1. Tim Hewison1 and all GSICS Developers EUMETSAT
01 May 2019
GSICS Corrections and Bias Monitoring for IR channels of GEO imagers. Data Access. Including plans to migrate references & handling diurnal variation
Tim Hewison1 and all GSICS Developers
EUMETSAT

2. Outline of Presentation
Introduction to Types of GSICS Products
GSICS Correction
GSICS Bias Monitoring
GSICS Reports & Guidelines
IR channels of GEO imagers
Current Status of GEO-LEO IR hyperspectral inter-calibration products
Data Access
Plans to Migrate References
Handling diurnal cycles & midnight calibration problems
Summary

3. GSICS Product Portfolio
Currently in Demonstration Mode
For Operational Satellites
− Geostationary – IR & Solar
− LEO – IR, Solar and Microwave – Conical & Cross-track Scanners
− Current Operational & Historic Instruments
− In near real-time and re-analysis modes
GSICS Bias Monitoring
− Routine comparisons of satellite radiances against reference
GSICS Correction
− Function to correct issued radiances
− For consistent calibration with reference
GSICS Reports & Guidelines
− Recommendations to modify practices
− Design and Operation of future satellite instruments

4. GEO-LEO IR: Basic Methodology
Comparison of level 1 data (radiances)
− issued operationally from the Monitored satellite instrument
− with those of a Reference instrument (or model)
Ensuring these are comparable
− Transformations (spectral, spatial, temporal)
− Including sampling/homogeneity checks
Typically ~1000 comparable samples/day for GEO-LEO
These are analysed to identify any significant sensitivities
− e.g. Incidence angle, latitude, radiance, time
and combined where possible
− defining the domain of validity and smoothing period

5. GSICS Products: Bias Monitoring
xMON ∆x
Comparing samples of xMON v xREF
− Over fixed domain,
− Period (e.g. 1 orbit/1 day)
Comparison by regression
Calc bias, ∆x=xMON-xREF
− at standard scene, xSTD
− with uncertainty
Plot time series of bias ∆x
− Compare recent results with long-term trend
− Valuable for instrument monitoring
xSTD
xREF ∆x t

6. GSICS Products: GSICS Correction
xREF
xMON
GSICS Correction function
Compare all xREF v xMON samples
− over smoothing period
Regression coefficients
with uncertainty (covariance)
Provide a function users can apply
to convert level 1 data, xMON
to be consistent with calibration of reference, xREF
Two versions:
Near Real-Time (asymmetric)
Re-Analysis (symmetric) ∆x t
Time series of bias estimated from:
– GSICS Near Real-Time Correction
– GSICS Re-Analysis Correction

7. GSICS Products: Guidelines
Underlying assumption of GSICS Correction:
Small errors (e.g. SRF errors, blackbody temperature, ...) introduce small departures from ‘true’ calibration
If these are linearly related a predictor (radiance, time, ...) we can apply empirical correction based on inter-calibration
Can analyse GSICS products
to diagnose root causes of calibration errors
Derive recommendations to modify
operating practices (e.g. adopt new SRF definition),
pre-launch characterisation, etc.
These GSICS Guidelines are distributed as written reports

8. Outline of Presentation
Introduction to Types of GSICS Products
GSICS Correction
GSICS Bias Monitoring
GSICS Reports & Guidelines
IR channels of GEO imagers
Current Status of GEO-LEO IR hyperspectral inter-calibration products
Data Access
Plans to Migrate References
Handling diurnal cycles & midnight calibration problems
Summary

9. GEO-LEO IR Product Status 2011-09
GPRC
Monitored Instrument
Reference Instrument
GSICS NRT Correction
GSICS Re-Analysis Correction
GSICS Bias Monitoring
EUMETSAT
Meteosat-9 }
Meteosat-8 }--
Meteosat-7 }
IASI
Demonstration
Prototype
JMA
MTSAT-1R }
MTSAT-2 }
IASI (+ AIRS)
NOAA
GOES-11 Imager
GOES-12 Imager
In development
GOES Sounder
CMA
FY2C }
FY2D } --
FY2E }
KMA
COMS

10. Comparison of Collocated Radiances
01 May 2019
Comparison of Collocated Radiances
Simultaneous near-Nadir Overpass
of GEO imager and LEO sounder
Collocation Criteria:
ΔLat<35° ΔLon<35°
Δt < 5 mins
Δsecθ < (Atmospheric path diff.)
Concentrated in tropics
~1000 collocations/orbit
~1 orbit/night

11. Data Transformations (Spectral and Spatial)
01 May 2019
Data Transformations (Spectral and Spatial)
Spectral Convolution:
Convolve LEO Radiance Spectra with GEO Spectral Response Functions
to synthesise radiance in GEO channels
Spatial Averaging:
Average GEO pixels in each LEO FoV
Estimate uncertainty
due to spatial variability
as Standard Deviation of GEO pixels
Use in weighted regression
LEO FoV~10km
~ 3x3 GEO pixels
The next step is to transform the collocated data to allow direct comparisons between the instruments.
Firstly, the IASI radiance spectra is convolved with the Spectral Response Functions of the Meteosat channels, to calculate the expected radiance in each Meteosat channel.
Then the Meteosat pixels within each IASI iFoV are averaged.
Their variance is also calculated to quantify the uncertainty due to spatial variability and
is used to weight each collocation in a linear regression …

12. GSICS Products for GEO IR
xREF
xMON
GSICS Correction function
GSICS Monitoring
GSICS Corrections
Near Real-Time & Re-Analysis
for Monitored instrument
IR channels of GEO imagers
against reference instrument
IASI (hyperspectral)
and AIRS (transfer radiometer)
by direct comparison
of collocated radiances
Typical Corrections ~1K
For GEO IR channels t ∆x
Time series of bias estimated from:
– GSICS Near Real-Time Correction
– GSICS Re-Analysis Correction

13. GEO-LEO IR Uncertainty Evaluation
Extended Error Budget for Meteosat-IASI GSICS Corrections
Following QA4EO / GUM
Processes at each step of ATBD introducing
Random Uncertainties
Dominated by spatial/temporal variability
over 3km/300s
Validated using time series statistics
Systematic Uncertainties
Dominated by spatial/temporal mismatches
Total uncertainties depend on radiance
Mostly dominated by random processes
Errors much lower in WV channels
Validated quoted errors as Quality Indicators

14. Outline of Presentation
Introduction to Types of GSICS Products
GSICS Correction
GSICS Bias Monitoring
GSICS Reports & Guidelines
IR channels of GEO imagers
Current Status of GEO-LEO IR hyperspectral inter-calibration products
Data Access
Plans to Migrate References
Handling diurnal cycles & midnight calibration problems
Summary

15. GSICS Data Access GSICS Bias Monitoring (prototype) GSICS Corrections
Hosted on websites of GSICS Processing & Research Centres (GPRCs)
GSICS Corrections
netCDF format, CF convention
GSICS Data & Products Servers
THREDDS-based system
See gsics.wmo.int for links

16. Outline of Presentation
Introduction to Types of GSICS Products
GSICS Correction
GSICS Bias Monitoring
GSICS Reports & Guidelines
IR channels of GEO imagers
Current Status of GEO-LEO IR hyperspectral inter-calibration products
Data Access
Plans to Migrate References
Handling diurnal cycles & midnight calibration problems
Summary

17. Combining Multiple References
Advantages
Disadvantages
Robustness
In case of failure of one reference
Allows transition between references – e.g. MetopA->B
Greater coverage of diurnal cycle
Both scene and instrument calibration variability
Important for 3-axis stabilized spacecraft
Political
Segal's law:
"A man with a watch knows what time it is. A man with two watches is never sure." [Bloch, Arthur (2003). Murphy's Law. Perigee. p. 36.]
It refers to the potential pitfalls of having too much conflicting information when making a decision.
User confusion
Metrological Traceability Problems
as the references can never be perfectly consistent
Define only one as the calibration reference
All others are regarded as calibration transfer standards

18. GSICS Correction using Reference 1
GSICS Correction for Monitored Instrument
Based on Reference 1
Function with uncertainty

19. GSICS Correction using Reference 2
GSICS Correction for Monitored Instrument
Based on Reference 2
Different function with different uncertainty

20. Delta Correction Reference 2->1
GSICS Correction for Reference 2
Based on Reference 1
Function with uncertainty

21. Delta Correction Reference 2->1
Delta GSICS Correction for Reference 2
Based on Reference 1
Function with uncertainty

22. GSICS Correction using Reference 2->1
Add Delta Correction to Reference 2
Maps GSICS Correction based on Reference 2 to Reference 1
Consistent function, but larger uncertainty

23. Implementation GSICS Correction Correction using Reference 1
Delta Correction Reference 2-1

24. Handling Diurnal Variations
Some GEO-LEO IR GSICS Corrections show significant diurnal variations
Mostly related to the midnight blackbody calibration problem
Could produce separate corrections for different periods of the diurnal cycle
Users' Guides should clearly specify the range of conditions over which they are valid
in particular, the periods of the diurnal cycle
Based on recent analysis by Fangfang Yu

25. Outline of Presentation
Introduction to Types of GSICS Products
GSICS Correction
GSICS Bias Monitoring
GSICS Reports & Guidelines
IR channels of GEO imagers
Current Status of GEO-LEO IR hyperspectral inter-calibration products
Data Access
Plans to Migrate References
Handling diurnal cycles & midnight calibration problems
Summary

26. Summary GSICS is Global Space-based Inter-Calibration System!
Focuses on Level 1 data => FCDR
GSICS does not (generally) issue re-calibrated archives of data
But provides products to correct calibration of real-time and archive data to be consistent with reference instruments
Also provides tools to monitor biases in near real-time
GSICS Product Types defined:
GSICS Correction – Near Real-Time and Re-Analysis
GSICS Bias Monitoring
GSICS Reports & Guidelines
Now have demonstration products
GSICS Correction of GEO imagers’ IR channels – beta testing
GSICS Calibration of AVHRR solar channels (PATMOS-X) – submitted to GPPA
Need to ensure consistency between products
For different instruments – contemporary and historic

27. 01 May 2019
Thank You
Any Questions?