DMSP F19 SSMIS Cal/Val Progress and NAVGEM Assimilation Results Tanya Maurer 1 , Steve Swadley 1 , Gene Poe 2 Ben Ruston 1 , Anna Booton 3 , Al Uliana 2 , Philip Shen 2 1 NRL, Monterey, CA 2 SAIC, Monterey, CA 3 UK Met Office, Exeter, UK 13 th JCSDA Technical Review & Science Workshop College Park, Maryland May 2015
29
Embed
DMSP F19 SSMIS Cal/Val Progress and NAVGEM Assimilation Results Tanya Maurer 1, Steve Swadley 1, Gene Poe 2 Ben Ruston 1, Anna Booton 3, Al Uliana 2, Philip.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
DMSP F19 SSMIS Cal/Val Progress and NAVGEM Assimilation Results
Tanya Maurer1, Steve Swadley1, Gene Poe2
Ben Ruston1, Anna Booton3, Al Uliana2, Philip Shen2
1NRL, Monterey, CA2SAIC, Monterey, CA
3UK Met Office, Exeter, UK
13th JCSDA Technical Review & Science Workshop College Park, Maryland
May 2015
Outline
• Background• F19 SSMIS Cal/Val status• F19 Cold Bias anomaly detection and mitigation strategy• F19 SSMIS assimilation in NAVGEM
• UPP updates and additions• SSMIS Upper Air Sounding (UAS) capabilities• SSMIS UPP radiance assimilation impacts
• Summary and Look Ahead
• Defense Meteorological Satellite Program (DMSP) heritage: 7 SSM/I, 4 SSMIS• Comprehensive Cal/Val program verify end-to-end instrument performance
DMSP & the SSMIS Cal/Val Program
Current DMSP MW Imager/Sounder Constellation Status
LTAN: Local Time of the Ascending NodeDMSP F19 Launched April 3, 2014
SSMIS deployed April 7, 2014
• Anomaly & systematic bias detection• SDR & EDR validation• Algorithm improvements• Rapid transition of data & products to users
Geolocation • Earth location errors ~20-30 km observed• Correction factors derived in terms of along-scan & along-track
partial derivatives• Errors are reduced to 3-5 km (spec=7km)
Environmental Data Records (EDR) Validation• OSWS and TPW EDRs validated against ocean buoys and Raob
stations• New APC coefficients derived • Retrieval skill within specification
Spec = 2m/s
F19 SSMIS Cal/Val Progress
Initial NRL RTM Departures Analysis uncovered a Large Cold Bias in the G-band channels. The OB-BK patterns for the 183 GHz channels were similar to F18, but colder by ~8-10K. Only the high frequency G-band channels were affected. Initial focus was on Reflector Roughness, then moved to possibility of FOD in optics via the damaged feedhorn cover.
• Histograms Plots of the Observed and Background computed TBs for F19 SSMIS Revs 64-66, 2014040806. Biased histograms are shifted and it was shown that altering the SPOFs would mitigate the cold bias.
• New SPOFs, αn, derived using ECMWF Analyses and RTTOV-10 computed TBK
• Computed Ocean Only Scene SPOFs are very stable over time.
• Method validated through vicarious calibration with F18
• Shows good performance across all surface types for 6 month period
• SPOF modification justified using APC physical model
• Vicarious Calibration SPOFs match those derived by the NWP RTTOV methodology (within ~0.5-1%)
Long Term Averages over Homogenous Surfaces
• Forecast Model: Navy Global Environmental Model (NAVGEM)– T359L50– 0.04 hPa model top (~70 km)– ~ 37 km horizontal resolution – Semi-Lagrangian/Semi-Implicit dynamical core, forecast model, explicit
clouds
• DA system: NRL Atmospheric Variational Data Assimilation System – Accelerated Representer (NAVDAS-AR)– T359 outer loop, T119 (~ 111 km) inner loop resolution– Radiance bias correction using variational bias correction approach
• Joint development by NRL and UK Met Office• Produces calibrated SSMIS TDR files suitable for radiance assimilation• SSMIS UPP V2 Operational at FNMOC (F16 - 07/2008, F17 - 04/2009, F18 - 04/2010,
F19 – 10/2014)• FNMOC distributes UPP data to NESDIS for use by the global NWP Community• Produces ASCII and BUFR TDR output files at full and/or filtered resolution• UPP V2 run at FNMOC includes:
• Reflector Emission Corrections (F16 and F17)• NRL-derived SPOF factors for cold-bias mitigation (F19)• Operational NGES Fourier Filtered Gain Files to Correct Gain Anomalies • Spatial Averaging to reduce NEΔT to 0.15 - 0.25 K level (NRL only)• Sensor-specific scan Non-uniformity and Geo-location corrections (corrects for
FOV intrusions and spacecraft misalignment)
SSMIS Radiance Assimilation by NAVDAS-AR
SSMIS Unified Pre-Processor (UPP)
• F19 UPP data Operational at FNMOC on 14 October 2014• Utilizes NRL SPOF correction methodology for Cold Bias mitigation• NRL Assimilating F19 channels 2-7, 9-11, 22-24 averaged at σ=37km
• Implemented Orbit Angle, φ, calculation for Bias Correction Predictor• Testing new orbital bias correction scheme using a Fourier Series in ϕ • Developed with Anna Booton (Met Office) during visit (Nov-Dec ‘14)• Using reference relative to the position of the sun results in a more stable predictor
Ecliptic plane
Ascending node φ – angle along the
orbital plane
β – bias coefficientP – bias predictorϕ – orbital angle
SSMIS UPP Updates and Additions
Assimilation of SSMIS by NAVDAS-AR
SSMIS UAS UPP Ready for Operations
High-peaking SSMIS Upper Atmosphere Sounding (UAS) channels: 19, 20, 21• Affected by Zeeman splitting of the oxygen absorption lines at alt > 50 km, due to
interaction of Earth’s magnetic field and SSMIS bore-sight angle• Require separate UAS UPP (operates on channels 19-24) • Retrievals provide key information about upper atmosphere
• Includes parameters: mean Bmag, B-dot-k, θB , and std_dev(B-dot-k) within averaging domain (100 closest UAS scenes)
• Gaussian spatial averaging with σ=75km• Orbit Angle, φ, calculation for Bias Correction
Predictor• Special Met Office file also available at full
resolution
UAS UPP includes:• Calculation of the SSMIS Propagation Vector, k, as the vector difference between the
spacecraft and the UAS scene position vectors• Extraction of the Geomagnetic field vector, B, components at 60 km
* Assimilation of SSMIS UAS channels into NAVGEM is currently limited to CH 21 due to model top
Assimilation of SSMIS by NAVDAS-AR
SSMIS Radiance Data QC/Prep• UPP averaging/thinning:
• Along-scan Gaussian averaging• Every 3rd scan, every other LAS scene (30 scenes per scan)• Every 6th scan, every other UAS scene (15 scenes per scan)
• Pre-assimilation thinning: 1.25°• Quality Control
• Ocean-only scenes for humidity sounding channels (F18, F19)• For temperature sounding channels 02, 03, 04, scenes are
rejected if:• Surface is not 100% homogenous• 20 FOVs in the spatial averaging domain have rain flag set
to “true”• Jacobian w.r.t. surface temperature exceeds threshold
Standard deviations of the bias-corrected innovations (bars) shown on top of weighting functions for each channel.