Add baseline for GeoColor composite including FCI, AHI and ABI recipes #2557
Conversation
…rence used for the Geocolor low-level cloud layer.
…LowCloudCompositor.
… of projectables and new land-sea mask data.
…n when converting NDVI to blend factor.
…uce false alarms with early FCI data.
|
There are currently two aspects that need more work (apart from the tests..):
UPDATE:
|
djhoese
added
enhancement
strandgren
changed the title
|
And here a couple of examples with ABI: ABI Sunrise ABI Sunset |
|
I love this. |
|
There's also another option for landmask -- NASA's blue marble next-gen. https://neo.gsfc.nasa.gov/archive/bluemarble/bmng/landmask/. That huge world.watermask.86400x43200.bin.gz is in about 500m resolution. The file is binary so you need an ENVI header to read and convert it to geotiff. |
…add_GeoColor_composite Conflicts: satpy/etc/composites/abi.yaml satpy/etc/composites/fci.yaml
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #2557 +/- ##
==========================================
+ Coverage 95.32% 95.38% +0.06%
==========================================
Files 371 371
Lines 52441 52521 +80
==========================================
+ Hits 49988 50097 +109
+ Misses 2453 2424 -29
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
|
While I think about it, could you add a description of geocolor to this page of the wiki? https://github.com/pytroll/satpy/wiki/Built-in-composites |
…ions in compositor code.
… again with wrong dimensions. Add unit test to catch this.
I could reproduce your initial error which happened when processing several images. I have fixed that with the latest commit and I do not get the error anymore. Wrt. to the other error it's hard to debug since you use you're own recipes. But can you double check that your land-sea mask look ok? |
I already have produced a low-cloud before using my own landmask, which is actually from that NASA link I provided. I haven't tested your latest update and will do it tomorrow. Thanks! |
…cially cold cloud tops.
…n/satpy into feature_add_GeoColor_composite
|
So I have finished the parts that were missing now and here are a few examples of the results using this PR at its current state: ABI (GEOS-East) ABI_GeoColor_goes_east_abi_f_2km_20231212-20231213_FR-6.mp4AHI AHI_GeoColor_himawari_ahi_fes_2km_20231212-20231213_FR-6.mp4FCI FCI_GeoColor_mtg_fci_fdss_2km_20230318-20230319_FR-6.mp4FCI FDHSI and HRFI data resampled onto 1km grid Full resolution: https://sftp.eumetsat.int/public/file/byvy1paveeghqg20wkgvew/FCI_GeoColor_FDHSI-HRFI_1km.png Some changes in implementation compared to imagery previously posted include:
For FCI we will have to update the configurations for the low-level cloud layer when the L1C data and calibration coefficients are updated. However, I'd leave that for a future small PR. The issue with false alarms over arid surfaces for the low-level cloud layer is due to surface emissivity differences between the two IR channels and something that we will work on to improve in the future. |
…ards compatibility.
|
@strandgren With #2690 this works pretty well. But I have a question here: I've tested the AHI data on UTC 2023-12-15 09:50, and compared my output with CIRA's. Despite the colors, there's a significant difference on LowCloud over China mainland. My image CIRA Here's your land-water mask: Mine: |
|
And my composite config: |
remove Please note that in one of the recent commits I changes the term |
Great! |
This PR adds the baseline for the GeoColor composite blend developed by Miller et al. 2020. The GeoColor implementation in this PR follows the general approach described in Miller et al., but uses equivalent or similar Satpy functionalities, methods and composites where available. Some notable differences are:
true_colorcomposite based in the green channel simulation in Satpy, with no use of the green channel simulation approach and hybrid green correction used in Miller et al. (2020).The PR implements the missing compositors/layers needed for the GeoColor composite as well as recipes for FCI, AHI and ABI. Based on this, it should be easy to add recipes for other sensors like AMI, AGRI or even polar orbiting satellites in the future.
Below two early examples for FCI and AHI:
FCI
AHI
