The Exploitation of Satellite Data at the U.K. Met Office Roger Saunders with the help of Steve English Bill Bell, Mary Forsythe, Brett Candy, James Cameron and many others © Crown copyright 2006 Page 1 Talk topics The Met Office satellite group and NWP SAF Current status of models and observation use Recent upgrades to use of satellite data Satellite data impacts Research into use of new data types Meteosat Second Generation METOP © Crown copyright 2006 Page 2 Satellite Group Head John Eyre Radiance Imagery for NWP Active sensors Data Processing Hi res models assimilation and forecasters Dave Offiler Dick Francis Bryan Conway Steve English Roger Saunders Reading SatelliteSatelliteSatelliteSatelliteNWP/nowcasting/imagery Radiance Imagery ActiveApplications ApplicationAssimilationSensing Reading Group Group Group © Crown copyright 2006 Page 3 Satellite Group Funding SA Funding 2006/07 31 Staff in total Core Govt EUMETSAT Defence Aviation ESA Climate © Crown copyright 2006 Page 4 NWP Satellite Application Facility The Met Office leads the NWP SAF In final year of initial operational phase Preparing proposal for follow-on operational phase (5 years) to start in March 2007. Major deliverables are: AAPP (ATOVS/AVHRR direct readout software) RTTOV (Fast radiative transfer model) 1DVAR (Met Office and ECMWF versions) Satellite data monitoring (Radiance, AMVs, O3) Scatterometer processor Reports on many aspects of satellite data Also involved in GRAS (GPS RO) SAF © Crown copyright 2006 Page 5 Summary of model –AIRS observations © Crown copyright 2006 Page 6 Talk topics The Met Office satellite group Current status of models and observation use Recent upgrades to use of satellite data Satellite data impacts Research into use of new data types Meteosat Second Generation METOP © Crown copyright 2006 Page 7 Met Office Operational Models : 2006 ¾Global ~40km ~50Level ¾North Atlantic/Europe 12km ~30level ¾UK 4km ~30Level ¾Re-locatable Defence and Civilian ¾Trial Ensemble (global & regional) at half horizontal resolution ¾Data assimilation 4DVar 6 hour window for global and regional models © Crown copyright 2006 Page 8 Operational data usage (Apr 2006) Observation group Observation Items used Daily extracted % used in Sub-group assimilation Ground-based TEMP T, V, RH processed to 1250 87,92,50 vertical profiles PILOT model layer average 850 92 As TEMP, but V only PROFILER As TEMP, but V only 6500 40 Satellite-based AMSU-A/B Radiances directly 430000 3 vertical profiles NOAA-15/16/18 assimilated with channel selection dependent on Aqua AIRS surface instrument and cloudiness Aircraft Manual T, V as reported with 24000 14 AIREPS duplicate checking and 185000 28 blacklist Automated AMDARS Satellite GOES 10,12 BUFR High resolution IR winds 110000 10 atmospheric Meteosat 5, 8 BUFR IR, VIS and WV winds 190000 5 IR and VIS winds motion vectors MTSAT SATOB IR and WV 4000 55 Aqua/Terra MODIS Satellite-based SSMI-13,15 In-house 1DVAR wind-speed 3000000 1 surface winds Seawinds, ERS-2 retrieval 1800000 1.5 NESDIS retrieval of ambiguous winds. Ambiguity removal in 4DVAR. Ground-based Land SYNOP Pressure only (processed to 28000 75 surface SHIP, Fixed Buoy model surface) 6700 94, 92 Pressure and wind Drifting BUOY 10000 76 © Crown copyright 2006 Pressure Page 9 Talk topics The Met Office satellite group Current status of models and observation use Recent upgrades to use of satellite data Satellite data impacts Research into use of new data types Meteosat Second Generation METOP © Crown copyright 2006 Page 10 Recent changes to use of satellite data Aqua AMSU-A replaced by NOAA-18 AMSU Better coverage with NOAA-18 Issue of Aqua AMSU-A antenna correction AIRS central fov replaced by AIRS warmest fov Reintroduction of ERS-2 scatterometer winds Meteosat-7 AMVs replaced by Meteosat-8 AMVs © Crown copyright 2006 Page 11 Satellite data delays Main run data cutoff SSMI F15 SSMI F14 SSMI F13 MODIS Terra MODIS Aqua SSMIS Max Delay QuikScat Average Delay AIRS ATOVS N18 ATOVS N16 ATOVS N15 ATOVS Aqua 0 60 120 180 240 300 360 420 480 540 Delay (minutes) © Crown copyright 2006 Page 12 Average AMSU Data Assimilated for a Main forecast Run Control Experiment N15 N16 N18 Aqua Total N15 N16 N18 Aqua Total 14000 14000 12000 12000 10000 10000 8000 8000 6000 6000 4000 4000 2000 2000 0 0 © Crown copyright 2006 Page 13 EARS-Mitigating Data Delay EARS System Overview © Crown copyright 2006 Page 14 EARS (2) EARS timeliness © Crown copyright 2006 Page 15 EARS (3) There are plans to: EARS coverage • extend coverage of EARS to the whole extra-tropical NH • develop other regional systems (e.g. S. America, Australia / NZ) © Crown copyright 2006 Page 16 Arrival Times of Data Global 109229 obs EARS 93979 Obs 100 • North Atlantic Main Run Update Run Region 90 NESDIS 25% both 100% • six-hour window EARS 75% 80 09/09/2003 09:00- 15:00 70 60 50 40 Percentage of Observations Percentage 30 20 First Overpass 10 0 09 00 10 00 11 00 12 00 13 00 14 00 15 00 16 00 17 00 18 00 © Crown copyright 2006 window Page 17 Moved to use AIRS warmest FOV Centre FOV Warmest FOV any of 9 This results in an increased coverage of AIRS clear sky radiances © Crown copyright 2006 Page 18 Removed Double Peaked AIRS Channels Channel 2107 at 2386cm-1 (4.19 microns), FWHM 1.880cm-1 The sharp, strong absorption CO2 lines cause a double peak in the Jacobian. © Crown copyright 2006 Page 19 AIRS q increments No AIRS AIRSWF © Crown copyright 2006 Page 20 Impact on AMSU-B Channels 183 ± 7 GHz First Iteration No AIRS AIRSWF Final Iteration No AIRS AIRSWF © Crown copyright 2006 Page 21 Summary Assimilating AIRS leads to significant humidity changes. All AIRS trials show improved fits to SSMI TCWV (not assimilated) and all AMSU-B channels. An improvement to RH is not confirmed by sondes, where no particular effect is apparent. We have not seen a big improvement in forecasts going from AIRS to AIRSWF. © Crown copyright 2006 Page 22 What is the NWP SAF AMV monitoring? NWP SAF –Numerical Weather Prediction Satellite Application Facility A EUMETSAT-funded initiative, led by the Met Office, with partners ECMWF, KNMI and Météo-France Aim to improve the benefits derived by European National Met. Services from NWP by developing techniques for more effective use of satellite data and to prepare for effective exploitation of new data/products. AMV Monitoring Displays comparable AMV monitoring output from different NWP centres in order to help identify and partition error contributions from AMVs and NWP models. Analysis reports produced periodically (planned every 2 years to coincide with winds workshops). Intended to stimulate discussion and to lead to improvements in AMV derivation and AMV use in NWP. http://ww© Crown copyrightw. 2006metoffice.gov.uk/research/interproj/nwpsaf/satwind_report Page 23 Data used - Mar 2006 MODIS polar winds (TERRA and AQUA) 60N equator 60S GOES-10 GOES-12 Meteosat-8 Meteosat-5 MTSAT-1R 1. 8th Feb 05 - Introduce NESDIS MODIS polar winds 2. 14th Jun 05 – GOES SATOB IR -> GOES BUFR IR SATOB and cloudy WV st BUFR 3. 1 Sep 05 – start using MTSAT-1R SATOB th © Crown copyright 2006 4. 14 Mar 06 – Meteosat-7 -> Meteosat-8 Page 24 Meteosat-7 Meteosat-8 Replace Meteosat-7 IR, WV and VIS with Meteosat-8 IR10.8, WV7.3, HRVIS and VIS0.8 using same QC as applied to Meteosat-7 Root mean square vector difference 100-400 hPa 400-700 hPa 700-1000 hPa Met-7 IR Met-8 IR 10.8 Impact neutral in Jul-Aug and moderately positive in Dec-Jan. Main benefit from tropical wind fields (W250 and W850) and southern hemisphere PMSL (against both observations and analyses). Few forecast parameters improved or degraded by more than 2%. © Crown copyright 2006 Page 25 MODIS winds © Crown copyright 2006 Page 26 Forecast error evolution Aug 14th, 2004 500 hPa geopotential height CONTROL MODIS T+24 T+48 T+24 T+48 T+72 T+96 T+72 T+96 T+120 T+144 T+120 T+144 © Crown copyright 2006 Page 27 AMVs Why do we struggle to see impact from AMVs? Is it poor height assignment of some winds? We are starting to investigate this Discussed at IWW in Beijing © Crown copyright 2006 Page 28 Comparison to model best-fit pressure 2 2 Vector Differencei = √((ObU – BgUi) + (ObV – BgVi) ) 100 200 Model best-fit at minimum in vector 300 difference profile. 400 AMV pressure and model best-fit pressure 500 agree well in this case. 600 Pressure (hPa) 700 u component 800 v component Diff 900 1000 -20 0 20 40 60 © Crown copyright 2006 m/s Page 29 Case studies – Sahara 8th Dec MSG wind height (hPa) MSG wind – model best fit (hPa) MODIS Cloud Top Pressure (hPa) From 7th December 21:00 UTC to 8th December 2005 3:00UTC (night time) Can also compare to other cloud top pressure products © Crown copyright 2006 Page 30 Talk topics The Met Office satellite group Current status of models and observation use Recent upgrades to use of satellite data Satellite data impacts Research into use of new data types Meteosat Second Generation METOP © Crown copyright 2006 Page 31 DirectDirect radianceradiance assimilationassimilation ATOVSATOVS && 3D-var3D-var © Crown copyright 2006 Page 32 Comparison of impact of observing sounding data 30 No Satellite Losing all microwave sounders Losing all infrared sounder Losing all radiosonde T, q and u Global degradation Losing all radiosonde T 01020 and q Analysis Ten years ago? TOVS NESDIS retrievals, AMV, © Crown copyright 2006more but lower quality radiosondes Page 33 Forecast skill vs time More forecast busts without satellite data © Crown copyright 2006 Page 34 Talk topics The Met Office satellite group Current status of models and observation use Recent upgrades to use of satellite data Satellite data impacts Research into use of new data types Meteosat Second Generation METOP © Crown copyright 2006 Page 35 Work in progress….