The Inter-Calibration of AMSR-E with WindSat, F13 SSM/I, and F17 SSM/IS Frank J. Wentz Remote Sensing Systems 1 Presented to the AMSR-E Science Team June 28, 2011 NCDC, Asheville, NC
The Inter-Calibration of AMSR-E with WindSat, F13 SSM/I, and F17 SSM/IS
Frank J. WentzRemote Sensing Systems
1
Presented to the AMSR-E Science TeamJune 28, 2011NCDC, Asheville, NC
The Problem• Volume: Nearly 100 satellite-years of observations from Microwave Radiometers.• Calibration: Each sensor has its own unique set of Sensor Calibration Problems• Precision: High precision required for Climate Studies
Satellite Microwave Radiometers
3
Satellite Calibration Methodology
4
F13 SSM/I, F17 SSM/IS, AMSR-E, and WindSat have been finalized at the RSS V7 Calibration Standard: Both TA/TB and Geophysical ProductsSST, wind, vapor, cloud, and rain have been thoroughly validated
Direct In Situ Comparisons Inter-satellite Comparisons A very large number of research studies
Geophysical products, including rain, are consistent across all platforms
Geophysical Products
Radiative Transfer Model
Simulated Antenna Temperature
Measured Antenna Temperatures
TA Anomaly Statistics
Satellite Calibration MethodologyFinal Step
0 50 100 150 200 250 300 3500
50
100
150
200
250
300
350
0
1
2
3
4
5
6
7
8
9
10x 10
4
Port Symmetry
WindSat 10.7 GHz Global TA(v) versus TA(h)TA(v) ≈T(h) over hot land: Indicate that ηV = ηh
5
ocean
land
0 K V-pol TA 350 K
0 K
H-p
ol T
A
3
50 K
6
1e c h
Bh c
C C TTC C
11e
Bh c h
c h hC C TTC C
TT
Absolute Errors in Hot Load and Spillover are Equivalent
Simplified TB equation (cold space and cross-pol set to 0)
Change in TB due to errors in specifying hot load temperature and spillover
Hot-Load/Spillover error dominates absolute calibration error Typical values for past satellites is 1 to 2 K (previous table) GMI Specification is 1.35 K RSS post-launch calibration assumes this error is the same for both polarizations with a few exceptions.
7
7 GHz 11 GHz 19 GHz 23 GHz 37 GHz 90 GHz
F13 -1.1 -0.5 1.2 1.1
F16 -0.8 -1.4 -0.9 1.6
F17 -1.7 -2.2 -0.5 2.8
AMSR-E -0.3 0.5 -0.1 0.3 0.9 2.0
WindSat 2.3 0.8 -1.3 -1.2 -0.6
Standard Rules Applied to All Sensors:0.3 K added to Planck cold space value (except JAXA added 0.7 K for AMSR-E)1.0 K subtracted from thermistor hot load readings
RSS derived Spillover minus pre-launch value reported by NRL or JAXA Spillover difference multiplied by 200 K to show typical TB difference in Kelvin
Color Coding: < -2K -2K to -1K -1K to +1K +1K to +2K >+2K
Absolute Calibration: Spillover
8
7 GHz 11 GHz 19 GHz 23 GHz 37 GHz 90 GHz
F13 0.1 0.3 0.5 0.7
F16 -0.2 0.0 0.2 0.0
F17 0.2 0.0 -0.1 0.1
AMSR-E 0.1 0.3 0.3 0.0 0.1 0.0
WindSat 0.1 -0.1 0.3 0.1 0.0
RSS derived Cross-Pol minus pre-launch value reported by NRL or JAXA Spillover difference multiplied by 50 K to show typical TB difference in Kelvin
Color Coding: < -0.5K -0.5K to -0.3K -0.2K to +0.2K + 0.3 to 0.5K >0.5K
Absolute Calibration: Cross-Pol
A compelling validation of the radiative transfer modelGMI has extremely small cross-pol precise RTM validation
9
A One-Parameter Absolute Calibration Model
For each frequency, one tuning parameter is requiredto match the brightness temperature observations to the RTM.
This single parameter is sufficient for both polarizations and for both ocean and land and presumably sea ice and snow.
This tuning parameter accounts for the combined knowledge error of the effective hot load temperature and spillover
The magnitude of this parameter is 1 - 2 K.
10
A One-Parameter Absolute Calibration ModelCaveats
1. For fine turning, we do make small (0-0.3K) adjustments to cross-pol.
2. For 3 cases out of 23, we find a polarization difference in spillover:AMSR-E 23.8 GHz 0.3 K @ 270KAMSR-E 89.0 GHz 0.5 K @ 270KF17 SSM/I 91.7 GHz 1.0 K @ 270K
All 3 cases show this polarization anomaly over both ocean and land
3. The RSS RTM is used as the Reference.Note however that when averaged over the ensemble of MW sensor, the absolute calibration adjustments are close to zero
4. There are relative adjustments that are made that are functions of scan position and sun angles
Note however these adjustments are all bias neutral.
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
11
7 GHZ, Hpol 11 GHZ, Hpol
19 GHZ, Hpol 23 GHZ, Hpol
37 GHZ, Hpol 89 GHZ, Hpol
Mission TA Anomaly Plots AMSR-E Problem at 19 GHz
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
1212
7 GHZ, Hpol 11 GHZ, Hpol
19 GHZ, Hpol 23 GHZ, Hpol
37 GHZ, Hpol 89 GHZ, Hpol
WindSat (750 MHz)
AMSR-E (200 MHz)
AMSR-E Adjusted
Mission TA Anomaly Plots60 Megahertz change to AMSR-E 18.7 GHz channel
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
1313
7 GHZ, Hpol 11 GHZ, Hpol
19 GHZ, Hpol 23 GHZ, Hpol
37 GHZ, Hpol 89 GHZ, Hpol
1313
Mission TA Anomaly Plots AMSR-E Cloud-Crosstalk Problem
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
0
0.02
0.04
0.06
0.08
0.1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
0
0.02
0.04
0.06
0.08
0.1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
0
0.02
0.04
0.06
0.08
0.1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
0
0.02
0.04
0.06
0.08
0.1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
0
0.02
0.04
0.06
0.08
0.1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
0
0.02
0.04
0.06
0.08
0.1
14
Location of Persistent Clouds
15
Problem with Cloud Cross-Talk In Vapor Retrieval Algorithm
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
1616
7 GHZ, Hpol 11 GHZ, Hpol
19 GHZ, Hpol 23 GHZ, Hpol
37 GHZ, Hpol 89 GHZ, Hpol
1616
Mission TA Anomaly Plots AMSR-E Cloud-Crosstalk Problem
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
03 04 05 06 07 08 09 10S.P.
Eq.
N.P.
Eq.
S.P.
-1
-0.5
0
0.5
1
1717
7 GHZ, Hpol 11 GHZ, Hpol
19 GHZ, Hpol 23 GHZ, Hpol
37 GHZ, Hpol 89 GHZ, Hpol
Mission TA Anomaly Plots AMSR-E Cloud-Crosstalk Removed
18
ConclusionsProper calibration requires that one starts with the raw radiometer counts and then apply consistent methods and algorithms to all sensors.
Major sources of calibration error are:1. Specification of antenna spillover and mean hot load temperature2. Specification of the variation of hot load temperature with varying sun angles3. Emissive antenna correction (TMI and SSM/IS)
There are many minor sources of calibration error4. Small Cross-Pol Adjustments5. Along-scan biases6. Moon in Cold mirror7. Etc.
The RSS V7 Calibration Standard is a highly evolved RTM that is applied to all satellite sensors.
AMSR-E, WindSat, F13 SSM/I and F17 SSM/IS are now at the RSS V7 Calibration Standard. The remaining SSM/I will soon follow.