUC, WRF and more) Waiting for a Global high resolution model is a possibility Validation of Regional Nature Run A major challenge Nature Run must produce statistically representative atmospheric state Major validation effort needed Cloud coverage: GOES imagery vs. Simulations Cloud liquid/ice statistics: Cloudsat/Calipso vs. Simulation Adequate regional data assimilation system beyond current state of science Non-hydrostatic atmosphere Analysis balance constraints Time dependency (4 D-Var) still under development “ 5 -year” development? ? Joint Center for Satellite Data Assimilation 14
Work Plan: Ocean OSSE State of science less developed than atmosphere Realistic real-time ocean models just coming on line Navy to take leading role? ? Joint Center for Satellite Data Assimilation 15
Work Plan: Phase II Conducting the OSSE Complete validation for ECMWF Nature Run Construct Conventional observations • RAOB, Air Craft, Cloud Track Wind Satellite radiances for all existing instruments • AQUA, IASI, ASCAT Observations from future instruments • Radiances AND • Retrieved temperature and humidity profiles – Observation errors generated by the retrieval method • Doppler Wind Lidar Calibration process Demonstrate impact of known instruments is statistically comparable in both Real and Nature Run worlds Joint Center for Satellite Data Assimilation 16
Work Plan: Phase III Conducting the OSSE Direct radiance assimilation Forward model and instrument errors Uses of temporal information Low and high spatial resolution data assimilation system Demonstrate the potential importance of all information derived from GOES-R instrument suite Joint Center for Satellite Data Assimilation 17
Expected Outcomes: Benefits to High-Impacts Events from Uses of GOES-R Sounders in NWP Reduced errors in predicting hurricane Intensity Reduced errors for predicting hurricane track Improved prediction of surface precipitation Reduced RMSE of temperature and water profiles at different forecast times Joint Center for Satellite Data Assimilation 18
Expected Outcomes: NWP Operational Readiness for GOES-R Provide the research environment and computational infrastructure necessary to assess operational and research computing needs for effectively transitioning GOESR data into operational use Establish an end-to-end process for testing and ingesting GOES-R data in NWP models Develop a GOES-R VAR system that will use high temporal/temporal information State-of-the art GOES-R Community Radiative Transfer Model 4 DVar is performed to assimilate the most recent observations, using a segment of the previous forecast as the background. This updates the initial model trajectory for the subsequent forecast Joint Center for Satellite Data Assimilation 19
JCSDA Community Radiative Transfer Model fullly Upgraded for GOES-R Joint Center for Satellite Data Assimilation VIS IR 20
Expected Outcomes: Fully Validated Nature Global and Mesoscale Nature Runs Used for other future GOES-R instrument OSSEs Used for NOAA satellite recapitalization plan Simulated GOES-R ABI 10. 35 micron band at 2 -km grid spacing Simulated GOES-R Joint Center for Satellite Data Assimilation spacing ABI 3. 9 micron band at 2 -km grid 21
Remarks Schedule is too ambitious Not attainable without additional computing resources Augment NOAA R&D computer Disk Cpu Increase funds for March 2008 upgrade Joint Center for Satellite Data Assimilation 22
Project Schedule August 1 2007 – GOES-R OSSE Kick-off meeting, NOAA Science Center December 8, 2007 – Preliminary Results Review, NESDIS HQ September, 2008 – Mid-term Review, NOAA September, 2009 – Final Review, NOAA Joint Center for Satellite Data Assimilation 23
Project Milestones and Deliverables 2007 – Complete preparations for global OSSEs and prepare for regional OSSE Create synthetic observations for conventional and all current satellite instruments Run calibration experiments for conventional and current satellite instruments Create synthetic IASI observations Apply ECMWF T 799 Nature Run for preliminary OSSE results for IASI Create 1 mesoscale Nature Run with validation 2008 – Complete additional 1 mesoscale Nature Run 2009 – Document impacts of GOES-R HES on both regional and global NWP forecast skill Joint Center for Satellite Data Assimilation 24
Remarks Need to mention FY 10 request, part of which supports an expanded OSSE capability Joint Center for Satellite Data Assimilation 25
Budget 2007 – 400 K 2008 – 500 K 2009 – 500 K Joint Center for Satellite Data Assimilation 26
Backslides: On-going Research on Validation of Joint OSSE global nature runs at NCEP Opportunities for Regional OSSE Joint Center for Satellite Data Assimilation 27
Generation of OSSE Synthetic Observations ECMWF 799 RAMS Nature runs Instrument Forward model (RTA) Atmosphere simulation Merge Orbit simulation IGBP land cover map AVHRR NDVI Digital Elevation Model Field of view simulation dataset Surface property simulation Joint Center for Satellite Data Assimilation Instrument Noise model Simulated observation
Simulated Cr. IS Observation (NPOESS) Joint Center for Satellite Data Assimilation 29
OSSE Flowchart and Validation Data assimilation data Data assimilation Withdraw or Experimental data Add data for Control data Real observed existing instruments. Real observed data Analysis impact test Analysis Forecast OSE using Real Data Analysis Forecast impact test Forecast Joint Center for Satellite Data Assimilation 30
Experimental data Simulation of Withdraw or data Control data add data for data existing instruments Simulated data Calibration Simulated analysis and forecast impacts are compared with real impact Data assimilation Simulated analysis and forecast are also evaluated against the Nature Run Data assimilation Simulated Nature Run Evaluation of new Instruments Analysis impact test Nature Run Analysis impact test Analysis Forecast impact test Forecast Control data + Simulated data for future instruments Data assimilation OSSE using simulated data Forecast Joint Center for Satellite Data Assimilation Analysis Forecast impact test Forecast 31
Real Simulation Nature Run Experimental data Real observed data add data for Nature Run Simulated analysis and forecast impacts are compared with real impact Analysis Forecast Data assimilation Simulated analysis and forecast are also evaluated against the Nature Run Data assimilation Simulated Control data Real existing instruments data observed Simulated data Calibration Evaluation of new Instruments Analysis impact test Analysis impact test Forecast impact test Control data + Simulated data for future instruments Data assimilation Simulation of Withdraw or data Forecast impact test Forecast Joint Center for Satellite Data Assimilation Forecast 32
Validation of ECMWF 511 nature Run Joint Center for Satellite Data Assimilation 33
Work Plan: Phase 2 Preparation A period of severe weather in T 511 NR will be identified Model (GFS, WRF) and data assimilation scheme (GSI and WRFVar) Nature run T 799 NR: Two 35 days long Capable of producing realistic hurricanes and frontal zones Validation of Nature Run Cloud statistics and radiance Tropical easterly waves Hurricane tracks Cyclone statistics Rossby waves Initial OSSE Data impact on synoptic waves Data impact on Hurricane tracks and development Data impact on frontal zones Joint Center for Satellite Data Assimilation 34
Work Plan: Phase 3 Preparation A period of 5 days for the OSSE will be identified A short description of the nature model, forecast model (WRF) and data assimilation scheme (WRF-Var) Nature run A 5 day nature run with 4 km resolution and 100 levels Grid points nonhydrostatic cloud resolving model Validation of nature run Cloud coverage: GOES imagery vs. Simulations Cloud liquid/ice statistics: Cloudsat/Calipso vs. simulation Initial OSSE Experiment of assimilating modeled temperature and humidity fields extracted from the nature run Joint Center for Satellite Data Assimilation 35
Скачать презентацию JCSDA Observation System Simulation Experiments OSSE Plan for
9d27d08e4c2e3d045e35e646ea359835.ppt