A Comprehensive Variational Approach to Remote Sensing in All-Weather, All-Surface Conditions -MiRS Algorithm- S.-A. Boukabara, K. Garrett, F. Iturbide-Sanchez, C. Grassotti, W. Chen, L. Moy, F. Weng and R. Ferraro NOAA/NESDIS Camp Springs, Maryland, USA NASA Sounder Science Team Meeting, Greenbelt, MD November 10, 2011
Contents General Overview and Mathematical Basis 1 2 Performance Assessment 3 Summary & Conclusion 2
Introduction / Context v Physical algorithm for microwave sensors (MiRS) v Cost to extend to new sensors greatly reduced v MiRS applies to imagers, sounders, combination v MiRS uses the CRTM as forward operator (leverage) v Applicable on all surfaces and in all-weather conditions v Operational for N18,19,Metop-A and F16/F18 SSMI/S. v On-going / Future: § Extension operations to Metop-B, NPP/ATMS and Megha-Tropiques (MADRAS and SAPHIR) § Get ready for the JPSS and GPM sensors. § Extend to FY-3 MWTS, MWHS and imager § Extend applications of MiRS (hydrometeors profiling) § Extend MiRS to Infrared Remote Sensing (CRTM is already valid) 3
All-Weather and All-Surfaces sensor Sounding Retrieval: Major Parameters for RT: • Temper emperatur ure e • Sensing Frequency • Mois oistur ure e • Absorption and scattering properties of material To account for cloud, rain, ice, we add the following in the state vector: • Geometry of material/wavelength interaction • Cloud loud (non-pr non-precipit ecipitating) ing) • Vertical Distribution • Temperature of absorbing layers • Liquid Liquid Precipit ecipitation ion • Pressure at which wavelength/absorber interaction occurs • Frozen en pr precipit ecipitation ion • Amount of absorbent(s) • Shape, diameter, phase, mixture of scatterers. To handle surface-sensitive channels, we add the following in the state vector: • Skin temperature Absorption • Surface emissivity (proxy parameter for all surface parameters) Scattering Effect Scattering Effect v Instead of guessing and then removing the impact of cloud and rain and ice on TBs (very hard), MiRS approach is to account for cloud, rain and ice within its state vector. v It is highly non-linear way of using cloud/rain/ice-impacted radiances. Surface 4
MiRS General Overview Vertical Integration and Post-Processing Radiances TPW Temp. Profile RWP IWP Rapid Humidity Profile Advanced Vertical Vertical 1 st Guess CLW Algorithms Retrieval Integration & Integration Liq. Amount Prof Post-processing (Regression) (1DVAR) Outputs 1DVAR Ice. Amount Prof -Sea Ice Concentration -Snow Water Equivalent selection Rain Amount Prof -Snow Pack Properties Post -Land Moisture/Wetness -Rain Rate Processing Emissivity Spectrum MIRS -Snow Fall Rate (Algorithms) -Wind Speed/Vector Products -Cloud Top Skin Temperature -Cloud Thickness Core Products -Cloud phase 5
1D-Variational Retrieval/Assimilation Measured Radiances MiRS Algorithm Yes Solution Comparison: Fit Simulated Radiances Within Noise Level ? Reached No Jacobians Update Initial State Vector State Vector Measurement & RTM Uncertainty Forward Operator New State Vector Matrix E (CRTM) Geophysical Mean Background Geophysical Covariance Forecast Field (1D-Assimilation Mode) Climatology (Retrieval Mode) Matrix B 6
Mathematical Basis: Cost Function Minimization v Cost Function to Minimize: 1 1 T ⎡ ⎤ ⎡ ⎤ T Jacobians & Radiance Simulation ( ) ( ) ⎥ J(X) X X B 1 X X Y m Y(X) E 1 Y m Y(X) ( ) ( ) − − = − × × − + − × × − 0 0 ⎢ 2 ⎥ ⎢ 2 from Forward Operator: CRTM ⎣ ⎦ ⎣ ⎦ J(X) ' ∂ J (X) 0 v To find the optimal solution, solve for: = = X ∂ ⎡ − y(x) y(x ) K x x v Assuming Linearity ⎤ = + 0 0 ⎢ ⎥ ⎢ ⎥ ⎣ ⎦ v This leads to iterative solution: 1 − ⎧ ⎫ 1 T 1 T 1 m ΔX B K E K K E Y Y(X ) K ΔX ⎪ ⎛ − − ⎞ − ⎪ ⎡ ⎤ ⎪ ⎪ ⎛ ⎞ = + − + n n n n n n ⎜ ⎟ n 1 ⎜ ⎟ ⎨ ⎬ ⎢ ⎥ + ⎜ ⎟ ⎝ ⎠ ⎣ ⎦ ⎪ ⎪ ⎝ ⎠ ⎪ ⎪ ⎩ ⎭ 1 − ⎧ ⎫ T T m ΔX BK K BK E Y Y(X ) K ΔX ⎛ ⎞ ⎪ ⎪ ⎡ ⎤ ⎛ ⎞ = + − + n n n n n n ⎜ ⎟ n 1 ⎨ ⎬ ⎜ ⎟ ⎜ ⎟ ⎢ ⎥ + ⎝ ⎠ ⎣ ⎦ ⎝ ⎠ ⎪ ⎪ ⎩ ⎭ More efficient Preferred when nChan << nParams (MW) (1 inversion) 7
Parameters are Retrieved Simultaneously If X is the set of parameters that impact the radiances Ym, and F the Fwd Operator Necessary Condition (but not sufficient) F(X) Fits Y m within Noise levels If F(X) Does not Fit Y m within Noise X is not the solution X is a solution X is the solution All parameters are retrieved simultaneously to fit all radiances together Suggests it is not recommended to use independent algorithms for different parameters, since they don ’ t guarantee the fit to the radiances 8
Solution-Reaching: Convergence v Convergence is reached everywhere: all surfaces, all weather conditions including precipitating, icy conditions v A radiometric solution (whole state vector) is found even when precip/ice present. With CRTM physical constraints. T 2 m 1 m Y Y X E Y Y X − ( ) ( ) ⎟ ⎛ ⎞ ⎛ ⎞ ϕ = − × × − ⎜ ⎟ ⎜ ⎝ ⎠ ⎝ ⎠ Current version Previous version (non convergence when precip/ice present) 9
Contents General Overview and Mathematical Basis 1 2 Performance Assessment 3 Summary & Conclusion 10
MiRS List of Products Official Products Products being investigated 1. Temperature profile 1. Cloud Profile Vertical Integration and Post-Processing 2. Moisture profile 2. Rain Profile 3. TPW (global coverage) 3. Atmospheric Ice Profile TPW Temp. Profile The following section about performance assessment 4. Land Surface Temperature 4. Snow Temperature (skin) RWP is a snapshot (focused on sounding mainly). IWP Humidity Profile 5. Emissivity Spectrum 5. Sea Surface Temperature Vertical CLW Integration 6. Surface Type (sea, land, snow, 6. Effective Snow grain size Liq. Amount Prof Outputs 1DVAR sea-ice) 7. Multi-Year (MY) Type SIC 7. Snow Water Equivalent Ice. Amount Prof 8. First-Year (FY) Type SIC -Sea Ice Concentration (SWE) -Snow Water Equivalent 9. Wind Speed 8. Snow Cover Extent (SCE) Rain Amount Prof -Snow Pack Properties Post 10. Soil Wetness Index -Land Moisture/Wetness 9. Sea Ice Concentration (SIC) -Rain Rate Processing Emissivity Spectrum -Snow Fall Rate (Algorithms) 10. Cloud Liquid Water (CLW) -Wind Speed/Vector -Cloud Top Skin Temperature 11. Ice Water Path (IWP) -Cloud Thickness Core Products -Cloud phase 12. Rain Water Path (RWP) 11
Temperature Profile Assessment (against ECMWF) Angle dependence taken care of very well, without any limb correction MIRS ECMWF Note: Retrieval is MIRS – ECMWF Diff done over all surface backgrounds but also in all weather conditions (clear, MIRS – ECMWF Diff cloudy, rainy, ice) N18 12
Moisture Profile (against ECMWF) Validation of WV done by comparing to: - GDAS - ECMWF - RAOB MIRS ECMWF Assessment includes: land Bias Sea - Angle dependence - Statistics profiles - Difference maps Stdev When assessing, keep in mind all ground truths (wrt GDAS, ECMWF, RAOB) N18 13
TPW Global Coverage MiRS GDAS MiRS TPW Retrieval (zoom over CONUS) Very similar features to GDAS Smooth transition over coasts 14
ATMS Expected Performances a) b) c) d) Theoretical performances for temperature sounding over ocean (a) and land (b) and water vapor sounding over ocean (c) and land (d). Simulations are performed in clear-sky for NPP with no noise added (black), N18 with noise (blue), NPP with N18-like noise (red dashed) and NPP expected noise (red). 15
NPP/ATMS REAL DATA Initial Assessment of Noise levels Noise levels for NPP/ATMS seem all to be within spec, and even lower (for some channels, significantly) than spec. To be monitored further with time. 16
NPP/ATMS Real Data (Initial Radiometric Assessment) NPP/ATMS data started flowing Nov 8 th 2011 Raw NPP/ATMS TB @ NPP/ATMS Simulated TB @ 57GHz (no correction) 57GHz(using GFS & CRTM) Hot off the press results. Non-corrected TBs fed to MiRS. 17
NPP/ATMS Real Data (Initial EDRs Assessment) Emissivity variation: - Ocean/Land Contrast - Ocean/Sea-ice contrast - Ocean angle variation Hot off the press results. Very encouraging results Non-corrected TBs fed to MiRS (no bias correction 18
Contents General Overview and Mathematical Basis 1 2 Performance Assessment 3 Summary & Conclusion 19
Recommend
More recommend
