GRWG/GDWG Web Meeting 2020-03-19

GSICS Web Meeting on GEO-LEO Inter-calibration Status in IR sub-group

Due to the postponement of 2020 GSICS annual meeting and as suggested by the GSICS executive panel, the IR group will host a series of web meetings to review all the presentations of the annual meeting. The first one will be held at 6:30-9:30am EDT on Thursday March 19 2020 (check your time here) and focused on GEO-LEO products. We will also discuss IR group research plan. Should you have any urgent issues related to the IR group, please bring them for discussion.

Agenda

1. Likun Wang (NOAA Affiliate) – Introduction and IR Group Activity Plan (5 Min)
2. Clemence Pierangelo (CNES) - Use of the Moon for calibration of IR sensors: IASI case (20 Min)
3. Arata Okuyama (JMA) - JMA GSICS IR product status (20 Min)
4. Fred Wu (NOAA) - GSICS correction for ABI IR bands (20 Min)
5. Ninghai Sun/Banghua Yan (NOAA) - Introduction to STAR ICVS GSCIS Portal (20 Min)
6. Minju Gu (KMA) - KMA GEO-LEO IR product status (20 Min)
7. Tim Hewsion(EUMETSAT) - EUMETSAT GEO-LEO IR product status and plan for ISSCP-NG (20 Min)
8. Dorothee Coppens (EUMETSAT) - Migration / Update on IASI-A/B/C (20 Min)
9. Chengli Qi (CMA) - HIRAS status update and IR recalibration for FY-3/IRAS and VIRR (20 Min)
10. Hui Xu (UMD) - Updates on IASI SW Gap Filling (20 Min)
11. Likun Wang (NOAA Affiliate) -Wrap Up (10 Min)

Attendees

Guest Chair: Likun Wang (NOAA Affliate) and Tim Hewsion(EUMETSAT)

CMA: Chengli Qi, Scott Hu, Hanlie Xu, Li Yuan, Yong Zhang, Na Xu, Qifeng Lu, Yuan Li

CNES: Clemence Pierangelo, Laura Le Barbier

ESA: Stefano Casadio

EUMETSAT: Tim Hewison, Dorothée Coppens, Sebastien Wagner, Alessandro Burini

JMA: Arata Okuyama, Kazuki Kodera

KMA: Dohyeong Kim, Minju Gu, Hyeji Yang, Jun Park,

NASA (Langley): Dave Doelling, Xu Liu, Wan Wu

NOAA: Likun Wang, Hui Xu, Fred Wu, Fangfang Yu, Jingfeng Huang, Mitch Goldberg, Manik Bali, Banghua Yan, Warren Porter, Ninghai Sun, Xin Jin

UW: Joe Taylor

WMO: Heikki Pohjola

Summary

Clemence Pierangelo (CNES) - Use of the Moon for calibration of IR sensors: IASI case

Clemence presented an analysis investigating the potential application of lunar calibration in the thermal infrared to IASI. She explained this required developing a model of the Moon’s surface temperature (based on LRO) and emissivity (based on ECOSTRESS for ASTER), with a split of soil types. Change in phase angle and surface temperature introduce radiance differences of ~1% between two orbits (101 minutes), while changing geometries introduce month-to-month differences ~10%. IASI observations are taken in the Space Views, but need dedicated coding tables to prevent saturation and cover the dynamic range from 90K to 380K.

Importantly, IASI’s Integrated Imaging System (IIS) was able to identify cases when the Moon fell cleanly within each of IASI’s iFOVs, which could be used to minimise the impact of edge effects, which introduce inter-band differences. The end results suggested absolute radiance differences between IASI-A, -B and –C within ±0.4% (<0.2K) were achievable – albeit based on a very small sample.

Q: Is the PSF considered? – Yes.

Q: Do you perform uncertainty analysis? – At this stage, only sensitivity study.

Q: How sensitive is the model to phase angle? - To be checked

Q: How to separate spectral calibration from sensitivity to viewing geometry?

It was also suggested that analysis of the inter-band differences could give information on the band-to-band co-registration.

R.GIR.20200319.1: CNES to consider making further lunar acquisitions with IASI to extend their valuable analysis of the potential of this technique and share with the lunar calibration community.

A.GIR.20200319.1: Sebastien Wagner (EUMETSAT) and Vinia Mattioli (EUMETSAT) to consider renaming the microwave session of the 3^rd Lunar Calibration Workshop to include the TIR.

Arata Okuyama (JMA) - JMA GSICS IR product status

JMA is in the process of switch the production of its GEO-LEO IR products using CrIS from Nominal to Full Spectral Resolution data, as the former will soon not be supported. Arata showed this has a negligible impact for most bands (<0.02K).

A.GIR.20200319.2: Arata Okuyama (JMA) to consider applying Hui Xu’s CrIS gap-filling method to CrIS for AHI. With CrIS gap filling, JMA can further investigate the cold scene bias at 3.9µm channel for by comparing IASI-AHI and CrIS -AHI results, which is also related to previous action A.GIR.2019.7i.2.

Arata also reviewed the uncertainty analysis performed at JMA on its GEO-LEO IR products for AHI, following Hewison, 2013, which confirmed the dominant terms are spatial and temporal variability. However, he pointed out that parallax errors are not explicitly included in this analysis – although they contribute to the spatial sensitivities and their influence can be minimized by applying a homogeneity filter to the collocations. (Post meeting note: Uncertainty analysis: Hewison (2013) also showed random components (day-to-day variation) were underestimated by 2-4 times, leading to Tim’s recommendation to inflate the current uncertainties, as his assessment misses some sources, such as the cloud parallax.)

With this analysis, JMA are now fully qualified to submit their products to the GPPA for acceptance as entering Operational mode.

D.GIR.20200319.1: Other agencies only need to assess the uncertainties due to spatial and temporal variability, when submitting products derived from IASI to the GPPA – with the addition of gap-filling uncertainty for CrIS. ??

A.GIR.20200319.3: Manik Bali (NOAA) to work with GRWG to define a template for a User Guide for GEO-LEO IR GSICS products - to include guidance on best before dates.

The need to use multiple references in GEO-LEO IR inter-calibration was also discussed, pointing out that NOAA already generate double-difference plots, and will put them on the web.

R.GIR.20200319.2: All GEO operators to include summaries of the double-difference between their GEO-LEO IR products generated from different reference instruments in their Annual Calibration Reports.

A.GIR.20200319.4: Fangfang Yu (NOAA) to include summaries of the double-difference between their GEO-LEO IR products generated from different reference instruments in the Annual Calibration Report template.

Fred Wu (NOAA) - GSICS correction for ABI IR bands

Fred reviewed several shortcomings in the current GSICS GEO-LEO IR algorithm a series of potential improvements, which centred on redefining form of the correction from a weighted linear regression of the collocations to a weighted ratio. This would require the development of a statistical test to check the hypothesis that the slope/intercept are not significantly different from 1/0.

It was pointed out that SRF errors can introduce statistically significant offset terms in the current regression – but that these may be better handled using nonlinear terms to ensure the correction remain zero for deep space views.

When asked how this would be communicated to the users, Fred explained that it would require a simple reformulation of the GSICS Correction – in most cases to make it much simpler that now. This could be supplementing by providing expert users with access to the collocations as intermediate products.

A more general question is whether we can define a process to analyse the collocations to identify optimal the form of the GSICS Correction for each instrument/channel, based on a series of statistical tests.

A.GIR.20200319.5: Fred Wu (NOAA) to investigate statistical significance tests and set up a web meeting to discuss.

Ninghai Sun/Banghua Yan (NOAA) - Introduction to STAR ICVS GSCIS Portal

NOAA’s Integrated Calibration/Validation System (ICVS) aims to provide a centralized platform to integrate STAR GEO-LEO inter-sensor comparison results for broader users, provide an interface for NOAA GEO and LEO sensor performance and stability trending products, contribute to the calibration and trending of data products of global weather/environmental satellite constellation and support NOAA operational weather satellite sensor cal/val activities.

Initially ICVS included LEO-LEO inter-comparison using SNOs, but has now been extended to include NOAA’s GEO-LEO IR work. Ninghai explained ICVS currently uses static images, which are pre-generated, but a dynamic website is currently under development.

The need for an ICVS User Guide was discussed. While it was agreed that the interface itself is well designed and intuitive, it is important that the underlying algorithms are documented. The ATBD type document including detailed information on how the data are collocated, filtered, and spectral matched is desired to put online for public.

Minju Gu (KMA) - KMA GEO-LEO IR product status

Minju provided a detailed analysis of the GEO-LEO IR results for AMI. In general, the results were excellent – with low bias in all channels for standard radiance scenes and small diurnal variations. She highlighted several interesting features. In particular:

• The IR3.8 channel shows significant biases for cold scenes wrt IASI (despite being completely covered by the IASI spectral coverage), but not CrIS. This is the topic of an existing action (A.GIR.2019.7i.2) for collaborative investigation.
• The Australian desert introduces strongly non-linear responses at high Tb range of window channels – although this can be overcome by rejecting all collocations south of 10°S, this compromises the full dynamic range coverage.
• IR13.3 channel showed a radiance-dependent bias, which was minimised by shifting the Spectral Response Function by 0.8cm-1.
  • It is important that the SRFs are checked during the commissioning period and any changes clearly communicated to user community

KMA are preparing to submit their GEO-LEO IR products for AMI as Demonstration GSICS products.

R.GIR.20200319.3: Minju Gu (KMA) to consider rejecting only land surfaces with highly variable topography or introducing Ray-matching to reduce azimuth effects.

A.GIR.20200319.6: Dohy Kim (KMA) to ensure revised SRFs for AMI IR13.3 is published on KMA website and inform RTM and NWP communities of the change. The revised SRFs can be found here.

Tim Hewison (EUMETSAT) - EUMETSAT GEO-LEO IR product status and plan for ISSCP-NG

Tim reviewed the status of the EUMETSAT GEO-LEO IR products. Both Near Real-Time and R-Analysis Correction are currently labelled as “Operational” for all combinations of SEVIRI on Meteosat-8, -9, -10 and –11 with IASI on Metop-A, -B and –C. However, those products generated with IASI-C have not formally been through the GPPA. Manik explained that EUMETSAT only need submit a statement that IASI-C is in the same family as the existing reference instrument, supported with evidence of the double difference statistics to the GPPA.

Tim also explained that EUMETSAT are continuing to generate GEO-LEO IR products from their prototype processor running in a development environment, which are labelled as “Demonstration” GSICS products. These provide coverage over an extended period, so will be kept in the GSICS Product Catalog.

Tim asked whether addition variables are needed to explain the applicability period of the GSICS products (in addition to the current time_coverage_start/end variables in the GSICS netCDF convention, such as a “Best Before Date”. It was agreed that this was best expressed in the products’ User Guide, and was added to Action A.GIR.20200319.3

A.GIR.20200319.7: Manik Bali (NOAA) to review guidance in the definition of time_coverage_start/end variables in the GSICS netCDF convention,

If GSICS is to be considered an integrated component in ISCCP-NG, there is a need to generate GSICS products for all channels of all future geostationary imagers - even if only monitoring to ensure calibration is within required accuracy and estimate their calibration uncertainty. But ultimately, GSICS Corrections may need to be developed where these requirements are not met.

Many of these had already been addressed by previous presentations, including the need for:

• gap-filling for CrIS ’ spectral coverage and in the short-wave end of IASI
• improved performance of GSICS Corrections for cold scenes, typical of cloud tops
• Improved coverage of diurnal cycles – initially to validate the GEO imagers’ temporal stability, but maybe in future refinements of the GSICS Correction.

Dorothee Coppens (EUMETSAT) - Migration / Update on IASI-A/B/C

Dorothée provided an update on the status of IASI-A, -B and –C, emphasising that IASI-B is now considered as Prime and should now be used by all GPRCs as reference instrument. She showed results of three different inter-comparison methods: direct comparisons using big averages, double differences using SEVIRI and Double differences using models (OBS-CALC). These all showed good agreements, with reduced differences wrt IASI-A since the update to its onboard processing on 30 Sept 2019. However, there were indications that the drift in the orbit of Metop-A (currently at 8:48pm/am equator crossing time) is starting to cause increasing differences in the short-wave end – which may be due to occasional solar illumination.

The orbital phase of Metop-C has now been moved into Trident formation, opposite Metop-B, which Metop-A is now continuing to drift until its planned End of Life in 2021 Q3/Q4.

A.GIR.20200319.8: Dorothee Coppens (EUMETSAT) to send notifications of IASI status updates impacting its calibration to the GSICS developers’ mailing list: gsics-dev@googlegroups.com.

Hui Xu (UMD) - Updates on IASI SW Gap Filling

Hui described the application of the PC-based gap-filling method he developed with Likun Wang to extend the spectral coverage of IASI from 2760cm^-1 to 3275cm^-1. This was applied to allow the inter-comparison of AVHRR Ch3b with IASI and resulted in there being no obvious scene or angle dependent after the gap was filled – even over cloudy scenes. However, there remains relatively large uncertainty in the very cold scenes - although it has been greatly reduced. Currently, the gap filling coefficients can only be applied to IASI spectra without negative values in the measured channels – although it was pointed out that it can be important to include these for statistical consistency – to avoid introducing biases in very cold scenes.

It was recommended to attempt to increase the spectral resolution. The validation of model-simulated spectra with real IASI spectra is also needed to ensure the validity of the training dataset.

A.GIR.20200319.9: Hui Xu (UMD) to evaluate uncertainties introduced by gap-filling method by applying it to an independent validation dataset – and by applying it to reconstruct a spectrally-truncated set of real IASI observations.

Chengli Qi (CMA) - HIRAS status update and IR recalibration for FY-3/IRAS and VIRR

• Postponed to next web meeting

Outcome

Overall the meeting are very productive. There are a total of 31 people attended. During the meeting, the participants were engaged and active so that the meeting got a productive output. A series of recommendations and actions were produced (listed above).

Finally, the team senses the urgency to further improve the existing GEO-LEO inter-calibration algorithms to meet new challenges (e.g., collocation, correction, gap-fillings, etc …) and requirements from ISSCP-NG and GEO-Ring inter-calibration. Another web meeting will be planned soon to discuss optimizing inter-calibration algorithm (see the preliminary slides here).

• SRF.zip: GEO-KOMPSAT-2A new spectral response function