And where did you decipher analyst falls? I’ve always been curious about these distinctions as well!

Looks great, glad this was useful!

I’ll take a look for sure, I wish you the best of luck with this!

Just to clarify I think OPENDAP is served by NetCDF which is why you probably get the 2D lat/lon vs. GRIB. Your project will make you a GRIB expert in no time though!

Are you getting GRIB from HTTPS? I would guess (but am not an OPENDAP expert) that it is being served by NetCDF on the backend which are probably nicer than the equivalent GRIB version. But I have the same questions you do ultimately!

Nice! So RTMA for HI also had this odd reverse row ordering quirk?

Hmm it still seems like something isn’t right with the data. Even after smoothing see how it looks like a mirror image down the center meridian of the image? The smoothing gets rid of the streaks but it still seems like every other row of data is backwards. Maybe you’re still working on it?

So happy to see this was useful! I hadn’t heard anything on the discussion but figured you would get back to it at some point. Nice work!

I think there are THREDDS data servers at NOMADS, NCEI which I think siphon would plug into but not sure about EC or ECMWF. Also check out “boto” for getting data from AWS buckets and “gustil” for downloading data from the Google cloud. That prob still doesn’t help with int’l data tho. Good luck!

www.unidata.ucar.edu/software/ldm/

Are you familiar with Siphon and THREDDS? www.unidata.ucar.edu/software/sip...

Follow up: It was an awesome talk!! Great job @nycoloradan.bsky.social and Julie!

Last link should actually be directly here: unidata.github.io/MetPy/latest...

Manually adding CRS: unidata.github.io/MetPy/latest...

Hawaii grid definition and projection attributes: graphical.weather.gov/docs/ndfdSRS...

Perhaps. Maybe try removing parse_cf to see if cfgrib returns something with x and y dims? Then you can isolate it to assigning the CRS, rather than an issue with xarray or cfgrib. You can manually assign the CRS if it comes to it…I’ll send two links

Are you using parse_cf from metpy when reading? Or are you assigning the CRS some other way (I see metpy_crs in your dataset). Seems like maybe parsing the projection info maybe is going awry. Based on Google this should be Mercator data with 321x225 so hopefully that metadata exists somewhere?