The AIRS Forward Model: Validation & In-Orbit RTA L. Larrabee - - PowerPoint PPT Presentation

▶
the airs forward model validation in orbit rta
SMART_READER_LITE
LIVE PREVIEW

The AIRS Forward Model: Validation & In-Orbit RTA L. Larrabee - - PowerPoint PPT Presentation

Aug 16, 2022 22 likes •493 views

September 2002 AIRS Science Team Meeting The AIRS Forward Model: Validation & In-Orbit RTA L. Larrabee Strow Scott E. Hannon Howard E. Motteler Sergio De Souza-Machado Rodrigo Ferreira de Souza (INPE) Lisa LeBlanc L A N Y D R A B

slide-1
SLIDE 1

September 2002 AIRS Science Team Meeting

The AIRS Forward Model: Validation & In-Orbit RTA

  • L. Larrabee Strow

Scott E. Hannon Howard E. Motteler Sergio De Souza-Machado Rodrigo Ferreira de Souza (INPE) Lisa LeBlanc

U M B C

U N I V E R S I T Y O F M A R Y L A N D B A L T I M O R E C O U N T Y 1 9 6 6

Department of Physics University of Maryland Baltimore County (UMBC) Baltimore, MD 21250

  • L. Strow, UMBC

1

slide-2
SLIDE 2

September 2002 AIRS Science Team Meeting

Overview

  • Validation data: ECMWF, PREPQC, ARM-CART
  • AIRS observations: July 20, Aug 31, Sept. 1,2,5,6,9,10,11
  • Aug/Sept. data validation done with the new AIRS-RTA at the

post-defrost frequencies

  • Changes to AIRS-RTA from pre-launch version include: (1) correct

frequencies, (2) fringes, (3) new water continuum, and (4) improved CO2 at 4.3 microns

  • Plan to deliver new AIRS-RTA after final testing next week. Includes 2 new

predictors for the water continuum, needed because we added temperature dependence to the foreign continuum.

  • This new AIRS-RTA contains no direct corrections from AIRS validation.
  • Will cover clear FOV selection, bias observations, differences between

AIRS-RTA and those based on GENLN2, and some very initial observations

  • f non-LTE and cirrus.
  • L. Strow, UMBC

2

slide-3
SLIDE 3

September 2002 AIRS Science Team Meeting

Clear Selection: 1st find uniform golfballs

Our FOV selection was based on three semi-independent tests: (1) the uniformity of the FOV radiances compared to all adjacent FOVs, (2) window channels at different wavelengths must have the same B(T), after correction for water, and (3) remove outliers by forcing agreement between observed and computed SST to ±4K. Uniform B(T) test:

  • test channels = [ 759, 903, 2328, 2333 ] = [900, 961, 2611, 2616 cm−1 ]
  • calculate BTmean of test channels for every FOV
  • calculate |∆BT| for all 8 adjacent FOVs
  • FOV is uniform only if max |∆BT| < ∆BTmax
  • ∆BTmax = 0.25 Kelvin
  • L. Strow, UMBC

3

slide-4
SLIDE 4

September 2002 AIRS Science Team Meeting

Clear Test

The clear test compares estimated surface temperatures for some subset of

  • channels. Observed window channel radiances are converted to surface

temperatures by adjusting them for the estimated atmospheric effects using the ECMWF model profile. The test channels are distributed into bins to allow channel (ie bin) averaging prior to comparing the temperatures. The bin temperature comparisons can either between two bins or between a bin and the model surface temperature.

  • L. Strow, UMBC

4

slide-5
SLIDE 5

September 2002 AIRS Science Team Meeting

Clear Test Procedures

  • use the nearest ECMWF profile and sea surface emissivity
  • derive an effective sea surface temperature for the test channels by

subtracting out the atmospheric effects. This also requires a sea surface emissivity model.

  • skip if Teff < 273 K since it is ice not open sea
  • sort individual channel surface temperatures into averaging bins
  • ignore shortwave test channels if daytime
  • compute bin average surface temperature
  • compare temperatures between bins
  • if delta BT is larger than max allowed difference, profile fails
  • ∆BT < 4 Kelvin for comparisons with ECMWF model Tsea
  • ∆BT < 0.4 Kelvin for comparisons between bins
  • max land fraction = 0.001
  • L. Strow, UMBC

5

slide-6
SLIDE 6

September 2002 AIRS Science Team Meeting Channel Bins bin id1 id2 id3 id4 id5 id6 id7 1 555 559 562 2 757 758 760 764 766 769 3 843 847 848 853 857 858 4 892 893 896 897 902 903 5 1291 1292 1293 1297 1298 6 2213 2214 2215 2216 2217 2218 2219 7 2327 2328 2331 2334 2346 2361 8 2388 Bin comparisons bin1 bin2 ∆T 1 2 0.4 2 3 0.4 3 4 0.4 4 5 0.4 4 4 4 7 0.4 6 7 0.4 8 4

  • L. Strow, UMBC

6

slide-7
SLIDE 7

September 2002 AIRS Science Team Meeting

800 1000 1200 1400 1600 1800 2000 2200 2400 2600 220 230 240 250 260 270 280 290 Wavenumber (cm−1) B(T) in K Clear Test Channels

  • L. Strow, UMBC

7

slide-8
SLIDE 8

September 2002 AIRS Science Team Meeting

2002.07.20 AIRS Visible False Colour Granule029 10 20 30 40 50 60 70 80 90 20 40 60 80 100 120

  • L. Strow, UMBC

8

slide-9
SLIDE 9

September 2002 AIRS Science Team Meeting

Forward Model Sensitivity to SRF Parameters

250 300 B(T) in K −0.4 −0.2 0.2 0.4 −0.4 −0.2 0.2 0.4 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 −0.4 −0.2 0.2 0.4 Wavenumber (cm−1) <−−−− Error in K −−−−> a. b. c. d.

1% centroid error

2% width error fringe error

  • L. Strow, UMBC

9

slide-10
SLIDE 10

September 2002 AIRS Science Team Meeting

Bias Calculations

  • ECMWF used for global bias calculations.
  • Bias calculations only over water, generally night only
  • Restrict latitude to ±60 degrees
  • All scan angles included, generally nadir is 2-3X more likely to be clear

than 45 degrees

  • Raob and ARM-CART site results extremely preliminary
  • L. Strow, UMBC

10

slide-11
SLIDE 11

September 2002 AIRS Science Team Meeting

220 240 260 280 B(T) in K

  • Sept. 10 Bias and Std

−1 1 2 O−C in K 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 0.5 1 1.5 2 Std in K Wavenumber (cm−1)

  • L. Strow, UMBC

11

slide-12
SLIDE 12

September 2002 AIRS Science Team Meeting

Fringe Effects, Aug 18, Post-Defrost

−2 −1 1 2 Bias in K w/o fringes w/ fringes 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 −0.6 −0.4 −0.2 0.2 0.4 ∆ B(T) in K Wavenumber (cm−1) Effect of fringes

  • L. Strow, UMBC

12

slide-13
SLIDE 13

September 2002 AIRS Science Team Meeting

Fringe Effects, Aug 18, Post-Defrost, Zoom

1250 1260 1270 1280 1290 1300 1310 1320 1330 −0.5 0.5 ∆ B(T) in K Wavenumber (cm−1) Effect of fringes −1 −0.5 0.5 1 Bias in K w/o fringes w/ fringes

  • L. Strow, UMBC

13

slide-14
SLIDE 14

September 2002 AIRS Science Team Meeting

Fringe Effects, Aug 18, Post-Defrost, Zoom

2190 2200 2210 2220 2230 2240 2250 −0.5 0.5 ∆ B(T) in K Wavenumber (cm−1) Effect of fringes −0.5 0.5 1 1.5 Bias in K w/o fringes w/ fringes

  • L. Strow, UMBC

14

slide-15
SLIDE 15

September 2002 AIRS Science Team Meeting

Bias Stability: Sept 10 vs Aug 31 Global Bias

220 240 260 280 B(T) in K (Sept 10 − Aug 31) ECMWF Derived Bias 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 −0.2 0.2 0.4 Wavenumber (cm−1) ∆ Bias

  • L. Strow, UMBC

15

slide-16
SLIDE 16

September 2002 AIRS Science Team Meeting

Mean and Std of Daily Global Bias over 7 days in Sept.

220 240 260 280 B(T) in K −1 1 2 O − C in K 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 0.1 0.2 0.3 Std in K

  • L. Strow, UMBC

16

slide-17
SLIDE 17

September 2002 AIRS Science Team Meeting

Mean and Std of Daily Global Bias over 7 days in Sept.

220 230 240 250 B(T) in K 0.5 1 1.5 O − C in K 650 655 660 665 670 675 680 0.1 0.2 Std in K

  • L. Strow, UMBC

17

slide-18
SLIDE 18

September 2002 AIRS Science Team Meeting

Mean and Std of Daily Global Bias over 7 days in Sept.

220 240 260 280 B(T) in K −0.5 0.5 1 O − C in K 690 700 710 720 730 740 750 0.1 0.2 Std in K

  • L. Strow, UMBC

18

slide-19
SLIDE 19

September 2002 AIRS Science Team Meeting

Mean and Std of Daily Global Bias over 7 days in Sept.

250 260 270 280 290 B(T) in K −1 1 O − C in K 800 850 900 950 1000 1050 1100 0.1 0.2 0.3 Std in K

  • L. Strow, UMBC

19

slide-20
SLIDE 20

September 2002 AIRS Science Team Meeting

Mean and Std of Daily Global Bias over 7 days in Sept.

220 240 260 280 B(T) in K −1 1 2 O − C in K 1250 1300 1350 1400 1450 1500 1550 1600 0.1 0.2 0.3 Std in K

  • L. Strow, UMBC

20

slide-21
SLIDE 21

September 2002 AIRS Science Team Meeting

Mean and Std of Daily Global Bias over 7 days in Sept.

220 240 260 280 B(T) in K −1 1 O − C in K 2200 2250 2300 2350 2400 2450 2500 2550 2600 2650 0.05 0.1 0.15 Std in K

  • L. Strow, UMBC

21

slide-22
SLIDE 22

September 2002 AIRS Science Team Meeting

ARM-CART July 25 Overpass

220 240 260 280 B(T) in K −2 2 O−C in K UMBC CKD2.4 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 1 2 Std in K Wavenumber (cm−1)

  • L. Strow, UMBC

22

slide-23
SLIDE 23

September 2002 AIRS Science Team Meeting

ARM-CART July 25 Overpass, Zoom

240 260 280 B(T) in K −2 2 O−C in K UMBC CKD2.4 1250 1300 1350 1400 1450 1500 1550 1600 0.5 1 1.5 Std in K Wavenumber (cm−1)

  • L. Strow, UMBC

23

slide-24
SLIDE 24

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

24

slide-25
SLIDE 25

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

25

slide-26
SLIDE 26

September 2002 AIRS Science Team Meeting

220 240 260 280 B(T) in K PREPQC Sondes, Sept 1−11, Bias and Std −2 2 O−C in K 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2 4 6 Std in K Wavenumber (cm−1)

  • L. Strow, UMBC

26

slide-27
SLIDE 27

September 2002 AIRS Science Team Meeting

220 240 260 280 B(T) in K Sonde Stats: Sept. 1−11 −1 1 O−C in K 650 700 750 0.5 1 1.5 2 Std in K Wavenumber (cm−1)

  • L. Strow, UMBC

27

slide-28
SLIDE 28

September 2002 AIRS Science Team Meeting

−1 1 O−C in K 220 240 260 280 B(T) in K July 20 Global Bias/Std with Pre−Launch RTA 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 0.5 1 1.5 2 2.5 Std in K Wavenumber (cm−1)

  • L. Strow, UMBC

28

slide-29
SLIDE 29

September 2002 AIRS Science Team Meeting

AIRS Bias with ECMWF, CO2 Corrected Using Laboratory Data

220 240 260 280 B(T) in K 1350 1400 1450 1500 1550 1600 −1 1 O−C in K Wavenumber (cm−1)

CKD2.4, Orig CO2 UMBC H2O Cont, Mod. CO2

  • L. Strow, UMBC

29

slide-30
SLIDE 30

September 2002 AIRS Science Team Meeting

Internal Consistency vs Accuracy, Water Vapor Continuum

220 230 240 250 260 −2 −1.5 −1 −0.5 0.5 1 Obs B(T) O−C in K CKD2.4 Cont. 1480 1500 1520 1540 1560 1580 1600 230 240 250 0.5 1 1.5 2 2.5 Obs B(T) UMBC Cont.

  • L. Strow, UMBC

30

slide-31
SLIDE 31

September 2002 AIRS Science Team Meeting

ECMWF SHIS in the 1600 cm−1 region

200 220 240 260 Mean Obs 1540 1560 1580 1600 1620 1640 1660 1680 −10 −5 5 Wavenumber cm−1 Obs−Calcs CKD2.4 UMBC

  • L. Strow, UMBC

31

slide-32
SLIDE 32

September 2002 AIRS Science Team Meeting

ECMWF SHIS in the 1400 cm−1 region

1300 1350 1400 1450 1500 200 220 240 260 280 Mean Obs 1300 1350 1400 1450 1500 −4 −2 2 Wavenumber cm−1 Obs−Calcs CKD2.4 UMBC

  • L. Strow, UMBC

32

slide-33
SLIDE 33

September 2002 AIRS Science Team Meeting

AIRS: kCARTA versus GENLN2 AIRS Bias with ECMWF

220 240 260 280 B(T) in K 715 720 725 730 735 740 745 750 −1.5 −1 −0.5 0.5 1 O−C in K Wavenumber (cm−1) GENLN2 kCARTA

  • L. Strow, UMBC

33

slide-34
SLIDE 34

September 2002 AIRS Science Team Meeting

WINTEX: O-C With and Without P/R Mixing

200 220 240 260 280 B(T) in K 710 720 730 740 750 −2 −1.5 −1 −0.5 0.5 Wavenumber (cm−1) Obs − Calc in K PR−Mixing Cousin

  • L. Strow, UMBC

34

slide-35
SLIDE 35

September 2002 AIRS Science Team Meeting

ECMWF vs S-HIS during CLAMS

710 720 730 740 750 −1.5 −1 −0.5 0.5 Wavenumber (cm−1) Obs − Calc in K PR−Mixing Cousin 220 240 260 280 B(T) in K

  • L. Strow, UMBC

35

slide-36
SLIDE 36

September 2002 AIRS Science Team Meeting

200 250 300 B(T) in K

  • ldkuniform_clear231.rtp Bias and Std

2330 2340 2350 2360 2370 2380 2390 2400 2410 2420 0.5 1 Std in K Wavenumber (cm−1) −1 1 2 O−C in K AIRS Mod kCARTA Lab Mod kCARTA

  • L. Strow, UMBC

36

slide-37
SLIDE 37

September 2002 AIRS Science Team Meeting

2386 2388 2390 2392 2394 2396 −0.04 −0.02 0.02 Obs − Calc Wavenumber (cm−1) 0.2 0.4 0.6 0.8 1 Transmittance Theory Data

  • L. Strow, UMBC

37

slide-38
SLIDE 38

September 2002 AIRS Science Team Meeting

Effect on Non-LTE on Sounding Channels

240 260 280 B(T) in K 2220 2240 2260 2280 2300 2320 2340 2360 2380 2400 5 10 O−C in K Wavenumber (cm−1)

  • L. Strow, UMBC

38

slide-39
SLIDE 39

September 2002 AIRS Science Team Meeting

Hitran 2K

Lineshapes Spectroscopy

UMBC-LBL

Monochromatic Absorption k's

SVD Compression

kCompressed Database kCARTA

Layer to Space Transmittances

SRF Convolutions

SRF's and Frequencies

Convolved Transmittances Transmittance Regressions

RTA Coefficients AIRS-RTA

AIRS Computed Radiances

Profiles Layering

Validation Data QA Reflected Thermal Inputs

  • L. Strow, UMBC

39

slide-40
SLIDE 40

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

40

slide-41
SLIDE 41

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

41

slide-42
SLIDE 42

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

42

slide-43
SLIDE 43

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

43

slide-44
SLIDE 44

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

44

slide-45
SLIDE 45

September 2002 AIRS Science Team Meeting

  • L. Strow, UMBC

45

slide-46
SLIDE 46

September 2002 AIRS Science Team Meeting

Conclusions

  • Radiometric accuracy looks very good, even for low radiances
  • Mid-altitude temperature channels agree to within 0.25-05K with ECMWF
  • Higher altitude channels at 15 microns suggest ECMWF is biased cold
  • 4 micron CO2 temperature channels are inconsistent with 15 micron

channels by ∼ 1K. We suspect the spectroscopy, with some SRF errors mixed in. More work needed.

  • Validation of the AIRS-RTA for water channels will be difficult. We need all

the validation data we can get with good sondes (RS-90’s) and lidar. ECMWF and PREPQC seem to give the same biases. Aircraft and laboratory data suggests both are wrong. Higher altitude water biased low in sondes/ECMWF, as expected.

  • non-LTE will limit use of some higher-altitude channels at 4.3 microns.

Model calculations qualitatively agree with observations (away from South Polar regions).

  • L. Strow, UMBC

46

slide-47
SLIDE 47

September 2002 AIRS Science Team Meeting

  • We need to emphasize the many new products AIRS can produce in order

to generate excitement at NASA HQ and in the scientific community, e.g. CO, CO2, CH4, cirrus properties, land surface properties, etc.

  • But first, we need to show that cloud-cleared radiances do not

significantly change NCEP/ECMWF bias and standard deviations.

  • We can “tune” the AIRS-RTA via transmittance adjustments. Should be

more physical (more accurate) than tuning radiances.

  • Determining the truth will take time, but is important for the climate

record.

  • L. Strow, UMBC

47