Vectorizing a multi-band raster

[Kamparia]

Hello Community,

Is there a feature for vectorizing a multi-band raster? the code below produces an error that I am looking to overcome.

import rioxarray
from geocube.vector import vectorize

raster = rioxarray.open_rasterio("/home/kamparia/Downloads/leeds_cir_compress_proj.tif", masked=False).squeeze()
df = vectorize(raster)
df.head()

Error message:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In [7], line 1
----> 1 df = vectorize(data)
      2 df.head()

Cell In [5], line 38, in vectorize(data_array)
     36 # nodata mask
     37 mask = None
---> 38 if np.isnan(data_array.rio.nodata):
     39     mask = ~data_array.isnull()
     40 elif data_array.rio.nodata is not None:

TypeError: ufunc 'isnan' not supported for the input types, and the inputs could not be safely coerced to any supported types according to the casting rule ''safe''

I look forward to any possible solution or alternative.

Thanks

[snowman2]

What is the value of raster.rio.nodata?

[Kamparia]

The value is None

[snowman2]

This may help: #142

[Kamparia]

Thanks. This worked just fine.

[snowman2]

For a multi-band raster, you likely want to vectorize each band separately:

vectorize(raster.sel(band=1))