GRWG IR Group Web Meeting 2020-06-24

GSICS IR Group Web Meeting on Refining Inter-Calibration algorithm and other Topics

The GSICS IR group is planning another IR web meeting to finish remaining talks from 2020 annual meeting agenda and also updates the efforts to improve inter-calibration algorithms. The meeting will be held at 7:00-10am EDT on Wednesday June 24 2020 (check your time here) . Should you have any urgent issues related to the IR group, please bring them for discussion.

Agenda

  1. Chengli Qi (CMA) - HIRAS status update and IR recalibration for FY-3/IRAS and VIRR (15 min)
  2. Alessandro Burini (EUMETSAT) - Update on SLSTR-IASI Inter-calibration (15 min)
  3. Hyeji Yang (KMA) - Inter-comparison between GK2A and Himawari-8 (15 min)
  4. Likun Wang (NOAA Affliate) - Improving GEO-GEO Inter-calibration with Parallax Correction (15 min)
  5. Tim Hewison (EUMETSAT) -Process to optimise inter-calibration algorithms (20 min)

Attendees

Guest Chair: Likun Wang
CMA: Chengli Qi, Scott (Xiuqing) Hu
EUMETSAT: Alessandro Burini, Tim Hewison, Sebastien Wagner, Viju John
ESA: Stefano Casadio
ISRO: Pradeep Thapliyal
JMA: Kazuki Kodera, Arata Okuyama, Hideaki Tanaka, Kazutaka Yamada
KMA: Hyeji Yang, Dohy Kim, Minjun Gu
NASA Langley: David Doelling, Connor Haney, Rajendra Bhatt
NOAA: Likun Wang, Fangfang Yu, Manik Bali, Fred (Xiangqian) Wu
UW: Dave Tobin, Hank Revercomb
WMO: Heikki Pohjola

Summary

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

FY-3D/HIRAS

This HIRAS now has stable operation status.
Decontaminations in December 2018 & October 2019 resulted in large gain changes, but not calibration wrt CrIS.
Feedback from NWP (UKMO): Anomalous behavior in Detector #3. Investigation showed this was seasonally and spectrally-dependent and related to Sun stray light.

Tim: Are that solar related calibration errors located at South Atlantic Anomaly region?
A: It’s a little a bit off (Fred Wu).

Fred: HIRAS has any contamination issues? What contents in contamination layer?
A: Yes. During the prelaunch test, we find the issues. It is not water or ice. We did two times’ decontamination on orbit.

Likun: CrIS and HIRAS inter-comparison results have spectral features at the midwave region? It is caused by spectral mismatch or noise?
A: At this stage, it is not clear. We will further investigate it.

Hank: When will be the next Launch for another HIRAS?
A: FY3E is now prelaunch test. Should be Launched soon.

FY-3A/B/C-IRAS Recalibration: 3 versions: v1 (Systematic Bias Correction), v2 (fine recalibration), v3 (FCDR with inter-satellite harmonisation and gridding). Validation against IASI: showed seasonal and diurnal differences – especially in sounding channels

Discussion: IRAS biases could be due to SRF errors.

Alessandro Burini (EUMETSAT) - Update on SLSTR-IASI Inter-calibration

Update on setting up MICMICS environment at EUMETSAT. This includes extended SNOs to cover bigger range of observation. So far only covers nadir view in 11/12µm window channels. But will include 3.8µm channel in future, using Hui Xu’s PCR-based gap-filling method to extend IASI coverage. Alessandro has set up a catalogue of all granules’ space/time coverage, which feeds a database. This can be cross-queried to identify overlapping granules, and provides global coverage for SLSTR-IASI.

Manik: It is software is in python? Did you the data remotely from the servers?
A: Yes. It is written in python. The remote server is only used internally in EUMETSAT.

Likun: Is finding the pixel within IASI FOV shape slow?
A: Using the pixel-in-polygon algorithm is really show. But we use the index search algorithms. It is fast.

Hyeji Yang (KMA) - Inter-comparison between GK2A and Himawari-8

The locations of these AHI and AMI imagers (140.7° and 128.2°E) provides good spatial matches. Their channels provide good spectral matches.
IR channels: Results are all good (including midnight period), except for IR3.8, as confirmed with inter-calibration against IASI and comparison with RTTOV and NWP model. There are also diurnal variations caused by orographic effects of the Australian terrain. VNIR channels: Results are very good (<1%).

Likun: Do you apply for spectral adjustment for geo-geo inter-calibration, for example NASA Langley’s tool?
A: Not yet. But will try it to see the difference. Dave Doelling comment they do have spectral adjustment tool for IR bands, which was generated based on IASI data.

Arata: Slide 4. It looks that comparison for AMI-AIRS is positive, it is different from AMI-IASI.
A: There a lot discussion on IASI negative radiance values at the short-wave region. Likun Wang will further look into it to check the results. David and Hank suggest that reducing the IASI resolution first to avoid negative values. At this point, it is too early to draw any conclusion.

Likun Wang (NOAA Affiliate) - Improving GEO-GEO Inter-calibration with Parallax Correction

Likun applied optical flow to correct for cloud parallax errors using inter-comparison of GOES-16 and -17 as an example, demonstrating the procedures for GEO-GEO collocation with .

Tim: Is the optical flow calculation expensive? It is easy to implement in operational environment.
A: The calculations is very fast. It also has C library and can be implemented for operational environment. But current version is used in Python.

Tim: It would be good to evaluate the uncertainties on the mean differences to check whether the parallax correction produces a significant improvement in bias as well as SD.

Manik: Does it introduces error when you project the images? Also, are there any geolocation errors for GOES-16?
A: It can introduce the errors during the projection. But we only use the re-projection to calculate the pixel displacement. After pairing the collocates pixel, we will go back to find the original BT or radiances values. We don’t manipulate the pixel values during the processing. For geolocation error, we assumed it is on the sub-pixel level. It is good enough for pixel displacement calculations. Fangfang commented that geolocations for GOES-16 perform well.

Tim: Can this be applied for visible channel geo-geo inter-calibration?
A: Visible channel geo-geo inter-calibration is more complicated than IR channels. I need to further look into it.

Tim Hewison (EUMETSAT) -Process to optimise inter-calibration algorithms

Tim’s presentation considered whether we could define a generic approach to optimise direct inter-calibration algorithms (SNO, GEO-LEO IR, Ray-matching). He recommended defining a simple metric to quantify the quality of the end result - uncertainty on bias estimate. There was good discussion on whether this metric should be defined in radiance (closest to the measurement of most instruments) or brightness temperature (closest to requirements).

Slide 4
Hank Revercomb (comments): We should do it in radiance domain. Otherwise, we will see large noise at the cold scenes at the shortwave regions.
Likun Wang (comments): Keep as many pixels as you can during collocation by relaxing colocation criteria. Later on, one can investigate the data to test the sensitive for different collocation variables (e.g., time differences, spatial difference etc.).

Slide 8:
Dave Tobin (UW) (comments): Using CrIS and IASI inter-calibration as an example, we should always make sure symmetric distribution of time difference (as well as other variables) to avoid potential sampling bias.

Slide 10:
Likun Wang (comments): We should later check scan angle dependence bias instead of local zenith angle. Because it is more instrument related. But we can collect data along with the zenith angle. But final results are better along with instrument scan angles.

Manik Bali: Is this going to be new inter-calibration products?
A: At this stage, we focus on LEO-LEO collocation algorithm but it may be apply for general inter-calibration algorithms. We may generate GSICS products for SLSTR, depending on its calibration and stability – TBC.

General comments after presentation:

Likun Wang (comments): Scene uniformity may be also a factor to consider.
A :Agree. General principle is to keep relatively large dataset for validation. It is possible to using Dave Tobin’s method to fit the data by assigning different weights according to the scene uniformity. This is implemented in the current GSICS GEO-LEO IR products. We agreed that as part of the process of designing the inter-calibration algorithm, a large dataset of collocations should be established to test for any significant dependences, such as scan angle, nonlinearity, diurnal, seasonal, long-term drift, etc. If any are significant dependences are found, the GSICS Correction should be (re-)formulated to account for them. If not, all the collocations may be combined.

Manik Bali: What can you share with the community?
A: We will share the IASI-SLSTR collocation software to the community.

David Doelling (comments): The best practice to keep the as much as noisy data samples in order to reserve possible dynamic range. Because for solar bands, the low-end signal is much more interesting. In addition, the bias analysis should also consider the orbital variation such as latitude-dependent bias.

Outcome

Action: A.GIR.20200624.1: Likun Wang and Tim Hewison to work on a wiki page with the inputs from the GSICS community to address the issues and best practice for generic inter-calibration algorithms.

Action: A.GIR.20200624.2 : David Doelling to give a presentation on intercalibration algorithm for visible channels during IR and Visible group Joint web meeting, which will summarize the common steps to optimizing the best inter-calibration results.

I Attachment Action Size Date Who CommentSorted ascending
2020.06.24-Chengli+Qi-GSICS+IR+Group+web+meeting-2.pptxpptx 2020.06.24-Chengli+Qi-GSICS+IR+Group+web+meeting-2.pptx manage 8 MB 24 Jun 2020 - 10:32 LikunWang  
ESL_SLSTR_IASI_Presentation.pptxpptx ESL_SLSTR_IASI_Presentation.pptx manage 7 MB 23 Jun 2020 - 15:10 TimHewison  
Likun_Wang_Geo_06242020.pptxpptx Likun_Wang_Geo_06242020.pptx manage 61 MB 23 Jun 2020 - 19:19 LikunWang  
KMA_The result of inter-comparison between <a class="foswikiNewLink" href="/bin/edit/Development/GK2A?topicparent=Development.20200624" rel="nofollow" title="Create this topic">GK2A</a> AMI and Himawari-8 AHI_20200624.pptxpptx KMA_The result of inter-comparison between GK2A AMI and Himawari-8 AHI_20200624.pptx manage 1 MB 24 Jun 2020 - 08:43 DohyKim AMI-AHI intercomparison
Process to optimise inter-calibration algorithms.pptxpptx Process to optimise inter-calibration algorithms.pptx manage 2 MB 25 Jun 2020 - 13:23 TimHewison updated post-meeting to include outcome of discussion on final slide
Topic revision: r13 - 28 Jun 2020, LikunWang
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding GSICS Wiki? Send feedback