Investigate the Effort Required For G2G to support NOAA ABI SST files

[kathys]

The ROFFS project requested us to investigate the possibility of providing them with hourly ABI Sea Surface Temperature (SST) GeoTIFF images over their Areas of Interest. The SSEC data center receives these files from NOAA. The SST files are not produced by CSPP GEO software, but the investigation should be to see what it would take to create images using Geo2Grid.

I have downloaded a series of hourly files here: bumi:/data/users/kathys/geo/sst/abi

These are full disk files, which are the only domain that is available to us. I thought they might be similar in format to the ACSPO files, but they are not.

The SST Array is ushort SST

And it must be filtered using the array ubyte DQF when it is equal to 0.

[djhoese]

Congrats!

geo2grid.sh -r abi_l2_nc -w geotiff -p SST -f OR_ABI-L2-SSTF-M6_G19_s20251810000208_e20251810059516_c20251810104170.nc

Produces:

It performs a linear stretch from 267.317 to 309.816.

So work required:

• Add "SST" as a possible/default product in the abi_l2_nc reader code in Geo2Grid so it knows how to load "SST" without needing it to be specified with the -p flag like above.

[djhoese]

Oh! I just noticed the part of your message about DQF. The above does not have DQF filtering. Is it possible it isn't necessary?

[djhoese]

Here's the DQF from the header with the relevant attributes:

        ubyte DQF(y, x) ;
                DQF:_FillValue = 255UB ;
                DQF:valid_range = 0UB, 3UB ;
                DQF:flag_values = 0UB, 1UB, 2UB, 3UB ;
                DQF:flag_meanings = "good_quality_qf degraded_quality_qf severely_degraded_quality_qf invalid_due_to_unprocessed_qf" ;

So I'm confused what a value of 0 represents? When using ncdump I see fill value, 2s, 3s, and a small amount of 1s.

[kathys]

I am confused too. Here is what the document that I found says, and this information is also included in the global attributes section:

Note that SST is reported over all water pixels, and only data with quality flag DQF=0 should be used to select clear-sky pixels.

The image that you included above is definitely land cleared but not cloud cleared.

Here is the document: https://noaasis.noaa.gov/pdf/ps-pvr/goes18/ABI/Sea%20Surface%20Temperature/Full/GOES-18_ABI_L2_SST_Full_ReadMe.pdf

[djhoese]

Similar sentence:

However, only clear-sky pixels (i.e., marked
with the best quality flag, DQF=0), are recommended for use.

But this clearly contradicts itself. I can try hacking Satpy without making a full pull request and see how DQF==0 and DQF==1 affect the image.

No wait...I think I'm misreading or misremembering how flag_values/flag_meanings work in CF. There are 4 flag values and 4 flag meanings. The first flag value is 0, the first flag meaning is good qualify. So DQF == 0 should be good quality only.

I'll try to look at adding this check today in Satpy or at least test it out.

[djhoese]

Only keep DQF == 0. Thoughts @kathys ?

[kathys]

Yes, I think DQF == 0 should be used for filtering. That is what both the doc and global attributes advise from NOAA.

[kathys]

@djhoese Thinking about this effort, it is for a particular client and they do not need cloud clearing. So let's start with no filtering, but with an option to if needed.