Index: trunk/tests/test_data/system_tests_nc/dummy_hs_gridded.nc
===================================================================
diff -u
Binary files differ
Index: trunk/SDToolBox/extract_data.py
===================================================================
diff -u -r75 -r76
--- trunk/SDToolBox/extract_data.py	(.../extract_data.py)	(revision 75)
+++ trunk/SDToolBox/extract_data.py	(.../extract_data.py)	(revision 76)
@@ -328,12 +328,14 @@
                 nn_lat_idx = self.__set_nn(
                     input_values=input_data._input_lat,
                     reference_list=ref_dataset.variables[self.lat_key][:],
-                    output_values=cases_dict[OutputData.var_lat_key]
+                    output_values=cases_dict[OutputData.var_lat_key],
+                    is_gridded=input_data.is_gridded
                 )
                 nn_lon_idx = self.__set_nn(
                     input_values=input_data._input_lon,
                     reference_list=ref_dataset.variables[self.lon_key][:],
                     output_values=cases_dict[OutputData.var_lon_key],
+                    is_gridded=input_data.is_gridded
                 )
             return nn_lon_idx, nn_lat_idx
 
@@ -359,19 +361,33 @@
                 self,
                 input_values: List[float],
                 reference_list: List[float],
-                output_values: List[float]) -> List[int]:
+                output_values: List[float],
+                is_gridded: bool) -> List[int]:
             """Sets the nearest neighbor for all the elements
             given in the points_list.
 
             Arguments:
                 input_values {List[float]} -- List of elements to correct.
                 reference_list {List[float]} -- Available neighbor list.
                 output_values {List[float]} -- Corrected values.
+                is_gridded {bool} -- Needs to extract nn as gridded dataset.
 
             Returns:
                 List[int] -- Indices of the nearest neighbors positions
             """
             output_idx = []
+            if is_gridded:
+                min_idx, min_value = self.get_nearest_neighbor(
+                    min(input_values),
+                    reference_list)
+                max_idx, max_value = self.get_nearest_neighbor(
+                    max(input_values),
+                    reference_list)
+                nn_idx = sorted([min_idx, max_idx])
+                output_values.extend(
+                    reference_list[nn_idx[0]:nn_idx[-1]])
+                return list(nn_idx)
+            # Non-gridded.
             for point in input_values:
                 idx, value = self.get_nearest_neighbor(
                     point,
@@ -394,15 +410,34 @@
             return index_found, value_found
 
         @staticmethod
-        def __get_case_subset(dataset, variable_name, input_data, lon, lat):
+        def __get_case_subset(
+                dataset: Dataset,
+                variable_name: str,
+                input_data: InputData,
+                lon: List[int],
+                lat: List[int]):
+            """Gets the subset given a lat-lon.
+
+            Arguments:
+                dataset {netCDF4.Dataset} -- Input Dataset.
+                variable_name {str} -- Name of the variable to extract.
+                input_data {InputData} -- Input parameters.
+                lon {List[int]} -- Longitude indices.
+                lat {List[int]} -- Latitude indices.
+
+            Returns:
+                netCDF4.array -- Dataset result subset.
+            """
+            # Dataset[Variable][time, latitude, longitude]
             if not input_data.is_gridded:
                 vc = []
                 # We access to the points individually.
                 for l in range(0, len(lat)):
                     vc.append(dataset[variable_name][:, lat[l], lon[l]])
                 return vc
 
-            # Is gridded, we get only an ortogonal of the lat-lon.
+            # Is gridded, we ge
+            # it only an ortogonal of the lat-lon.
             return dataset[variable_name][
                 :,
                 lat[0]:lat[-1],
Index: trunk/SDToolBox/input_data.py
===================================================================
diff -u -r75 -r76
--- trunk/SDToolBox/input_data.py	(.../input_data.py)	(revision 75)
+++ trunk/SDToolBox/input_data.py	(.../input_data.py)	(revision 76)
@@ -157,8 +157,8 @@
             self.__get_corrected_longitude(lon)
             for lon in self._input_lon]
         if self.is_gridded:
-            self._input_lat = set(self._input_lat)
-            self._input_lon = list(set(self._input_lon))
+            self._input_lat = sorted(set(self._input_lat))
+            self._input_lon = list(sorted(set(self._input_lon)))
         self._input_lat = list(self._input_lat)
 
     def __get_corrected_longitude(self, longitude: int):
Index: trunk/tests/test_data/system_tests_nc/dummy_hs_non_gridded.nc
===================================================================
diff -u
Binary files differ
Index: trunk/SDToolBox/output_data.py
===================================================================
diff -u -r75 -r76
--- trunk/SDToolBox/output_data.py	(.../output_data.py)	(revision 75)
+++ trunk/SDToolBox/output_data.py	(.../output_data.py)	(revision 76)
@@ -591,5 +591,10 @@
                 continue
             # Assign the data from the frame towards the netCDF-parameter
             print('writing variable {}'.format(var_name))
-            netcdf.variables[var_name.upper()][:] = \
-                np.transpose(variables[var_name])
+            if self.is_gridded:
+                netcdf.variables[var_name.upper()][:] = \
+                    variables[var_name]
+            else:
+                # When non-gridded we need to transpose the dimensions.
+                netcdf.variables[var_name.upper()][:] = \
+                    np.transpose(variables[var_name])