selecting an xarray dataset by coordinates instead of dimensions (or switching coordinates to dimensions)

Question

I have a, xarray dataset that is ocean color of the Atlantic Ocean called olci_ds. This dataset has dimensions of rows and columns and coordinates of lat, lon as in:

I want to grab a transect across this dataset going from a certain start lat and lon to a certain end lat and lon. For example something like:

lon_list = np.arange(-77.2, -75.5, 0.002)
lat_list = np.arange(34.48, 33.25, -0.002)

My primary challenge is that because this dataset has dimensions of (row,column) instead of (lat,lon). I think ideally I could just make these coordinates dimensions then my problem would be solved but I'm not sure how to do that. I have this working, but it is through a very slow process where for each point in the transect line I find the row and column of the lat and lon I'm interested in and then find the data at that row and column by this function:

def grab_transect_data(lon_list, lat_list, olci_ds):
    variable_list = []
    xloc_list = []
    yloc_list = []

    ratio = len(lat_list)/len(lon_list)

    for idx in range(len(lon_list)):
    # for idx in range(300):

        lat = y_list[math.floor(idx*ratio)]
        lon = lon_list[idx]

        # First, find the index of the grid point nearest a specific lat/lon.   
        abslat = np.abs(olci_ds.lat-lat)
        abslon = np.abs(olci_ds.lon-lon)
        c = np.maximum(abslon, abslat)

        try:
            ([yloc], [xloc]) = np.where(c == np.min(c))

        except ValueError:
            # sometimes there are two equally near and I just grab the first one
            yloc = np.where(c == np.min(c))[0][0]
            xloc = np.where(c == np.min(c))[1][0]

        xloc_list.append(xloc)
        yloc_list.append(yloc)


        if idx % 50 == 0:
            print(idx)

        # Now I can use that index location to get the values at the x/y diminsion
        point_ds = olci_ds.sel(columns=xloc, rows=yloc)
        variable_list.append(point_ds['variable I want'].values)

    return(xloc_list, yloc_list, variable_list)

I have to imagine it would be much simpler just to change these coordinates to dimensions then I can use a normal .sel() something like:

olci_ds.sel(latitude=lat_list, longitude=lon_list, method="nearest")

But I haven't been able to successfully make this conversion. Any help would be much appreciated!

Answer 1

I believe this should answer your question: https://gis.stackexchange.com/questions/353698/how-to-clip-an-xarray-to-a-smaller-extent-given-the-lat-lon-coordinates Basically you can create a mask based on the (non-dimension) coordinates and then use xarray.Dataset.where() with drop=True in order to create the subset

If the rows and columns correspond to latitude and longitude, it may be possible to use xarray.Dataset.swap_dims() https://xarray.pydata.org/en/stable/generated/xarray.Dataset.swap_dims.html but from what you show here, it looks like the longitude (and latitude) depends on both the row and column, so that will not work