reform_nan_array

HSTB.kluster.xarray_helpers.reform_nan_array(dataarray_stack, orig_idx, orig_shape, orig_coords, orig_dims)

To handle NaN values in our input arrays, we flatten and index only the valid values. Here we rebuild the original square shaped arrays we need using one of the original arrays as reference.

See stack_nan_array. Run this on the stacked output to get the original dimensions back.

Parameters

  • dataarray_stack (DataArray) – flattened array that we just interpolated

  • orig_idx (tuple) – 2 elements, one for 1st dimension indexes and one for 2nd dimension indexes, see np.where

  • orig_shape (tuple) – original shape of array before stack_nan_array

  • orig_coords (DataArray) – coordinates from array before stack_nan_array

  • orig_dims (tuple) – original dims of array before stack_nan_array

Returns

values of arr, filled to be square with NaN values, coordinates of ref_array

Return type

xr.DataArray