search

NOAA-EMC / NCEPLIBS-bufr

The NCEPLIBS-bufr library contains routines and utilites for working with the WMO BUFR format.
Other
40 stars 19 forks source link

Missing GOES SATWND in prepbufr #597

Closed Nuo-Chen closed 1 month ago

Nuo-Chen commented 1 month ago

I was trying to create synthetic observation using NCEP PREPBUFR files as the template. Looking at the prepbufr data on July 15 and August 15 from year 2020 to year 2023 as sample data, I noticed the data void in satellite wind field in all these files over North and South Americas (possibly missing GOES-east/west). Satellite wind from August 2016 is complete.

image

The data I used was from NCAR's RDA archive ds337.0 and they are downloaded from NCEP NOMADS server according to the dataset manager at NCAR. The NCAR staff also found the satellite wind is missing from the SATWND BUFR file (RDA ds351.0).

jbathegit commented 1 month ago

I don't think this is a library issue, but in any case I'm bringing @ilianagenkova in on the discussion since she heads our observational processing team which generates the satwnd and prepbufr files.

ilianagenkova commented 1 month ago

@Nuo-Chen, some years ago a decision was made to stop adding new data to SATWND in the prepbufr files because the volume of global coverage winds continues to increase while the prepbufr file was meant to be a smaller size file. Please use the satwndbufr_d files on NOMADS for the full satellite winds data set.

When you say "The NCAR staff also found the satellite wind is missing from the SATWND BUFR file (RDA ds351.0)" I don't know what RDA ds351.0 means. In our archive the GOES winds are present for 2023 July and August. If you'd like the data for just the dates you mentioned I could share these with you:

[clogin09 /lfs/h2/emc/obsproc/noscrub/iliana.genkova/TEMP_Nuo-Chen]$ sinv GDA/gfs.20230815/00/atmos/gfs.t00z.satwnd.tm00.bufr_d 003 METOP-1 (Metop-B 572 591
056 METEOSAT 9 121735 207
057 METEOSAT 10 153698 207
174 Himawari-9 132314 000
209 NOAA 18 226 000
223 NOAA 19 443 000
224 NPP 12873 616
225 NOAA 20 12015 616
270 GOES 16 876099 617
272 GOES 18 2039878 617
471 INSAT 3D 9540 000
473 INSAT 3DR 57724 000
783 TERRA 122 000
784 AQUA 4350 000
854 Non-specific mix 28102 000
3449691

[clogin09 /lfs/h2/emc/obsproc/noscrub/iliana.genkova/TEMP_Nuo-Chen]$ sinv GDA/gfs.20230715/00/atmos/gfs.t00z.satwnd.tm00.bufr_d 003 METOP-1 (Metop-B 8112 591
005 METOP-3 (Metop-C 3003 591
056 METEOSAT 9 141987 207
057 METEOSAT 10 142865 207
174 Himawari-9 140733 000
206 NOAA 15 312 000
223 NOAA 19 387 000
224 NPP 26325 616
225 NOAA 20 19668 616
270 GOES 16 874105 617
272 GOES 18 1886985 617
473 INSAT 3DR 73170 000
854 Non-specific mix 30663 000
3348315

tcram commented 1 month ago

Hi, I manage the SATWND bufr_d archive at NCAR that @Nuo-Chen is referring to.

Our BUFR decoder retrieves latitude and longitude from the Table B mnemonics CLAT and CLON, which as it turns out contain missing values for most satellite platforms. I've updated our decoder to check for missing CLAT and CLON, and, if missing, use CLATH and CLONH instead.

This issue can be closed now.

Tom

ilianagenkova commented 1 month ago

Tom, good move, as some providers (JMA, i.e. NC00504*) still use CLAT and CLON.