Crop xarray given polygon corners

[vic1309]

Hey,

I was given some latitude and longitude pairs corresponding to the corners of a polygon:

lon_llc = 3.31265187; lat_llc = 55.95295157
lon_lrc = 6.6760554; lat_lrc = 55.95295157
lon_ulc = 3.31265187; lat_ulc = 57.8149327
lon_urc = 6.6760554; lat_urc = 57.8149327

My netcdf has latitude and longitude as 2D variables, but these (and the other variables) are defined with 1-D coords as eta_rho and xi_rho (x and y). Given the corners, I found the corresponding indexes and cropped my original file using them:

(eta_indices, xi_indices) = np.array([499, 256, 375, 603]), np.array([19, 133, 368, 261])

eta_min, eta_max = eta_indices.min(), eta_indices.max()  # 256–603
xi_min, xi_max   = xi_indices.min(), xi_indices.max()    # 19–368

cropped_ds = ds.isel(
    eta_rho=slice(eta_min, eta_max + 1),
    xi_rho=slice(xi_min, xi_max + 1)
)

However, when I plot the cropped domain, it doesn't really correspond to the polygon. See figure below.

Has anyone had this issue before? Tips?

[keewis]

can you post an image of the HTML repr of your dataset (or at least its coordinates section)? That might help with figuring out what is going on.