1. Dorothee Coppens

2. Outline of the presentation
IASI processing stability
IASI-A reprocessing
Purpose
Validation of the improvements
Schedule
3) IASI-B Non-linearity update
Why changing the non-linearity correction?
Preliminary results
Next steps

3. 1) IASI-A used as GSICS IR reference
IASI characteristics:
Continuous spectral coverage: cm-1
Spectral resolution: 0.5 cm-1 (sampling of 0.25 cm-1)
Spectral accuracy better than 2 ppm
IASI L1 products quality:
Radiometric and spectral calibrations are done every 8 seconds
Level 1c products are harmonized (IASI SRF removed)  Easy use of the data for the users
IASI L1 processing stability:
Accurate IASI on-board/on-ground monitoring to perform updates before any degradation can be observed 
Only one small adjustment in 10 years !!! (February 2011)

4. IASI-A reprocessing

5. 2) IASI-A reprocessing IASI calibration is very stable since 10 years
Small remaining instrument effect before 2011 in pixel 2
Format evolution in 2010 based on operations needs:
 New information has been added on the L1 products:
Cloud information
Detailed on-board flagging for the monitoring
To take the opportunity to fill the gaps in the data records due to any anomaly on the processing chain

6. 2) Improvement of Spectral Harmonisation
Since July 2007, inter-pixel radiance differences in the IASI level 1C products exceeding 0.1 K were observed in some regions of the spectra
The on-ground processing algorithm has been update in February 2011.
Reference file
Reprocessed file

7. 2) New Information in the Products: Cloud information
Cloud and land mask information from AVHRR have been included in the IASI products which did not exist in the processed data before 2010
The following figure shows the cloud and land fraction in a reprocessed product from October 2007:
Cloud Fraction
Land Fraction

8. 2) IASI-A reprocessing: Filling Gaps
Missing products due to some anomaly are recovered in the reprocessed products.
Reference file
Reprocessed file
IASI-A reprocessing

9. IASI-B non-linearity correction

10. 3) Why Changing the non-linearity correction ?
Comparison IASI-B with:
IASI-A
 Radiometric residual bias, higher in band 1 ~ 0.1 K

11. 3) Why Changing the non-linearity correction ?
Comparison IASI-B with:
Observed bias of -0.1K
AIRS
 Observed bias of -0.05K
CrIS

12. 3) EUMETSAT: Final ISSWG Statement in June 2017
 The justifications were:
The new corrections will improve the quality of IASI data
The correction will be implemented for IASI-C  Must be applied on IASI-B in order to get a IASI-B / IASI-C consistent system
The sooner the better, as IASI-B will replace IASI-A as the GSICS reference
The request to apply the correction on board IASI-A in order to establish a consistent system IASI-A / IASI-B / IASI-C during a period before Metop-A drift becomes too large
Delay of 6 months between the IASI-B and IASI-A corrections should be sufficient to characterize the effect of the change on IASI-B alone
Implementation schedule following a clear statement from ISSWG:
The non-linearity correction for IASI-B shall be updated as soon as possible
 New non-linearity implemented on-board as for 2 August 2017
The non-linearity correction for IASI-A shall be implemented 6 months after IASI-B

13. 3) Preliminary results in December 2017 – CNES
Comparison IASI-B with:
IASI-A
AIRS
CrIS
 Better agreement with IASI-A
Residual bias of +0.05K with AIRS
 Residual bias of +0.1K with CrIS

14. 3) Status of the new Non-Linearity – EUMETSAT
Based on OBS-CALC methodology
CALC are done with:
RTM: RTTOV 10.2
Temperature, humidity and O3 from ECMWF forecasts
Selection of scenes:
close to nadir (satza < 20 deg.)
night time (sunza ≥ 120 deg.)
No polar regions (|lat| ≤ 60 deg.)
Over sea
clear sky

15. 3) IASI cold scenes
Following Manning (SPIE, 2017), we average IASI scenes with Tb=200 ± cm-1 i.e. thick tropical clouds
These scenes are expected to be identical from a geophysical point of view; and indeed, the s of the mean spectrum is below 0.2%  possibility to compare different instruments

16. 3) Cold Scenes – results IASI/B-IASI/A
2014
2015
2016
2018

17. 3) IASI warm scenes
A similar approach can be followed for warm scenes: we average IASI scenes with Tb=288 ± cm-1
The variability is much higher; however, using grand averages, it is still possible to compare instruments from year to year

18. 3) Warm scenes – results IASI/B-IASI/A
2014
2015
Current status of the new Non-Linearity – EUMETSAT
Current status of the new Non-Linearity – EUMETSAT
2016
2018

19. 3) Status in March 2018
On EUMETSAT side: the first results are promising and show that IASI-A should be updated as well
First estimation from ISSWG members were inconsistent  Need more time to have a full picture of the impact
Update of IASI-A in February 2018 has been put on hold
A Task Force (ISSWG members) has been organised – collocation meeting is April 6th, 2018.
Outcome of the task force will be discussed at the ISSWG and presented at the IASI RevEx, in June 2018, where the decision will be taken regarding the update of IASI-A (and IASI-C)

20. Summary
IASI stability: Very stable since 10 years, with only one small adjustment in 2011.
IASI-A reprocessing:
IASI-A has been reprocessed for the period to:
Update the spectral harmonization before 2011
Include the cloud/land information, missing before 2010
Fill small gaps in the data records
Validation will be completed by the end of March 2018
3) IASI-B non-linearity correction
IASI-B first results are encouraging
More time is needed on the users side to assess the impact
A Task Force (ISSWG members) has been organised – collocation meeting is April 6th, 2018  Trigger the IASI-A update or not