Loading MSWEP for area small areas might result in an empty array

[maawoo]

from sdc.load import load_product

mswep = load_product(product="mswep", vec="site03")

bounds = (24.385, -29.124, 24.392, -29.111)
mswep.sel(longitude=slice(bounds[0], bounds[2]), latitude=slice(bounds[3], bounds[1]))

This will result in an empty DataArray:

This combination of .sel and using index slicing is how I've implemented it in the load_mswep function:

sdc-tools/sdc/products/mswep.py

Lines 55 to 56 in 7d57239

	ds = ds.sel(longitude=slice(bounds[0], bounds[2]),
	latitude=slice(bounds[3], bounds[1]))

[maawoo]

Something like method='nearest' can't be used with index slicing. An alternative way to index is the following:

mswep.sel(longitude=[bounds[0], bounds[2]], 
          latitude=[bounds[3], bounds[1]], 
          method='nearest')

However, this results in duplicate latitude/longitude values, which has also been described in pydata/xarray#7042. The conclusion from this discussion is to adjust the area of interest so that it contains or is closer to a grid point (pixel center).