Aquarius Cal/Val: Salinity Retrieval Algorithm V1.3 Thomas Meissner Frank Wentz
L2 Algorithm: Overview Aquarius Radiometer Counts Earth + Calibration View Radiometer Calibration Algorithm RFI flagging Total Antenna Temperature Sea-Surface Brightness Temperature Remove Surface Roughness Effects Remove Space Contributions: Galaxy, Sun, Moon, CS Earth Antenna Temperature Remove the Antenna Pattern Effect Earth Brightness Temperature (TOI) Correct for Faraday Rotation Specular Brightness Temperature Find Salinity for which emissivity of Meissner-Wentz dielectric model matches specular TB (v-pol only for now) Salinity Top of the Atmosphere Brightness Temperature (TOA) Remove Atmospheric Contribution
V1.2.3 Updates from V1.1 1. Correction for radiometer drift using CND/ND deflection ratios (see J. Piepmeier) 2. Surface roughness correction: NCEP wind speed + scatterometer σ 0 Change in scatterometer requires re-deriving roughness correction and reprocessing of L2 processing. V1.3 to be done 1. Implementation of overflow flag 2. Flag consolidation
Residual TB error after drift correction (RSS) 7 day backward linear interpolation ND update
Surface Roughness Correction (V1.2.3) Tb surf0 flat surface scatterometer σ 0 radiometer TB surf measurement NCEP Wind Direction ΔTB surf ancillary data pre computed remove SCAT wind direction signal add RAD wind direction signal action output scatterometer σ 0 (isotropic) NCEP Wind Speed ΔTB surf (isotropic) decision 2-dim lookup table roughness correction flag set [W, σ 0 ] bin underpopulated No Yes use only NCEP wind speeds use σ 0 and NCEP wind speeds
Surface Roughness: Wind Speed + Scatterometer σ 0 2-dimensional lookup table: [wind speed, σ 0 ] ΔT B
Surface Roughness Correction (V1.3 prop.) Tb surf0 flat surface scatterometer σ 0 radiometer TB surf measurement NCEP Wind Direction ΔTB surf ancillary data pre computed remove SCAT wind direction signal add RAD wind direction signal action output scatterometer σ 0 (isotropic) NCEP Wind Speed ΔTB surf (isotropic) decision 2-dim lookup table roughness correction flag set [W, σ 0 ] bin underpopulated or SCAT RFI flag bit 29/31 set No Yes use only NCEP wind speeds use σ 0 and NCEP wind speeds
V1.3 Radiometer Data + Flags (1) Lists all TB steps from TOA to specular surface Handling of land correction Variables w/i land correction are default. Variables w/o land correction are labeled _nolc. Flag iqc_rough 0: regular case = used W NCEP + σ 0VV for roughness correction. 1: [W NCEP, σ 0VV ] bin underpopulated = used only W NCEP for roughness correction. Statistics: iqc_rough = 1 in: 0.5% (horn 1), 0.7% (horn 2), 1.3% (horn 3) of all cases with no SCAT RFI (bit 29/31 in scat flag not set). Variables dtb_v + dtb_h are obsolete. TB errors (specular surface) T B,err = T B,surf,0,meas T B,surf,0,RTM v,h 5
V1.3 Radiometer Data + Flags (2) TB at TOA Variable TB at surface Default (with land correction) rad_toa_v rad_toa_h rad_tbv rad_tbh No land correction (nolc) rad_toa_v_nolc rad_toa_h_nolc rad_tbv_nolc rad_tbh_nolc TB after removing roughness correction (specular) rad_tbv_rc rad_tbh_rc rad_tbv_rc_nolc rad_tbh_rc_nolc TB error rad_tb_err rad_tb_err_nolc Sea surface salinity SSS SSS_nolc
Negative TA/TF (1) RSS V1.1, V1.2 + current V1.3 L2 processing produces negative TA in 2 instances: 1. Cold space maneuvers Minimum TA (v-pol) close to 0K, can get slightly negative ( > -1.0K) Minimum TA (h-pol) about +6K Calibration problem with cold space non-linearity correction off? not a processing bug needs to be looked at but left alone for V1.3 V 1.3 does have flag for cold space look 2. During very strong RFI Limited number of bits reserved for counts 13 bits for SA 1 2 12 bits for other SA Strong RFI causes overflow of bits SA bits get wrapped around Very small numbers for SA Causes negative TA after count to TA
Proposed Fix for Case 2 in V 1.3 L1A files contain an overflow flag radiom_nrt_tlm. Radiometer Non-Real Time Telemetry. See Table 5 of L1A manual. Indicates that overflow occurs at which cycle (frame + subframe) for radiometer (1-3). It does NOT tell you which one of the SA (1 6), which subcycle (1-12) and which polarization is overflowing. Simple fix: Check flag. If set, set all SA for all polarizations this whole cycle (frame + subframe) and radiometer to missing/invalid at beginning of L2 processing. Internally within L2 processing If SA is missing/invalid for cycle, no L2 processing is done. Carry overflow flag from L1A into L2. If overflow flag is applied correctly no negative TA are produced beside for Case 1 (cold space). If overflow flag is applied correctly no negative TF are produced beside for Case 1 (cold space). RFI algorithm does NOT produce negative TF.
TA/TF Distributions cold space ocean land/ ice RFI
Invalid / NaN SSS SSS retrieval Newton iteration looking for 0 of TB surf,0 [measured RTM] (SSS). Breaks down in very few cases. Cold space maneuvers shortly before boresight goes off the Earth. RTM breaks down. TB surf,0 [measured RTM] (SSS) = 0 has no realistic solution. Can produce unrealistic or NaN SSS values. Fix Safeguard Newton iteration. Check on χ 2 = TB surf,0 [measured exp] 2 for convergence. If failed set SSS to missing/invalid (not yet implemented). Set SSS to missing/invalid if gland or gice > 0.5
Discussion Item: Flag Consolidation ADPS radiometer_flags Status unclear iqc_rough overflow flag L2 Q/C flag missing/invalid radiometer observation missing/invalid time or geolocation boresight off Earth currently produced in L2 algorithm but not in ADPS files