ra-2 & mwr ccvt – esa/esrin – 25 to 27 march, 2003 mwr calibration & validation status...
TRANSCRIPT
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
MWR Calibration & Validation status
MWR Cal/Val team: Ngan Tran, Estelle Obligis, Laurence Eymard,
Michel Dedieu,
Claire Marimont, Joël Dorandeu, Yannice Faugère
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Instrumental parameters monitoring
-2.8%
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Instrumental parameters monitoring
-3.0%
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Instrumental parameters monitoring
-3.8%
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Instrumental parameters monitoring
=> Impact direct sur Ta(ERS2 < 0.5K)
-2.2K
1.3K
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Instrumental parameters monitoring
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Impact of drift on the Brightness Temperatures
Drift of -1.06K/year on ENVISAT TBs
Drift of -1.90K/year on ENVISAT TBs
Monitoring over continental areas
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Impact of drift on the Brightness Temperatures
Monitoring over coldest ocean points
Selecting method of the coldest points over ocean: -1.2K /year
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
MWR 36 GHZ Drift Analysis
• The sky horn counts, hot load counts and gain have decreased by about 4% since launch
• Different methods provide a similar estimation for the drift of the brightness temperatures
• Differences between ERS and ENVISAT TBs present a drift, which is between +1 and +2 K/year
=> A drift of 1 to 2K/year corresponds to a drift between 0.7mm and 3 mm per year depending on the brightness temperature.
• Action on ALS (MWR manufacturer) to perform the investigations on the too high Te and counts and gain drifts cause and provide recommendations
• If no MWR stabilization, a Level 1 B algorithm upgrade to correct for the drift could be envisaged in a later stage
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Methodology for the MWR Calibration
• Retrieval algorithms formulated using the 2000/2001 analyses of the
ECMWF model / simulations by a radiative transfer model / regression
• Important change in the model in January 2002
Mean bias(K) 2000/2001 2002 difference
SIMU/ERS2 23.8 GHz +4.91 +2.75 +2.16
36.5 GHz +2.16 +0.30 +1.86
• Algorithms and calibration have to be consistent
• Not suitable to calibrate ENVISAT/MWR over 2002 analyses
> Calibration performed on ERS2 TBs adjusted on 2000/2001 database
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
+4.91 K +2.16 K
Comparison measurements/simulations ERS2 - 2000/2001 database
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
ERS2 vs ENV dh computed with
adjusted sigma0_Ku
High |dh| higher
Low |dh| lowerBias=4.5 mmStdev=10mm
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
ENV dh computed with adjusted sigma0_Ku vs
ECMWF dh
Bias=3.7 mm
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
ERS2 vs radiosounding wet
tropo
With Envisat we will get a better agreement with radiosounding measurements
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
MWR Algorithm Upgrades• Level 2 Neural Network Algorithm
Already implemented and validated in Reference Processor Expected operational date in IPF and CMA by mid-2003
Bias +1.46 cm Bias +4.5 mmNeural productsBetter behavior for dry atmosphere (no more necessary to add a particular correction)
Current products
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
MWR Algorithm Upgrades
• Level 2 MWR Side Lobe Algorithm
– Standard algorithm :
• Tsl constant : Tsl_24=8.21K and Tsl_36=0.413K
– Improved Side Lobe algorithm
• Tsl=eta_sun x Tsun + eta_sky x Tsky +eta_earth x Tearth + eta_sat x Tsat
• The satellite is supposed to be at the earth temperature
• Tsl=eta_sun x Tsun + eta_sky x Tsky +(eta_earth + eta_sat) x Tearth
• Tearth estimated from 1 year of ERS2/MWR measurements
• function of the position(1°x1° meshes)/season(4)/frequency(2)
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Mean Side Lobe Contribution for 23.8 GHz channel - Spring
Actual constant value
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Difference between TB_23.8 computed with improved and operational algo for cycle 10
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Difference between TB_36.5 computed with improved and operational algo for cycle 10
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Difference between Dh computed with improved and operational algo for cycle 10
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Validation of the new Side Lobe algorithm
Validation method– Comparison between BTs Envisat and ERS-2 over strongly
contaminated zones– With the new algorithm, the difference in BT between ERS and
Envisat shall be closer than the one observed in open ocean
But– Can we consider the ERS-2 BT as a reference ?– Are all the Envisat efficiencies values valid ?– Can the Envisat Platform be considered as a perfect reflector ?
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Conclusions / Recommendations
• The Envisat MWR has been calibrated to get consistency with the 2000/2001 database used for the computation of the retrieval algorithms
• Mean bias around 5 mm between ERS2 and ENVISAT dh => we will get a better agreement with radiosounding measurements
• Neural algorithm ready for operational application– Linear correction factors shall be added on BTs and Sigma0– Sigma0_Ku 1 dB higher (in MWR NN level 2 processing) to provide
reliable radiometer products
• The impact of the new side lobe algorithm has been estimated (up to 1 cm near the costs). The validation is on the way
• There is a drift of the ENVISAT 36.5 GHz gain. Investigation is on-going. Not taken into account because the calibration has been performed at the end of September and is valuable for October data
RA-2 & MWR CCVT – ESA/ESRIN – 25 to 27 March, 2003
Perspectives • Validation of the Side Lobe algorithm near the costs using ERS2 TBs (-
> mid 03)
• Comparison with other radiometers (Jason, AMSU-A - 1 common channel)
• Comparison with radiosounding measurements (when enough coincidences : mid 2003 ?)
• Check/change the calibration depending on the drift explanation (-> April/May 03)
• After adjustment of the ERS2 TBs as well as correction of the 23.8 GHz drift, use of the same algorithms than for Envisat => better agreement and continuity between the 2 missions