Monitoring SMOS brightness temperatures at global scale. A preliminary overall quality assessment   Thanks to: ECMWF opera...
Outline <ul><li>ECMWF main objectives using SMOS data, </li></ul><ul><li>Some aspects about the implementation of SMOS dat...
Outline <ul><li>ECMWF main objectives using SMOS data, </li></ul><ul><li>Some aspects about the implementation of SMOS dat...
Main objectives <ul><li>Global monitoring of Level-1C brightness temperatures at the satellite antenna reference frame, at...
Main objectives <ul><li>Assimilation of SMOS Level-1C brightness temperatures over land     investigate the meteorologica...
Implementing SMOS data in the IFS. First version ODB: Observations Data Base used by the Integrated Forecasting System
<ul><ul><li>Volume of SMOS data, </li></ul></ul><ul><ul><ul><li>Much computing resources and time is needed to process and...
L1C-NRT BUFR product Convert to L1C-NRT  ECMWF BUFR product <ul><li>Pre-process data: </li></ul><ul><li>Consistency checks...
$Instrument_$SensingTime1_$SensingTime2_$Satellite_$orbit_$datatype_$GeneratingTime_$datalevel.bufr NRT latency – December...
NRT latency – Jan-Feb-Mar-Apr 2010 January February March April
NRT latency –  May-June-July 2010 May June July
<ul><ul><li>SMS Supervisor Monitor Scheduler </li></ul></ul><ul><ul><li>Routinely checks, </li></ul></ul><ul><ul><li>Valid...
Outline <ul><li>ECMWF main objectives using SMOS data, </li></ul><ul><li>Some aspects about the implementation of SMOS dat...
Observations monitoring  <ul><ul><ul><li>, SMOS offline monitoring webpage </li></ul></ul></ul><ul><ul><ul><li>Available s...
TBxx TByy 20-Nov-09 20-Dec-09 16-Jan-10 Θ = 40° NRT NRT
Monitoring observations – Urulu (Australia) TBV
First - guess First-guess    CMEM initial config  <ul><ul><li>Community Microwave Emission Model (CMEM), modular radiativ...
First-guess departures  (obs - model) ‏ H-pol V-pol <ul><li>After implementation bugs removal, some departures are still t...
First-guess departures (obs – model) ‏ H-pol FG departures Orography Case Study:  22 January 2010 FG departures > 75K
First-guess departures (obs – model) ‏ H-pol FG departures incidence angle Case Study:  22 January 2010 FG departures < -1...
First-guess departures  V-pol Case Study:  22 January 2010 -25 K < FG departures < 25 K
Preliminary assessment of main FG departures  <ul><li>Departures too large     observations >> model : </li></ul><ul><ul>...
Global statistics over land  Average of Observations, TBH polarisation, spatial resolution 0.25 °   01-07 June
Global statistics over oceans Average of Observations, TBV polarisation, spatial resolution 1°  01-07 April Incidence angl...
Global statistics over oceans  Number of Observations per grid square, TBV polarisation, spatial resolution 1°  01-07 Apri...
Histogram of departures over oceans <ul><li>03 June 2010, </li></ul><ul><li>All observations over open seas (first 4D-Var ...
<ul><li>Maps of Observations Standard Deviation (STD) </li></ul><ul><li>TB STD [K] </li></ul><ul><li>H-pol </li></ul><ul><...
Global statistics: Standard monitoring maps  Map of Mean First Guess Departure over land (Obs – Model) 01-07 March
Global statistics : Standard monitoring maps  Map of Mean First Guess Departure over land (Obs – Model) 01-07 April
Global  statistics: Standard monitoring maps  Map of Mean First Guess Departure over land (Obs – Model) 03-10 May
Global statistics: Standard monitoring maps  <ul><ul><li>RFI impact on FG departures STD is large,  </li></ul></ul><ul><ul...
<ul><li>STD of OBS </li></ul><ul><li>01-07 June </li></ul><ul><li>H-pol </li></ul><ul><li>V-pol </li></ul>Global statistic...
Global statistics: Time Series  <ul><ul><li>TB average (OBS) </li></ul></ul><ul><ul><li>Nbr observations </li></ul></ul><u...
<ul><ul><li>ECMWF main objectives using SMOS data are:  monitoring  and  data assimilation </li></ul></ul><ul><ul><li>Impl...
<ul><ul><li>On going activities:  </li></ul></ul><ul><ul><ul><li>Re-structuration of internal data base to efficiently han...
Thanks   !
Back-up slides
Faraday Effect over land  H_pol TBH with Faraday – TBH without Faraday
Faraday Effect over land  TBH with Faraday – TBH without Faraday V_pol
Observations monitoring – Sea West Coast Australia  TBxx TByy
Monitoring observations – Urulu (Australia) TBxx TByy
Monitoring observations – Urulu (Australia) TBV TBH
Global statistics: Standard monitoring maps  Map of STD of  First Guess Departure over land (Obs – Model) 01-07 March
<ul><li>STD of first-guess departures </li></ul><ul><li>01-07 april </li></ul><ul><li>H-pol </li></ul><ul><li>V-pol </li><...
Global statistics: Time Series  Routine production of time series in different geographical areas  Global area,  02-09 May...
Global statistics over oceans  Standard deviation of Observations, TBV polarisation, spatial resolution 1°  01-07 April In...
Upcoming SlideShare
Loading in …5
×

FR2.L10.1: MONITORING SMOS BRIGHTNESS TEMPERATURES AT GLOBAL SCALE. A PRELIMINARY OVERALL QUALITY ASSESSMENT.

540 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
540
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
17
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

FR2.L10.1: MONITORING SMOS BRIGHTNESS TEMPERATURES AT GLOBAL SCALE. A PRELIMINARY OVERALL QUALITY ASSESSMENT.

  1. 1. Monitoring SMOS brightness temperatures at global scale. A preliminary overall quality assessment Thanks to: ECMWF operations team, Matthias Drusch (ESA/ESTEC), Susanne Mecklenburg (ESA/ESRIN), Steven Delwart (ESA/ESTEC), Norrie Wright(ESA/ESRIN), Philippe Richaume (CESBIO). Joaquín Muñoz Sabater Anne Fouilloux, Patricia de Rosnay, Mohamed Dahoui
  2. 2. Outline <ul><li>ECMWF main objectives using SMOS data, </li></ul><ul><li>Some aspects about the implementation of SMOS data in the Integrated Forecasting System (IFS), </li></ul><ul><ul><li>Main challenges of the implementation , </li></ul></ul><ul><ul><li>NRT product latency, </li></ul></ul><ul><ul><li>Pre-processing, </li></ul></ul><ul><li>SMOS data: preliminary results </li></ul><ul><ul><li>SMOS offline data monitoring webpage – observations monitoring, </li></ul></ul><ul><ul><li>Preliminary assessment of main first-guess departures sources, </li></ul></ul><ul><ul><li>Global statistics </li></ul></ul><ul><li>Current and future activities </li></ul>
  3. 3. Outline <ul><li>ECMWF main objectives using SMOS data, </li></ul><ul><li>Some aspects about the implementation of SMOS data in the Integrated Forecasting System (IFS), </li></ul><ul><ul><li>Main challenges of the implementation , </li></ul></ul><ul><ul><li>NRT product latency, </li></ul></ul><ul><ul><li>Pre-processing, </li></ul></ul><ul><li>SMOS data: preliminary results </li></ul><ul><ul><li>SMOS offline data monitoring webpage – observations monitoring, </li></ul></ul><ul><ul><li>Preliminary assessment of main first-guess departures sources, </li></ul></ul><ul><ul><li>Global statistics </li></ul></ul><ul><li>Current and future activities </li></ul>
  4. 4. Main objectives <ul><li>Global monitoring of Level-1C brightness temperatures at the satellite antenna reference frame, at several incidence angles. </li></ul><ul><ul><li>For Numerical Weather Prediction (NWP) applications, monitoring compares forecast (or analysis) and observed data. </li></ul></ul><ul><ul><li>Results available in NRT through the ECMWF satellite monitoring webpage. </li></ul></ul>Observed TB (OBS) ‏ Modelled TB (FG) ‏ Passive monitoring First Guess (FG) departures
  5. 5. Main objectives <ul><li>Assimilation of SMOS Level-1C brightness temperatures over land  investigate the meteorological impact of SMOS data assimilation. </li></ul><ul><ul><li>Extended Kalman Filter (EKF) soil moisture (w a ) analysis: </li></ul></ul>Soil moisture first-guess of j layer by ECMWF land surface scheme Background error matrix; a-priori knowledge of soil moisture variances Linearised version of the observation operator CMEM by small perturbations of the initial moisture state Soil layer defined in H-TESSEL Observation error matrix  inputs provided by monitoring statistics Multi-angular, multi-polarised SMOS TB observations w a,j = w b,j + (B -1 +H T R -1 H) -1 H T R -1 (TB 0 - H[w b,j ] ) ‏
  6. 6. Implementing SMOS data in the IFS. First version ODB: Observations Data Base used by the Integrated Forecasting System
  7. 7. <ul><ul><li>Volume of SMOS data, </li></ul></ul><ul><ul><ul><li>Much computing resources and time is needed to process and test SMOS data, </li></ul></ul></ul><ul><ul><ul><li>Which data to assimilate? </li></ul></ul></ul><ul><ul><li>Particular measuring principle (observation of the same area with different incidence angles at different time stamps) produces very large internal data bases which are difficult to handle, </li></ul></ul><ul><ul><ul><li>S tructure of SMOS ODB in the IFS needs to be revised to make it more efficient and use less memory resources  Is a ‘MUST’ for operational purposes, </li></ul></ul></ul><ul><ul><li>Observation operator (CMEM: Community Microwave Emission Model) implementation in the IFS, </li></ul></ul><ul><ul><ul><li>Compatibility with IFS is only guaranteed if CMEM code is adapted to a multi-thread environment </li></ul></ul></ul>Main obstacles (and challenges) in the implementation
  8. 8. L1C-NRT BUFR product Convert to L1C-NRT ECMWF BUFR product <ul><li>Pre-process data: </li></ul><ul><li>Consistency checks </li></ul><ul><li>Parallel data thinning </li></ul><ul><li>per angular bins </li></ul>ESAC MARS ECFS Store in ECMWF archives Computations in model space (gp_model) <ul><li>Get SMOS data in grid point </li></ul><ul><li>call smos_process </li></ul><ul><li>Forward model (CMEM) </li></ul><ul><li>physics interface routines </li></ul><ul><li>call callpar </li></ul><ul><li>Back to observation space </li></ul><ul><li>call smos_update </li></ul>BUFR files Mapping and load data to ODB tables ODB data passive monitoring of L1C TB over land & sea <ul><li>call smos_screen </li></ul><ul><li>CMEM interface </li></ul><ul><li>call mwave_screen </li></ul><ul><li>RTTOVS interface </li></ul>T atm ε Distribution per processor and grid point Implementing SMOS data in the IFS. New version
  9. 9. $Instrument_$SensingTime1_$SensingTime2_$Satellite_$orbit_$datatype_$GeneratingTime_$datalevel.bufr NRT latency – December 2009
  10. 10. NRT latency – Jan-Feb-Mar-Apr 2010 January February March April
  11. 11. NRT latency – May-June-July 2010 May June July
  12. 12. <ul><ul><li>SMS Supervisor Monitor Scheduler </li></ul></ul><ul><ul><li>Routinely checks, </li></ul></ul><ul><ul><li>Validity of data, </li></ul></ul><ul><ul><li>Data thinning, </li></ul></ul><ul><ul><li>Others checks can potentially be implemented at this level (noise filtering, RFI mitigation algorithms, etc.) </li></ul></ul>SMOS data pre-processing
  13. 13. Outline <ul><li>ECMWF main objectives using SMOS data, </li></ul><ul><li>Some aspects about the implementation of SMOS data in the Integrated Forecasting System (IFS), </li></ul><ul><ul><li>Main challenges of the implementation , </li></ul></ul><ul><ul><li>NRT product latency, </li></ul></ul><ul><ul><li>Pre-processing, </li></ul></ul><ul><li>SMOS data: preliminary results </li></ul><ul><ul><li>SMOS offline data monitoring webpage – observations monitoring, </li></ul></ul><ul><ul><li>Preliminary assessment of main first-guess departures sources, </li></ul></ul><ul><ul><li>Global statistics </li></ul></ul><ul><li>Current and future activities </li></ul>
  14. 14. Observations monitoring <ul><ul><ul><li>, SMOS offline monitoring webpage </li></ul></ul></ul><ul><ul><ul><li>Available since November-2009. Since January-2010 only NRT data is monitored and published, </li></ul></ul></ul><ul><ul><ul><li>Daily global maps of Level-1C NRT product, </li></ul></ul></ul><ul><ul><ul><li>Polarisations in the antenna reference frame at 0°, 10°, 20°, 30°, 40°, 50° and 60°, </li></ul></ul></ul><ul><ul><ul><li>http://www.ecmwf.int/research/ESA_projects/SMOS/monitoring/smos_monitor.html </li></ul></ul></ul>
  15. 15. TBxx TByy 20-Nov-09 20-Dec-09 16-Jan-10 Θ = 40° NRT NRT
  16. 16. Monitoring observations – Urulu (Australia) TBV
  17. 17. First - guess First-guess  CMEM initial config <ul><ul><li>Community Microwave Emission Model (CMEM), modular radiative transfer code used to compute first-guess: </li></ul></ul><ul><ul><li>- Drusch et al., 2009, JHM </li></ul></ul><ul><ul><li>- de Rosnay et al., 2009, JGR </li></ul></ul><ul><ul><li>- Mu ñ oz-Sabater et al., 2010, IJRS </li></ul></ul>dielectric Wang effect. temp Choudhury smooth surface Fresnel roughness Choudhury vegetation Kirdyashev atmosphere Pellarin
  18. 18. First-guess departures (obs - model) ‏ H-pol V-pol <ul><li>After implementation bugs removal, some departures are still too cold or too warm. </li></ul><ul><li>Case Study: </li></ul><ul><li>22 January 2010, </li></ul><ul><li>First 4D-Var 12h cycle, </li></ul><ul><li>Global scale, </li></ul><ul><li>All incidence angles included, </li></ul><ul><li>No mask applied on vegetation or snow </li></ul>
  19. 19. First-guess departures (obs – model) ‏ H-pol FG departures Orography Case Study: 22 January 2010 FG departures > 75K
  20. 20. First-guess departures (obs – model) ‏ H-pol FG departures incidence angle Case Study: 22 January 2010 FG departures < -100K
  21. 21. First-guess departures V-pol Case Study: 22 January 2010 -25 K < FG departures < 25 K
  22. 22. Preliminary assessment of main FG departures <ul><li>Departures too large  observations >> model : </li></ul><ul><ul><li>Location: mainly in Europe and Central-West Asia, </li></ul></ul><ul><ul><li>Contributing causes: </li></ul></ul><ul><ul><ul><li>mountainous areas (snow and slope effects), </li></ul></ul></ul><ul><ul><ul><li>areas contaminated by RFI. </li></ul></ul></ul><ul><li>Departures too negative  model >> observations. </li></ul><ul><ul><li>Location: South-Europe, North-Africa and some areas of China and Australia. </li></ul></ul><ul><ul><li>Contributing causes: </li></ul></ul><ul><ul><ul><li>coastlines, </li></ul></ul></ul><ul><ul><ul><li>dry areas at large incidence angles, </li></ul></ul></ul><ul><ul><ul><li>areas contaminated by RFI </li></ul></ul></ul><ul><li>These results need to be confirmed and further investigated with systematic statistics determined at global scale and at different incidence angles. </li></ul>
  23. 23. Global statistics over land Average of Observations, TBH polarisation, spatial resolution 0.25 ° 01-07 June
  24. 24. Global statistics over oceans Average of Observations, TBV polarisation, spatial resolution 1° 01-07 April Incidence angle 46-48°
  25. 25. Global statistics over oceans Number of Observations per grid square, TBV polarisation, spatial resolution 1° 01-07 April Incidence angle 46-48°
  26. 26. Histogram of departures over oceans <ul><li>03 June 2010, </li></ul><ul><li>All observations over open seas (first 4D-Var 12h cycle), </li></ul><ul><li>All incidence angles included, </li></ul><ul><li>Only flat component is accounted for. </li></ul>Observations – First_guess
  27. 27. <ul><li>Maps of Observations Standard Deviation (STD) </li></ul><ul><li>TB STD [K] </li></ul><ul><li>H-pol </li></ul><ul><li>45 ° -50 ° </li></ul><ul><li>01-07 March 2010 </li></ul><ul><li>03-10 May 2010 </li></ul>Global statistics: standard monitoring maps Areas affected by RFI  large STD of TB
  28. 28. Global statistics: Standard monitoring maps Map of Mean First Guess Departure over land (Obs – Model) 01-07 March
  29. 29. Global statistics : Standard monitoring maps Map of Mean First Guess Departure over land (Obs – Model) 01-07 April
  30. 30. Global statistics: Standard monitoring maps Map of Mean First Guess Departure over land (Obs – Model) 03-10 May
  31. 31. Global statistics: Standard monitoring maps <ul><ul><li>RFI impact on FG departures STD is large, </li></ul></ul><ul><ul><li>Excluding RFI contaminated areas, most of first-guess departures STD are below 9 K. Larger values are found in snow, boreal forests and dry areas. </li></ul></ul>Map of STD of First Guess Departure over land (Obs – Model) 01-07 March
  32. 32. <ul><li>STD of OBS </li></ul><ul><li>01-07 June </li></ul><ul><li>H-pol </li></ul><ul><li>V-pol </li></ul>Global statistics: Standard monitoring Hovmoeller Diagrams
  33. 33. Global statistics: Time Series <ul><ul><li>TB average (OBS) </li></ul></ul><ul><ul><li>Nbr observations </li></ul></ul><ul><ul><li>First-guess departures (FG_DEPAR) </li></ul></ul><ul><ul><li>45 ° -50 ° </li></ul></ul><ul><ul><li>std ( OBS, FG, FG_DEPAR) </li></ul></ul>Routine production of time series in different geographical areas Global (land), 01-06 Jun-2010 (30 ° - 35 ° ) ‏ TB ‏ H TBV ‏
  34. 34. <ul><ul><li>ECMWF main objectives using SMOS data are: monitoring and data assimilation </li></ul></ul><ul><ul><li>Implementation of SMOS data in the IFS is very complex and challenging , </li></ul></ul><ul><ul><li>The ‘SMOS chain’ depends critically on the NRT product latency , </li></ul></ul><ul><ul><li>An offline data monitoring webpage is available since Dec.09 and regularly updated  a follow-up update will substitute current data monitoring by temporal statistics. </li></ul></ul><ul><ul><li>Preliminary analyses on first-guess departures suggest that: </li></ul></ul><ul><ul><ul><li>RFI is the most important source of positive and negative bias, </li></ul></ul></ul><ul><ul><ul><li>As expected, snow, ice, mountains, boreal forests and dry areas produce also a significant disagreement with observations, greater in H than in V-pol, </li></ul></ul></ul><ul><ul><ul><li>Operational monitoring products of SMOS observation characteristics permit to identify any source of systematic differences with observations. </li></ul></ul></ul><ul><ul><li>Lot of information in the multi-angular and multi-polarised signal ! </li></ul></ul>Summary
  35. 35. <ul><ul><li>On going activities: </li></ul></ul><ul><ul><ul><li>Re-structuration of internal data base to efficiently handle SMOS data, </li></ul></ul></ul><ul><ul><ul><li>Implementation of the Ocean Emission Model from Argans, </li></ul></ul></ul><ul><ul><ul><li>Preparing SMOS chain for operations, </li></ul></ul></ul><ul><ul><ul><li>Testing routine production of global statistics in NRT, </li></ul></ul></ul><ul><ul><li>Future activities: </li></ul></ul><ul><ul><ul><li>Noise filtering, </li></ul></ul></ul><ul><ul><ul><li>Sensitivity of the SM analysis to different multi-angular and multi-polarised configurations of the observations, </li></ul></ul></ul><ul><ul><ul><li>Development of a bias correction scheme, </li></ul></ul></ul><ul><ul><ul><li>… many others… </li></ul></ul></ul><ul><ul><ul><li>Assimilation of SMOS data in the IFS. </li></ul></ul></ul>Current and future activities
  36. 36. Thanks !
  37. 37. Back-up slides
  38. 38. Faraday Effect over land H_pol TBH with Faraday – TBH without Faraday
  39. 39. Faraday Effect over land TBH with Faraday – TBH without Faraday V_pol
  40. 40. Observations monitoring – Sea West Coast Australia TBxx TByy
  41. 41. Monitoring observations – Urulu (Australia) TBxx TByy
  42. 42. Monitoring observations – Urulu (Australia) TBV TBH
  43. 43. Global statistics: Standard monitoring maps Map of STD of First Guess Departure over land (Obs – Model) 01-07 March
  44. 44. <ul><li>STD of first-guess departures </li></ul><ul><li>01-07 april </li></ul><ul><li>H-pol </li></ul><ul><li>V-pol </li></ul>Global statistics: Standard monitoring Hovmoeller Diagrams
  45. 45. Global statistics: Time Series Routine production of time series in different geographical areas Global area, 02-09 May-2010 (H-pol) ‏ <ul><ul><li>TB average </li></ul></ul><ul><ul><li>Nbr observations </li></ul></ul><ul><ul><li>First-guess departures </li></ul></ul><ul><ul><li>25 ° -30 ° </li></ul></ul><ul><ul><li>45 ° -50 ° </li></ul></ul>
  46. 46. Global statistics over oceans Standard deviation of Observations, TBV polarisation, spatial resolution 1° 01-07 April Incidence angle 46-48°

×