1. AHI IR Tb bias variation diurnal & at low temperature
Arata Okuyama and Masaya Takahashi
Meteorological Satellite Center, Japan Meteorological Agency
2017 GRWG/GDWG Annual Meeting, March 2017, Madison, USA

2. 1. Diurnal variation analysis of AHI-8 and AHI-9
as an ideal case of GEO-GEO approach, SBAF ~= 1.0, so to say, full-time ray-matching

3. Diurnal variation in IR Tb
Tb Biases at the std.radiance
Himawari-8/AHI
[K]
[K]
MTSAT-2/Imager
[K]
[K]
AHI-8 doesn’t have significant diurnal variation seen in MTSAT-2, but small variation still exists.
[K]
[K]
[K]
[K]
[K]
[K]
[K]
[K]
[K]
Reference sensors:
Metop-A/IASI
Metop-B/IASI
Aqua/AIRS
S-NPP/CrIS
[K]
midnight
Dec. 2014
[UTC] [UTC] [UTC]
Dec. 2015 3

4. AHI-8 and AHI-9 Tb bias at the std. scene
Diurnal variation of Tb bias for AHI-9 is as small as AHI-8.
Hyper-sounder based approach is effective to know diurnal variation, but “gap” exists.
Himawari-8/AHI 　　　　 Himawari-9/AHI
Tb biases [K]
Tb biases [K]
[UTC]
[UTC]
Feb 2017
Feb 2017

5. GEO-GEO approach by AHI-8/9
B08 (6.2 um)
AHI-9
EW gradation?
AHI9 – AHI8
stray light
AHI-9 Tb matches that of AHI-8 very well.
GEO-GEO approach can be of help to know spatial pattern of the Tb bias.
[K]
[K]
AHI-8 [K]
AHI-9 [K]
AHI9 – AHI8 [K]
Averaged Tb in 19x19 pix grid is plotted.
Data region is SSP +/- 30 deg.

6. GEO-GEO approach by AHI-8/9
GEO-LEOs (AHI-9)
GEO-GEO (AHI9–AHI8)
0.0K
-0.5K
+0.5K
Band 8 (6.2 um)
Tb biases [K]
0.0K
-0.5K
+0.5K
Band 9 (6.9 um)
0.0K
-0.5K
+0.5K
Band 10 (7.3 um)
14 – 19 Feb Vertical lines show 00 UTC.
[UTC]
Global averaged Tb difference at the standard temp. The Tb diff. is corrected based on SRF difference.
Feb 2017

7. GEO-GEO approach by AHI-8/9
AHI-9 has Tb bias jump, around 0.1 K, twice per day.
Not so many GEO-LEO collocation data in a timing of the jumps
GEO-GEO approach is effective in such case.
GEO-GEO approach:
is available in 24 hrs.
can reveal diurnal variation in finer temporal resolution.
can illustrate spatial distribution of bias.
is relative comparison, which can’t utilize state-of-the-art LEO instruments.
Best way would be a blending of GEO-LEO + GEO-GEO approaches, but, how to decide the biases and its uncertainty at each time of a day?

8. 2. AHI-8 Tb bias for lower temperature
Response to: Action GIR o.1:
Arata to check how the cold end corrections are behaving using AIRS.

9. From my presentation in the last annual meeting...
Bias in “cold end”
Uncertainty of brightness temperature (Tb) bias tends is larger in lower Tb range, especially in SWIR.
Some meteorologists are interested in cold region.
It is caused by non-linearity between radiance and Tb and is significant in shorter wavelength.
Is such the large bias true?
Tb bias is -5 K at 220 K, -15 K at 210K. The corrected radiance become minus in Tb < 207 K.
Planck’s law says...
[W･m-2･sr-1･(μm)-1]
3.9 um
6.2 um
10.4 um
300 K
0.6021
5.682
9.823
200 K
0.1187
0.9703
Energy is 1/470

10. AHI-8 Tb bias stability for lower temperature
Tb bias is stable.
Estimated Tb bias deviation is larger for lower temperature.
Time series of Tb biases between the hyper sounders and Himawari-8/AHI Band 12 (9.6μm)
w.r.t. S-NPP/CrIS 　　 w.r.t. Aqua/AIRS 　　 w.r.t. Metop-A/IASI
at 290K
at 290K
at 290K
+1.5
+1.5
+1.5
-1.5
-1.5
-1.5
2015
2016
2015
2016
2015
2016
at 250K
at 250K
at 250K
+1.5
+1.5
+1.5
-1.5
-1.5
-1.5
2015
2016
2015
2016
2015
2016
at 220K
at 220K
at 220K
+1.5
+1.5
+1.5
-1.5
-1.5
-1.5
2015
2016
2015
2016
2015
2016
Stable AHI Tb biases w.r.t. CrIS at cold scenes
Daily
29 days statistics for GSICS correction

11. Standard error of Tb bias
The standard error based on AIRS and CrIS tend to be smaller in Band 7, 8, and 11, even though there are wide gap-channels around the bands.
Standard error of Tb bias at 220 K
B07 (3.9 um)
B08 (6.2 um)
B09 (6.9 um)
B10 (7.3 um)
B11 (8.6 um)
B12 (9.6 um)
B13 (10.4 um)
B14 (11.2 um)
B15 (12.4 um)
B16 (13.3 um)
Reference sensors:
Metop-A/IASI
Metop-B/IASI
Aqua/AIRS
S-NPP/CrIS

12. Band 11
Band 8
Band 7

13. Summary GEO-GEO comparison: Tb bias for lower temperature:
The approach is available in 24 hrs,
can reveal diurnal variation in finer temporal resolution.
can illustrate spatial distribution of the Tb bias, and
is relative comparison, which can’t utilize state-of-the-art LEO instruments.
Best way would be a blending of GEO-LEO + GEO-GEO approaches, but, how to decide the biases and its uncertainty at each time of a day?
Tb bias for lower temperature:
Likun commented that the IASI on-board processing is not performing well enough in the cold end and noted that AIRS do not have that particular problem. (c.f. minutes of the last annual meeting)
Confirmed. Action GIR o.1  close

15. Tb bias
Tb bias shows large variation for Band 7, 8, and 11 w.r.t. AIRS and CrIS because of wide gap-channels around the bands.
Tb bias at 220 K
Reference sensors:
Metop-A/IASI
Metop-B/IASI
Aqua/AIRS
S-NPP/CrIS

16. Standard error of Tb bias at 250 K

17. Standard error of Tb bias at 290 K

18. AHI data quality (Tb bias vs location)
What is an uneven plot in the diurnal variation?
Collocation data location depends on time of day and equatorial crossing time of LEO satellite.
Assuming that GEO Tb bias depends on observed location, the uneven plot might be appeared at the timing of reference sensor change.
Further investigation is required.
AHI Band 12 (9.6um) vs. IASI-A
# of collocation data Mean TB bias TB bias in each longitudinal bin
~clear scene
2 x 2 degrees grid statistics for 2015 (Sep. - Nov.)

19. B08
AHI9 – AHI8
AHI8
AHI9