Feature/issue 267

CHANGELOG.md

@@ -6,6 +6,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 ### Added
+- [issue/267](https://github.com/podaac/l2ss-py/pull/261): Add xtrack and atrack dimension options for get_nd_indexers when bounding box subsetting is performed on SNDR.
 ### Changed
 ### Deprecated 
 ### Removed

podaac/subsetter/gpm_cleanup.py

@@ -30,7 +30,7 @@ def compute_new_time_data(time_group, nc_dataset):
     return new_time_list, time_unit_out
 
 
-def change_var_dims(nc_dataset, variables=None, time_name="_timeMidScan"):
+def change_var_dims(nc_dataset, variables=None, time_name="__timeMidScan"):
     """
     Go through each variable and get the dimension names from attribute "DimensionNames
     If the name is unique, add it as a dimension to the netCDF4 dataset. Then change the

podaac/subsetter/xarray_enhancements.py

@@ -69,12 +69,20 @@ def get_indexers_from_nd(cond: xr.Dataset, cut: bool) -> dict:
         Indexer dictionary for the provided condition.
     """
     # check if the lat/lon coordinate numpy array has 2 or more dimensions
+    transpose = False
     if cond.values.squeeze().ndim == 2:
         x_axis = 1
         y_axis = 0
     else:
-        x_axis = 2
-        y_axis = 1
+        if any('xtrack' in dim for dim in list(cond.dims)) and\
+           any('atrack' in dim for dim in list(cond.dims)):
+            x_axis = list(cond.dims).index('xtrack')
+            y_axis = list(cond.dims).index('atrack')
+            transpose = True
+        else:
+            x_axis = 2
+            y_axis = 1
+
     rows = np.any(cond.values.squeeze(), axis=x_axis)
     if cut:
         cols = np.any(cond.values.squeeze(), axis=y_axis)
@@ -102,11 +110,15 @@ def get_indexers_from_nd(cond: xr.Dataset, cut: bool) -> dict:
         }
     else:
         # if the lat/lon had 3 dimensions the conditional array was identical in the z direction - taking the first
-        rows = rows[0]
-        cols = cols[0]
+        if transpose:
+            rows = rows.transpose()[0]
+            cols = cols.transpose()[0]
+        else:
+            rows = rows[0]
+            cols = cols[0]
         indexers = {
-            cond_list[1]: np.where(rows)[0],
-            cond_list[2]: np.where(cols)[0]
+            cond_list[y_axis]: np.where(rows)[0],
+            cond_list[x_axis]: np.where(cols)[0]
         }
 
     return indexers

tests/test_subset.py

@@ -1311,8 +1311,10 @@ def test_cf_decode_times_sndr(data_dir, subset_output_dir, request):
     sndr_files = ['SNDR.J1.CRIMSS.20210224T0100.m06.g011.L2_CLIMCAPS_RET.std.v02_28.G.210331064430.nc',
                   'SNDR.AQUA.AIRS.20140110T0305.m06.g031.L2_CLIMCAPS_RET.std.v02_39.G.210131015806.nc',
                   'SNDR.SNPP.CRIMSS.20200118T0024.m06.g005.L2_CLIMCAPS_RET.std.v02_28.G.200314032326_subset.nc']
-    bbox = np.array(((-180, 180), (-90, 90)))
-    for sndr_file in sndr_files:
+    # do a longitude subset on these files that doesn't alter the resulting shape
+    sndr_spatial = [(-180,-150), (-15,180), (-180,30)]
+    for sndr_file, box in zip(sndr_files, sndr_spatial):
+        bbox = np.array(((box[0], box[1]), (-90, 90)))
         output_file = "{}_{}".format(request.node.name, sndr_file)
         shutil.copyfile(
             os.path.join(SNDR_dir, sndr_file),
@@ -1327,6 +1329,14 @@ def test_cf_decode_times_sndr(data_dir, subset_output_dir, request):
             max_time='2021-02-24T01:09:55Z'
         )
 
+        out_ds = xr.open_dataset(join(subset_output_dir, output_file),
+                                    decode_coords=False)
+        in_ds = xr.open_dataset(join(SNDR_dir, sndr_file),
+                                    decode_coords=False)
+
+        # do a longitude subset that cuts down on the file but the shape should remain the same
+        assert out_ds['lon'].shape == in_ds['lon'].shape
+
         if not isinstance(box_test, np.ndarray):
             raise ValueError('Subset for SNDR not returned properly')