Make sure a discrete data type is preserved through resampling

satpy/resample.py

@@ -851,8 +851,15 @@ def resample(source_area, data, destination_area,
     return res
 
 
+def get_fill_value(dataset):
+    """Get the fill value of the *dataset*, defaulting to np.nan."""
+    if np.issubdtype(dataset.dtype, np.integer):
+        return dataset.attrs.get('_FillValue', np.nan)
+    return np.nan
+
+
 def resample_dataset(dataset, destination_area, **kwargs):
-    """Resample the current projectable and return the resampled one.
+    """Resample *dataset* and return the resampled version.
 
     Args:
         dataset (xarray.DataArray): Data to be resampled.
@@ -862,7 +869,8 @@ def resample_dataset(dataset, destination_area, **kwargs):
         **kwargs: The extra parameters to pass to the resampler objects.
 
     Returns:
-        A resampled DataArray with updated ``.attrs["area"]`` field.
+        A resampled DataArray with updated ``.attrs["area"]`` field. The dtype
+        of the array is preserved.
 
     """
     # call the projection stuff here
@@ -874,7 +882,8 @@ def resample_dataset(dataset, destination_area, **kwargs):
 
         return dataset
 
-    new_data = resample(source_area, dataset, destination_area, **kwargs)
+    fill_value = kwargs.pop('fill_value', get_fill_value(dataset))
+    new_data = resample(source_area, dataset, destination_area, fill_value=fill_value, **kwargs)
     new_attrs = new_data.attrs
     new_data.attrs = dataset.attrs.copy()
     new_data.attrs.update(new_attrs)

satpy/tests/test_resample.py

@@ -33,6 +33,29 @@
     import mock
 
 
+class TestHLResample(unittest.TestCase):
+    """Test the higher level resampling functions."""
+
+    def test_type_preserve(self):
+        """Check that the type of resampled datasets is preserved."""
+        from satpy.resample import resample_dataset
+        import xarray as xr
+        import dask.array as da
+        import numpy as np
+        from pyresample.geometry import SwathDefinition
+        source_area = SwathDefinition(xr.DataArray(da.arange(4, chunks=5).reshape((2, 2)), dims=['y', 'x']),
+                                      xr.DataArray(da.arange(4, chunks=5).reshape((2, 2)), dims=['y', 'x']))
+        dest_area = SwathDefinition(xr.DataArray(da.arange(4, chunks=5).reshape((2, 2)) + .0001, dims=['y', 'x']),
+                                    xr.DataArray(da.arange(4, chunks=5).reshape((2, 2)) + .0001, dims=['y', 'x']))
+        expected = np.array([[1, 2], [3, 255]])
+        data = xr.DataArray(da.from_array(expected, chunks=5), dims=['y', 'x'])
+        data.attrs['_FillValue'] = 255
+        data.attrs['area'] = source_area
+        res = resample_dataset(data, dest_area)
+        self.assertEqual(res.dtype, data.dtype)
+        self.assertTrue(np.all(res.values == expected))
+
+
 class TestKDTreeResampler(unittest.TestCase):
     """Test the kd-tree resampler."""
 
@@ -369,6 +392,7 @@ def suite():
     mysuite.addTest(loader.loadTestsFromTestCase(TestNativeResampler))
     mysuite.addTest(loader.loadTestsFromTestCase(TestKDTreeResampler))
     mysuite.addTest(loader.loadTestsFromTestCase(TestEWAResampler))
+    mysuite.addTest(loader.loadTestsFromTestCase(TestHLResample))
 
     return mysuite
 

setup.py

@@ -31,7 +31,7 @@
 
 from setuptools import find_packages, setup
 
-requires = ['numpy >=1.12', 'pillow', 'pyresample >=1.10.0', 'trollsift',
+requires = ['numpy >=1.12', 'pillow', 'pyresample >=1.10.3', 'trollsift',
             'trollimage >=1.5.1', 'pykdtree', 'six', 'pyyaml', 'xarray >=0.10.1',
             'dask[array] >=0.17.1']