While exporting the rasters using rioxarray, nodata value is undefined after the mask

Question

I have used xarray to open the netcdf file, and using rioxarray exported the raster as a .tif file. However, the exported raster doesn't have definition for nodata value, and hence, the GIS software such as ArcGIS, or QGIS cannot recognize the NoData. I could reclassify it in arcGIS afterwards, but later I need to implement this on large number of datasets, so automation is preferred. How can we define nodata while exporting raster?

Here is my code sample:

  ds_mask = xr.open_dataset(path_mask)
  nodata_value = np.nan
  ds_mask = xr.where(ds_mask == 0, nodata_value, ds_mask)
  ds_mask = coordAssign(ds_mask)

  mask_filename = f"/content/LANDMASK/landmask1.tif"
  ds_mask.rio.to_raster(mask_filename, nodata=np.nan)

Also, here is my implementation function to assign the coordinates:

def coordAssign(xrData):
    # convert longitude from (0-360) to (-180 to 180) (if required)
    xrData['lon'] = (xrData['lon'] + 180) % 360 - 180
    xrData = xrData.sortby(xrData.lon)
    xrData = xrData.rio.set_spatial_dims('lon', 'lat')
    # print(data.rio.crs)
    xrData.rio.set_crs("epsg:4326")
    return xrData

Find the image for your reference, while clicking on masked region, the ArcGIS should recognize it as NoData but at this time it's set to nan.

Answer 1

Documentation reference:

• nodata management
• managing data loss

Summary: You can lose metadata after performing xarray operations and you have to be careful to manage it.

Methods to use before/after xarray operations:

• rio.nodata
• rio.write_nodata

Example:

nodata = raster.rio.nodata

raster = raster.where(raster != nodata)

raster.rio.write_nodata(nodata, encoded=True, inplace=True)