Radar Composite Conversion of x, y into latitude and longitude

[frontogenesis]

Hello all,

I'm trying to convert x, y coords into latitude and longitude so that I may extract dBZ values. It appears I've had success doing so -- or so I thought.

Perhaps an examination of the code would better elaborate the challenge I'm bumping into:

I get an error saying dimensions or multi-index levels ['latitude', 'longitude'] do not exist, despite directly assigning it via the assign_crs method. Printing out the dataset also shows the latitude and longitudes, and it looks OK.

I'm not exactly sure where to take this next. Pyproj is perhaps a solution where I specify a lon, lat and it converts to x, y, and then pass it to the sel method is one way. Though, I'm thinking just converting x, y to lon, lat is better for future things I may want to do.

Thanks so much for any pointers!

[dopplershift]

I think seeing the code would be needed to be sure, but the problem is that, if you're assigning lat/lons to a projected grid of data (I'm assuming your radar composite is something like MRMS, which is on a Lambert conformal projection), you end up with 2D arrays of lat/lon. Currently, xarray does not allow you to use those 2D indexes to do selection of data.

The easiest way, as you alluded to, is to take your desired lat/lon pair, use Pyproj (var.metpy.pyproj_crs can give you the proper object) to transform that pair to x,y, and then use that to select into your data.

[frontogenesis]

Yes, this was hugely helpful. Thanks, Ryan! (Not sure why my code didn't paste from the last post. Sorry about that!) Here is the solution for transforming, for those who may stumble across this thread.
reflectivity = radar_ds['Base_reflectivity_surface_layer'].squeeze()
crs = reflectivity.metpy.cartopy_crs
location = crs.transform_point(-82.60, 30.82, ccrs.PlateCarree())
x, y = location

We get the x, y and then plug that into the .sel method that xarray exposes.

[ahuang11]

May be relevant pydata/xarray#5009 too.