Add cloud_type_chmi

[nedelceo]

• Closes #xxxx
• Tests added
• Fully documented
• Add your name to AUTHORS.md if not there already

Hi, my colleague Martin Setvák recently modified enhancements of FCI Cloud type RGB so it is not oversaturated and it better detects cirrus clouds. @ameraner suggested to create PR for it. This PR is adding cloud_type_chmi.

However I am not sure how to properly call the new product, so I renamed original CIMSS to CHMI. The bigger issue might be that in composites.yml there are now two same composites that differ only in name and standard_name. But in my opinion this is the only way of keeping both products.

Martin's description:
The difference is namely in the range for NIR1.3 band - we use broader range not to over-saturate the brightest parts of clouds, to maintain their structure, morphology, and also partial transparency (to keep the lower clouds underneath somewhat better visible). To compensate for the brightness loss resulting from the broader range, we use stronger gamma. I'm aware that our product settings results also in somewhat increased noise in darkest parts, but I prefer this negative impact, rather than loosing some of the thinner-most cirrus clouds.

Image examples:
CIMSS Cloud type

CHMI Cloud type

[gerritholl]

Thanks, this looks good!

About the name: according to #3269, the existing Cloud Type RGB should anyway be renamed day_cloud_type. So I would name your RGB day_cloud_type_chmi.

Have you tried the Li and Shibata correction as an alternative to sunz_corrected + sunz_reduced? Li and Shibata is also a WMO recommendation from the RGB workshop. It is available in satpy under modifiers.geometry.EffectiveSolarPathlengthCorrector.

[nedelceo]

Thanks Gerrit for your suggestions to follow WMO guidliness. I have changed the names, and replaced sun corrections as you suggested. Based on my very simple test the change improved the plotting of the transition zone.

[ameraner]

Indeed, it looks very nice! Much less flashy saturation all over. Are these settings good for the full-disk or rather tuned for mid latitudes?
cc @strandgren

[nedelceo]

Martin tuned it based on Europe area. However, I would expect much less (if any), sensitivity to tropospheric height compared to the IR channels. And it seem so be working properly for the full-disk. Example here (on CHMI cloud): https://ecloud.chmi.cz/s/4RSNoDigpz7a6DJ?dir=undefined&openfile=821964

[strandgren]

Looks really good indeed, thanks for sharing the recipe. Although I like the intense colours of the CIMSS version, this recipe definitely seems to be more useful showing more cloud structures🙂 Thanks for changing the solar correction modifiers as well, with that we clearly get more data at the terminator, without any obvious artefacts.