Merging Datasets

[Ixiodor]

I loaded N files(single bands) in a Dataset:

collected_ds = []
datasetname = xarray.Dataset()
datasetname['aname'] = rioxarray.open_rasterio(...)).squeeze() # Doing squeeze to resolve error about non 2D/3D array on to_raster
...
collected_ds.append(datasetname)
...
collected_ds.append(datasetname1)

and i did a clip on each dataset.
Now i collected N datasets that need to be geospatially merged for a mosaic:
mosaic_test = rioxarray.merge.merge_arrays(collected_ds, precision=50, nodata=0)
but that throws:

rioxarray.exceptions.InvalidDimensionOrder: Invalid dimension order. Expected order: ('y', 'x')You can use DataArray.transpose('y', 'x') to reorder your dimensions.

Why happens?

[snowman2]

Do you have to convert it to a dataset? If so, I recommend using:

rioxarray.merge.merge_datasets(collected_ds, precision=50, nodata=0)

EDIT: Fix in #245

Workaround: wait to do squeeze until after merging.

[snowman2]

What happens if you don't do squeeze()?

[Ixiodor]

Seems to get merged but to_raster() throws:

../earthalytics_core/earthalytics_engine/Sentinel2l2a.py:284: in generate_images
mosaic_10m.rio.to_raster(m10_filepath)
/home/ubuntu/.cache/pypoetry/virtualenvs/earthalytics-core-YgAKQ8v--py3.8/lib/python3.8/site-packages/rioxarray/raster_dataset.py:410: in to_raster
return data_array.rio.to_raster(
/home/ubuntu/.cache/pypoetry/virtualenvs/earthalytics-core-YgAKQ8v--py3.8/lib/python3.8/site-packages/rioxarray/raster_array.py:915: in to_raster
count=int(self.count),
/home/ubuntu/.cache/pypoetry/virtualenvs/earthalytics-core-YgAKQ8v--py3.8/lib/python3.8/site-packages/rioxarray/rioxarray.py:684: in count
extra_dim = self._check_dimensions()

---

self = <rioxarray.raster_array.RasterArray object at 0x7efbcf155520>

def _check_dimensions(self):
    """
    This function validates that the dimensions 2D/3D and
    they are are in the proper order.

    Returns
    -------
    str or None: Name extra dimension.
    """
    extra_dims = list(set(list(self._obj.dims)) - set([self.x_dim, self.y_dim]))
    if len(extra_dims) > 1:

      raise TooManyDimensions(

            "Only 2D and 3D data arrays supported."
            f"{_get_data_var_message(self._obj)}"
        )

E rioxarray.exceptions.TooManyDimensions: Only 2D and 3D data arrays supported.

repr:

<xarray.Dataset>
Dimensions: (band: 1, x: 249, y: 148)
Coordinates:

• y (y) float64 4.526e+06 4.526e+06 ... 4.525e+06 4.525e+06
• x (x) float64 4.263e+05 4.264e+05 ... 4.288e+05 4.288e+05
• band (band) int64 1
  spatial_ref int64 0
  Data variables:
  blue (band, y, x) float64 dask.array<chunksize=(1, 148, 249), meta=np.ndarray>
  green (band, y, x) float64 dask.array<chunksize=(1, 148, 249), meta=np.ndarray>
  red (band, y, x) float64 dask.array<chunksize=(1, 148, 249), meta=np.ndarray>
  nir (band, y, x) float64 dask.array<chunksize=(1, 148, 249), meta=np.ndarray>

[snowman2]

I think the fix here is to wait to do squeeze() until right before to_raster().

[snowman2]

Fix in #245

[Ixiodor]

Worked, Thanks