Future ABI L1b RGBs and priorities

[djhoese]

These are the list of RGBs that are NOAA/NWS priorties and appear in the AWIPS menus as something forecasters can see:

Product Ingredients (ABI bands)

• Day Cloud Phase Distinction ABI02, ABI05, ABI13
• Nighttime Microphysics ABI07, ABI13, ABI15
• CIRA Geocolor ABI01, ABI02, ABI03, ABI07, ABI13
• Air Mass RGB ABI08, ABI10, ABI12, ABI13
• Day Cloud Convection ABI02, ABI13
• VIS/IR Sandwich ABI02, ABI13
• Fire Temperature ABI05, ABI06, ABI07
• Day Snow-Fog ABI03, ABI05, ABI07, ABI13
• Day Convection ABI02, ABI05, ABI07, ABI08, ABI10, ABI13
• Day Land Cloud ABI02, ABI03, ABI05
• CIMSS Natural Color ABI01, ABI02, ABI03
• Differential Water Vapor ABI08, ABI10
• Ash ABI11, ABI13, ABI14, ABI15
• Dust ABI11, ABI13, ABI14, ABI15
• SO2 ABI09, ABI10, ABI11, ABI13
• Blowing Snow ABI02, ABI05, ABI07, ABI13
• Simple Water Vapor ABI08, ABI10, ABI13
• Day Cloud Type ABI02, ABI04, ABI05
• Day Ocean Cloud Convection ABI03, ABI13

A check in the above list means it is available in Geo2Grid and is documented in the reader as being available.

It was noted that the first 7 or so (up to Fire Temperature) are the highest priority/useful. We'll need to consider which exact recipe to use as Satpy may include some of these but with different per-band scaling limits or for more complex RGBs like geocolor different correction algorithms/LUTs may be used (ex. rayleigh correction by CIRA's code versus Satpy's sibling project pyspectral).

NOTES

1. "CIMSS Natural Color" I believe is the same as a no-rayleigh version of our "true color", possibly no / cos(SZA) term either. So I think if we're taking our opinions into account we'll just keep our true_color. We also have a "natural_color" but I believe that follows the EUMETSAT traditional recipe (not very "natural").
2. Satpy has a geocolor implementation but it likely uses different algorithms or approaches to the exact CIRA recipe. This work was added to Satpy mostly by EUMETSAT with permission from CIRA to get a GeoColor composite for the FCI instrument.
3. We have something called "fog" which if I'm reading the configs correctly is a (12.0 - 10.8, 10.8 - 8.7, 10.8) RGB. I'm not sure where this came from.
4. day cloud phase distinction in satpy
5. geo color in satpy
6. Possible day cloud convection called "cira_day_convection" in satpy This is apparently the same as "convection" mention in point 9 below and neither are "day cloud convection" which I can't find a record of.
7. fire temperature awips in satpy
8. day snow fog in satpy
9. day convection as "convection" in satpy
10. day land cloud in satpy
11. differential water vapor as "water_vapors2" in satpy
12. so2 in satpy
13. simple water vapor as "water_vapors1" in satpy
14. day cloud type possibly as "cimss_cloud_type" in satpy

CC @graemely @ScottLindstrom @kathys

[djhoese]

As mentioned above in my edit a couple minutes ago, the "cira_day_convection" I had found is not equivalent to "day cloud convection", but is instead equal to another Satpy RGB called "convection" which is equal to "Day Convection" (lower in the list).

There is a "day cloud convection" PDF that is not implemented in Satpy from what I can tell but could be added: https://rammb2.cira.colostate.edu/wp-content/uploads/2020/01/QuickGuide_DayCloudConvectionRGB_final-1.pdf

Also, "fire temperature" seems to have a "cira" and "cimss" variant in satpy, but the "cira" one is the only one that matches exactly to the scaling limits of the CIRA/NASA PDF. The difference is 0.15 Kelvin where it was clear the CIRA recipe was converted exactly from C to K as far as "273.15K" being 0 C, while all the other fire temperature recipes use "273.0" as a starting point and some have different maximum limits on the red channel. Bottom line, we should probably use the CIRA recipe.

Other notes:

• Blowing Snow PDF: https://rammb2.cira.colostate.edu/wp-content/uploads/2024/11/GOES-BlowingSnowRGB1_QuickGuide_24April2024.pdf
• Day Cloud Type PDF but only mentions gamma correction not scaling limits: http://cimss.ssec.wisc.edu/goes/OCLOFactSheetPDFs/ABIQuickGuide_Day_Cloud_Type_RGB.pdf
• Day Ocean Cloud Convection does not exist on CIRA's website: https://rammb2.cira.colostate.edu/training/visit/quick_reference/#tab17

[djhoese]

I'm running out of time today to look through all the ones I've tried out, but this "cira_fire_temperature" looks pretty good:

Day Cloud Phase Distinction (cloud_phase_distinction)

Details

Raw (no sunz/rayleigh):

Details

Fire temperature (cira_fire_temperature)

Details

Day Convection (convection)

Details

Day Land Cloud (land_cloud)

Details

Day Snow Fog (snow_fog)

Details

SO2 (so2)

Details

Simple Water Vapor (water_vapors1)

Looks like the gamma is missing from the PDF:

https://rammb2.cira.colostate.edu/wp-content/uploads/2020/01/Simple_Water_Vapor_RGB-1.pdf

Details

Differential Water Vapor (water_vapors2)

The Satpy recipe gamma doesn't seem to match the gamma used in the PDF:

https://rammb2.cira.colostate.edu/wp-content/uploads/2020/01/QuickGuide_GOESR_DifferentialWaterVaporRGB_final-1.pdf

YAML recipe

  water_vapors2_default:
    standard_name: water_vapors2
    operations:
    - name: stretch
      method: !!python/name:satpy.enhancements.stretch
      kwargs:
        stretch: crude
        min_stretch: [30, 278.15, 243.9]
        max_stretch: [-3, 213.15, 208.5]
    - name: gamma
      method: !!python/name:satpy.enhancements.gamma
      kwargs:
        gamma: [3.5, 2.5, 2.5]

Details

Blowing Snow (blowing_snow)

Details

Day Cloud Type (cloud_type)

Details

TODO: I'll update this comment with other RGBs when I get the time.

[djhoese]

@ScottLindstrom and others, what do you think about including the / cos(SZA) correction for reflectance bands? And what about rayleigh correction? I ask as I was looking at "Day Cloud Phase Distinction" in Satpy which has this recipe:

https://rammb2.cira.colostate.edu/wp-content/uploads/2020/01/QuickGuide_DayCloudPhaseDistinction_final_v2.pdf

  cloud_phase_distinction:
    compositor: !!python/name:satpy.composites.GenericCompositor
    prerequisites:
      - name: C13
      - name: C02
        modifiers: [sunz_corrected, rayleigh_corrected]
      - name: C05
        modifiers: [sunz_corrected]
    standard_name: cloud_phase_distinction

So that is the C02 has the SZA correction and rayleigh and the C05 only has SZA as I think rayleigh correction is negligible at that wavelength if I'm not mistaken.

[kathys]

@djhoese Graeme and I spoke about the CIRA geocolor product, and we do not want to support it at this time.

[djhoese]

Just so everyone in this issue is aware, there was an RGB workshop last week in Sweden with attendees from EUMETSAT, SMHI (Sweden), WMO, CIRA, and others. They worked on standardizing RGB recipes, talked about path length corrections (pytroll/satpy#3096), and other related tasks. The plan is for a summary of the meeting to come out soon which will guide Satpy's updates to its builtin recipes. Eventually a standards document (or documents) should be published by WMO.

One such name change seems to be cloud_phase_distinction to cloud_type_distinction. There was also a request to WMO to standardize instrument channel names in addition to what they do already with satellite and instrument names.

[djhoese]

For G2G 1.3, we're going to focus on a limited subset and will try to only support:

• Day Cloud Type (similar to cloud phase distinction which is being replaced by day cloud type apparently)
• Day Convection
• SO2
• Blowing Snow

[djhoese]

I've been spending longer than I'd like to admit deciphering the RGB workshop's recommendations for the above RGB's that I/we want to add to G2G for ABI. Here is what I've got so far:

1. Day Cloud Type: We can add this as-is (day_cloud_type). There is some confusing language around what is happening to day_cloud_phase_distinction but day_cloud_type is a thing so we will use it. If we want to add day_cloud_phase_distinction in the future we'll need to define (probably in Satpy) day_cloud_type_distinction which seems to be the preferred name now. The difference is what bands are used.
2. Day Convection: Called convection in existing Satpy. The preferred name seems to be day_severe_storms so I think we should use that.
3. SO2: Renamed "Volcanic Emissions (SO2)". Satpy needs updates to call this volcanic_emissions_so2. I'm asking some Pytroll folks to contact WMO as it seems "Volcanic Emissions (SO2 and Ash)" is the name of the recipe.
4. Blowing Snow: Should be renamed to it's preferred name "day_blowing_snow".

So:

• day_cloud_type
• day_severe_storms
• volcanic_emissions_so2 (possibly volcanic_emissions_so2_ash)
• day_blowing_snow

[djhoese]

This list was discussed in today's GeoSphere meeting as the added set of RGBs into Geo2Grid directly affects what is possible in GeoSphere. The main concern from @graemely was that we weren't adding any/many RGBs from the top 7 of the original list which were identified as "high priority" or most useful. Here is a summary of my response:

1. 3 of the 7 are controversial or at least not straight forward:

• GeoColor: complex and requires large static background image. There is a long term hope that a different night time image could be found as an alternative.
• Vis/IR sandwich: what I know of this composite is that many people have different opinions about how it should be made and how it should look. Seemed best to skip it for now.
• Fire Temperature: My understanding is that there are 2 or 3 different versions of this RGB. The original was presented to NWS to be compatible with AWIPS which only has access to the CMIP calibrated version of the bands. However, the preferred recipe would involve using the channel's radiances instead of Refl/BT calibration of the channel. I don't recall the specifics off the top of my head, but seems best to skip for now.

2. 2 of the 7 are already included in G2G: night microphysics and air mass
3. "Day Cloud Phase Distinction" is a recipe originally developed for AHI. This recipe uses the 10.3um channel (ABI C13) for the red channel. However, this is only done because AHI doesn't have the preferred 1.3um channel (ABI C04). The recipe that uses 1.3um is called "Day Cloud Type" and is included in the list-to-add in the above comments. Additionally, the RGB Workshop recommends renaming "Day Cloud Phase Distinction" to "Day Cloud Type Distinction" to avoid confusion as there is already a "Day Cloud Phase" RGB recipe that looks much different than Day Cloud Phase Distinction.
4. "Day Cloud Convection" seems to be listed in the "legacy" section of the RGB Workshop document. There is a "Day Convection" RGB whose preferred name is now apparently "Day Severe Storms" and is in the list-to-add above so maybe that's a good consolation.

The RGB workshop meeting website is:

https://community.wmo.int/en/meetings/rgb-experts-and-developers-workshop-1-3-april-2025

And link to the document I'm referring to here.