Get point from HRRR grid nearest a latitude/longitude location

[blaylockbk]

One question I get asked often (and still don't know the best answer to), is "How do I get the grid point nearest a specific latitude/longitude location?"

As far as I understand, there still isn't a method to select a point in a grid nearest a latitude/longitude from an xarray.Dataset with curvilinear grid (like HRRR, which is in a Lambert Conformal projection).

I have devised a straightforward (at least to me) method to pluck out a point from a grid in a few lines of python described in this notebook.

This method is used in my function pluck_points for an xarray.Dataset that has coordinates 'latitude' and 'longitude'.

There is some additional discussion of this topic at StackOverflow. One alternative method is to transform the lat/lon grid and point to the projection coordinates and do a ds.sel on that (idea from Karl Schneider).

There is work in progress on the xarray front to make it possible to select nearest lat/lon points along multidimensional indecies. Check out the discussion in xarray's "Explicit Index" project: https://github.com/pydata/xarray/projects/1

[blaylockbk]

Here is a short about this same issue, and another link to the github issue that will make this type of indexing possible.
https://twitter.com/Cyclogenesis_au/status/1369656440424701958?s=20

Hey @xarray_dev gurus. I have a dataset with dims of t,z,y,x and with extra coordinates of lat and lon (see image) slicing with .sel(x=something) is easy but .sel(lon=something) is not happy.. How to you tell xarray that lat/lon are indexable dimensions? pic.twitter.com/8fQve1o534

— Scott Collis (He/Him) (@Cyclogenesis_au) March 10, 2021

This will hopefully become much easier with the big upcoming flexible indexes refactor (https://t.co/bvNnpgcPlJ), but for now, you'll probably want to use swap_dims (https://t.co/jalAC6F2Ci).

— Jon Thielen (@wx0jt) March 10, 2021

<script async