CONAE Microwave Radiometer (MWR) Counts to Tb Algorithm (Version 6.0) and On-orbit Validation

Transcription

1 CONAE Microwave Radiometer (MWR) Counts to Tb Algorithm (Version 6.) and On-orbit Validation Zoubair Ghazi W. Linwood Jones María Marta Jacob CFRSL Central Florida Remote Sensing Lab CONAE Comisión Nacional de Actividades Espaciales MWR Calibration Workshop Buenos Aires, AR Nov.,

2 SW Matrix MWR Channel Block Diagram T ant T in T N T ref Counts On/off Three Dicke radiometer states: C C C a N ref ( T ( T ( T ant ant ref T TN T recv ) Grecv V Trecv ) G ) G V recv recv off set recv V off set off set () () () Subtracting () from () yields the radiometer gain, which varies in time G recv C N C T N a

3 During MWR on-orbit Cal/Val testing, all three radiometer channels exhibited similar brightness temperature (Tb) anomalies: The system gain was found to depend upon input scene Tb Gain, calc using the excess noise ratio = (C N C a )/T N, changed abruptly at Ocean/Land boundaries where high Tb contrast exist This anomaly is believed to be the result of a slightly non-linear radiometer transfer function (T in to counts)

4 Rad_counts V.S V6. Note: gain land/water boundaries Dynamic range = 6 counts

5 New version V6. developed by CFRSL and CONAE incorporates : Smear correction counts (same as V.S) Counts linearization Removes radiometer transfer function non-linearity Reanalysis of pre-launch TV calibration data To improve front-end loss coefficients Antenna Pattern Correction (APC) To convert T ant to Tb The results are presented in the following charts

6 Rad_counts, counts Quadratic regression T in- = T ant + T N T in- = T ant T in- = T o Ref Load Full-dynamic range Radiometer Input to the antenna port of Dicke switch, T in (Kelvin) 6

7 Averaging nd order regression coeff s from 7 DSC orbits, the instantaneous counts linearization equation is: For 7 V _ linear x (-7.677e-)* T in C x C For 7 H _ linear x (-6.96e - )* T in C x C For H _ linear x (-.78e - )* T in C x C Where x = ant, N, and ref Tin is the input Tb to the Dicke switch, which is estimated using smear-corrected (non-linear) counts 7

8 Rad_counts (gain normalized) Rad_counts (gain normalized) V.S (non-linear counts) V6. (linear counts) y = 6.9x - (T in ) +.8(T in )+6 y =.x -6 (T in ) +.8(T in )+ 6 A reduction of > Radiometer Input to the antenna port of Dicke switch, T in (Kelvin) 8

9 APC and residual bias correction were applied by inter-satellite XCAL MWR = target & WindSat = reference MWR and WindSat have different incident angles, therefore, Tb s were adjusted using theoretical radiative transfer model values for both satellites (MWR sim and WS sim ) WS adj = WS obs + (MWR sim WS sim ) Double Difference Technique DD = MWR obs WS adj 9

10 MWR Observed Tb, K Slope =.99 Offset =.7 Land Linear Regression Ocean WindSat Adjusted Tb, K

11 Double Difference Radiometric Biases MWR/WindSat (Five Days Average) Sept, Feb 8,

12 Version.S B B B B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8 Version 6. B B B B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8

13 Version.S Version 6. North Pole - North Pole Sep - feb Sep - feb 8 Sep - Feb 8 Sep - Feb 8 - -

14 Version.S B B B6 B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8 Version / /7 / Sep - feb 8-9/ /7 / Sep - feb 8 B B B6 B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8

15 Version.S Version 6. North Pole - North Pole Sep - feb Sep - Feb 8 Sep - Feb 8 Sep - feb 8 - -

16 Version.S B B B B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8 Version 6. B B B B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8

17 Version.S Version 6. North Pole - North Pole Sep - feb Sep - feb 8 Sep - Feb 8 Sep - Feb

18 Version.S B B B B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8 Version 6. B B B B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8 8

19 Version.S Version 6. North Pole - North Pole Sep - feb Sep - Feb 8 Sep - Feb 8 Sep - feb

20 MWR Counts to Tb algorithm V6. has been developed and a beta test version has been distributed to the AQ Cal/Val team via PODAAC MWR transfer function non-linearity in V.S has been characterized and corrected in V6. A count-linearization procedure has been shown to make the radiometer transfer function linear and removed the effect of gain dependence on the input scene brightness On-orbit inter-satellite radiometric calibration (XCAL) between MWR and WindSat have produced the antenna pattern correction (APC) and removed small Tb biases

21

22 MWR V6. Counts to Tb algorithm be adopted by CONAE for L-B processing All MWR data should be reprocessed using V6. L- Geophysical retrieval algorithms should be retuned using V6. Tb s MWR Calib and Application team reviews be held in Spring to formally release MWR data products

23 Version.S B B B6 B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8 Version 6. B B B6 B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8

24 Version.S Version 6. North Pole - North Pole Sep - feb Sep - Feb 8 Sep - Feb 8 Sep - feb 8 - -

25 Version.S B B B6 B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8 Version 6. B B B6 B / /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8-9/ /7 / Sep - feb 8

26 Version.S Version 6. North Pole - North Pole Sep - feb Sep - Feb 8 Sep - Feb 8 Sep - feb

27 Version.S Version 6. 7

28 Under typical on-orbit condition, the radiometer system gain will vary cyclically (once/orbit) due to the receiver physical temperature changes However, to determine the radiometer transfer function, the gain must be constant Therefore, the time variable gain was removed using the Dicke state- (viewing the reference load) G i = C i (To+Trec) i where G i is the instantaneous channel gain C o is the reference load counts T o is the receive physical temperature T rec is the receiver noise temp 8

29 Time variable gain was removed and all counts were normalized using the following equation: C x_ norm = C xi * mean{co/ (To+Trec)} [Co i / (To+Trec) i ] x = a, N, o & i = instantaneous value 9

30 Normalized Relative Gain Normalized_relative_ gain = G i mean(g i ) Radiometer physical Temp, T o (Kelvin)

31 Noise Diode Deflection, counts Noise Diode Deflection, counts Scene Brightness Tb, K Before Count (gain) Normalization After Count (gain) Normalization Effects of variable gain Linear dependence on scene Tb Radiometer Input Tin, K Radiometer Input Tin, K

32 Seven Deep Space Calibration (DSC) orbits that included, space, ocean, and land observations were used to cover wide range of scene Tb s After counts (gain) normalization, the radiometer transfer function was established Rad_counts = f(t in ) Quadratic regression for 7V channel yielded the following Rad_counts = 7.79x - (T in ) (T in ) + 7.9

33 Gain, counts/k Gain, counts/k Scene Brightness Tb, K V.S V6. Space Linear Regression Ocean Ocean Land Land Brightness Temperature, K Brightness Temperature, K

34 Gain, counts/k Gain, counts/k Scene Brightness Tb, K Note: gain jumps V.S V6. Gain variation due to changing recvr phy temp Time Time

35

36 Characterization of injected noise diode temperature (T N ) over physical temperature Retrieve switch matrix loss coefficients Empirical method (regression model) was applied Assumption: All transmission and reflection coeff s are constant and are NOT expected to change during MWR's mission life time 6

37 T ap =[T in -(b *T o +b *T +b *T +b *T +b 6 *T )]/b T in calc from rad_counts & linear transfer function T ap & T ap_regress 7

38 T ap =[T in -(b *T o +b *T +b *T +b *T +b 6 *T )]/b where T ap is the scene brightness temp at horn aperture T in is the input brightness temperature to antenna port of Dicke switch T o, T, T, T,& T are MWR physical temps b, b, b, b, b and b 6 are coefficients derived using the regression model 8