1. Inter-satellite Calibration of HIRS OLR Time Series
Hai-Tien Lee
University of Maryland, CICS/ESSIC-NOAA
Arnold Gruber
Robert G. Ellingson
Florida State University, Dept. of Meteorology
Istvan Laszlo
NOAA/NESDIS
NOAA/NESDIS Cooperative Research Program 2nd Annual Science Symposium
Madison, WI. July 13-14, 2005

2. Outline Cal/Val Issues for the OLR Climate Data Record (CDR)
Inter-satellite Calibration of HIRS OLR Time Series
Findings and Discussion

3. Cal/Val Issues for the OLR Climate Data Record
What is CDR?
“A time series of measurements of sufficient length, consistency, and continuity to determine climate variability and change.” – Committee on Climate Data Records from NOAA Operational Satellites, NRC/NAS, 2004.

4. Satellite OLR Products

5. Equator Crossing Time for NOAA Polar Orbiters

6. Typical Satellite Track Crossing for a Morning and an Afternoon POESPM

7. Inter-satellite calibration for HIRS OLR
Collocation:
• 1°x1° lat/lon
• ±30 minutes
• n > 1
Homogeneity filter:
• Std error of mean OLR < 1 Wm-2
Satellites
Bias (Wm-2) TN
0.15
N06
1.80
N07
2.13
N08
2.03
N09
Reference
N10
0.53
N11
-5.36
N12
-2.42
N14
-5.14
N15
-3.65
N16
-3.25
HIRS OLR retrievals were collocated for each of the overlapping pairs of POES over a target of 1°x1° within plus/minus 30 minutes. Minimum number of retrievals is 2.
The black curve histogram shows the distribution of the OLR differences. This example is for TIROS-N and NOAA6. (we label TIROS-N as N05)
A homogeneity filter was applied to improve the data quality - only mean OLR with a standard error of less than 1 Wm-2 is used. The red dashed curve histogram shows the distribution of OLR differences from homogeneous scenes, with corresponding scatter plot shown at the bottom.
There are two approaches for the calibration. Method 1 determines mean biases referenced to NOAA9. Method 2 determines linear adjustments also referenced to NOAA9. Table on the right shows method 1 results.

8. Improvement with inter-satellite calibration
The blended HIRS monthly mean OLR data with the adjustments determined by inter-satellite calibration agrees much better with the CERES.
This is the time series of monthly mean OLR averaged over the tropics (20S-20N).
The brown curves are the HIRS OLR monthly mean derived from individual satellites, including NOAA11, 12, 14, 15 and 16, for various periods. The differences in sampling time and the systematic biases caused the deviation within these time series.
The black solid (dotted) line is the blended product of the HIRS OLR that the individual satellite retrievals were first adjusted according to the inter-satellite calibration method 1 (2) results and then temporally integrated.
CERES OLR is from broadband measurements that is considered here as the “ground truth” reference. Apparently, the blended HIRS OLR with the inter-satellite adjustment agrees much better with the CERES than any of the individual ones.
Tropical Mean
Magenta - CERES (TRMM, Terra, Aqua)
Black solid/dotted - HIRS, blended with calibration method 1/2
Brown - HIRS from individual satellites: NOAA11, 12, 14, 15, 16

9. Global Mean Magenta - CERES (TRMM, Terra, Aqua)
Black solid/dotted - HIRS, blended with calibration method 1/2
Brown - HIRS from individual satellites: NOAA11, 12, 14, 15, 16

10. Instrument Continuity & Algorithm Consistency
N10-N09
N11-N10
HIRS/2
vs.
HIRS/2I
HIRS/2
N14-N11
N15-N14
HIRS/2I
vs.
HIRS/3
HIRS/2I

11. A B C D
N05
N06
Collocation occurs in the
rhombus ABCD:
AB & CD: ZA5 > ZA6
BC & AD: ZA6 > ZA5
N06 > N05
N06 < N05

13. Summary
The HIRS OLR inter-satellite calibration method worked very well. It significantly reduced the discontinuity in the OLR time series that were introduced by various sources of inconsistency and errors.
Identified inconsistent OLR retrievals caused by the changes in HIRS instruments and OLR algorithm.
Found some large OLR differences that might point to the limb correction and/or modeling for extreme conditions.
Call for revision of the Operational algorithm.
This procedure may apply to the inter-satellite calibration of other thematic CDR products.

14. Backup Slide

16. Multi-spectral HIRS OLR Algorithm
=local zenith angle
ai=regression coefficients
(Ellingson et al., 1989)

17. Validation of Multi-spectral OLR Algorithms
Ellingson et al., 1994: Validation of a technique for estimating outgoing longwave radiation from HIRS radiance observations J. Atmos. Ocean. Technol., 11,
Ba et al., 2003: Validation of a
technique for estimating OLR with the GOES sounder. J. Atmos. Ocean. Technol., 20, 79–89.

18. Channel selections NOAA16 HIRS/3 GOES8 Sounder GOES8 Imager GOES-R ABI
OLR Spectrum
Mid-lat Summer

19. Earth Radiation Budget
Longwave Radiation Products
Outgoing Longwave Radiation at the top of the atmosphere
Atmospheric Layer Heating Rate
Downward Longwave Radiation at the earth’s surface
Kiehl and Trenberth, Bull. Amer. Meteor. Soc., 78,