Simplified type casting and removes unreachable lines.

improver/nbhood/nbhood.py

@@ -314,9 +314,9 @@ def _calculate_neighbourhood(
         if self.sum_only:
             area_sum = None
         else:
-            area_sum = self._do_nbhood_sum(valid_data_mask, np.int64)
+            area_sum = self._do_nbhood_sum(valid_data_mask)
         # Where data are all ones in nbhood, result will be same as area_sum
-        data = self._do_nbhood_sum(data, out_data_dtype, max_extreme=area_sum)
+        data = self._do_nbhood_sum(data, max_extreme=area_sum)
 
         if not self.sum_only:
             with np.errstate(divide="ignore", invalid="ignore"):
@@ -325,18 +325,15 @@ def _calculate_neighbourhood(
             # For points where all data in the neighbourhood is masked,
             # set result to nan
             data[area_sum == 0] = np.nan
-            data = data.clip(min_val, max_val).astype(out_data_dtype)
+            data = data.clip(min_val, max_val)
 
         if self.re_mask:
             data = np.ma.masked_array(data, data_mask, copy=False)
 
-        return data
+        return data.astype(out_data_dtype)
 
     def _do_nbhood_sum(
-        self,
-        data: np.ndarray,
-        out_data_dtype: Type,
-        max_extreme: Optional[Union[int, np.ndarray]] = None,
+        self, data: np.ndarray, max_extreme: Optional[Union[int, np.ndarray]] = None,
     ) -> np.ndarray:
         """Calculate the sum-in-area from an array.
         As this can be expensive, the method first checks for the extreme cases where the data are:
@@ -386,12 +383,8 @@ def _do_nbhood_sum(
             # Determine default array for the extremes around the edges, or everywhere
             if isinstance(when_all_extremes, np.ndarray):
                 untrimmed = when_all_extremes
-            elif when_all_extremes is None:
-                raise NotImplementedError(
-                    "Don't know what to do when default is None. Shouldn't get here."
-                )
             else:
-                untrimmed = np.full(data_shape, when_all_extremes, dtype=out_data_dtype)
+                untrimmed = np.full(data_shape, when_all_extremes)
         if size:
             # Trim to the calculated box
             data = data[ystart:ystop, xstart:xstop]
@@ -408,7 +401,7 @@ def _do_nbhood_sum(
         if data.shape != data_shape:
             untrimmed[ystart:ystop, xstart:xstop] = data
             data = untrimmed
-        return data.astype(out_data_dtype)
+        return data
 
     def process(self, cube: Cube, mask_cube: Optional[Cube] = None) -> Cube:
         """