Added support for only offshore data

src/geofabrics/dem.py

@@ -527,6 +527,18 @@ def _sample_offshore_edge(self, resolution) -> numpy.ndarray:
         offshore_dense_data_edge = self.catchment_geometry.offshore_dense_data_edge(
             self._raw_extents
         )
+        if offshore_dense_data_edge.area.sum() == 0:
+            # No offshore edge. Return an empty array.
+            offshore_edge = numpy.empty(
+                [0],
+                dtype=[
+                    ("X", geometry.RASTER_TYPE),
+                    ("Y", geometry.RASTER_TYPE),
+                    ("Z", geometry.RASTER_TYPE),
+                ],
+            )
+            return offshore_edge
+        # Otherwise proceed as normal
         offshore_edge_dem = self._raw_dem.rio.clip(offshore_dense_data_edge.geometry)
 
         # If the sampling resolution is coaser than the catchment_geometry resolution
@@ -990,11 +1002,14 @@ def _define_chunk_region(
         )
 
         # Define region to rasterise inside the chunk area
-        chunk_region_to_tile = geopandas.GeoDataFrame(
-            geometry=region_to_rasterise.buffer(radius).clip(
-                chunk_geometry.buffer(radius), keep_geom_type=True
+        if region_to_rasterise.area.sum() > 0:
+            chunk_region_to_tile = geopandas.GeoDataFrame(
+                geometry=region_to_rasterise.buffer(radius).clip(
+                    chunk_geometry.buffer(radius), keep_geom_type=True
+                )
             )
-        )
+        else: 
+            chunk_region_to_tile = region_to_rasterise
         # remove any subpixel polygons
         chunk_region_to_tile = chunk_region_to_tile[
             chunk_region_to_tile.area
@@ -1325,7 +1340,8 @@ def add_lidar(
 
         # Clip DEM to Catchment and ensure NaN outside region to rasterise
         dem = dem.rio.clip(self.catchment_geometry.catchment.geometry, drop=True)
-        dem = dem.rio.clip(clip_region.geometry, drop=False)
+        if clip_region.area.sum() > 0: # If area of 0 all will be NaN anyway
+            dem = dem.rio.clip(clip_region.geometry, drop=False)
 
         # If drop offshrore LiDAR ensure the foreshore values are 0 or negative
         if (
@@ -1715,13 +1731,16 @@ def add_coarse_dems(
                 .count()
                 .isnull()
             )
-            no_value_mask &= (
-                xarray.ones_like(self._dem.z)
-                .rio.clip(
-                    self.catchment_geometry.land_and_foreshore.geometry, drop=False
-                )
-                .notnull()
-            )  # Awkward as clip of a bool xarray doesn't work as expected
+            if self.catchment_geometry.land_and_foreshore.area.sum() > 0:
+                no_value_mask &= (
+                    xarray.ones_like(self._dem.z)
+                    .rio.clip(
+                        self.catchment_geometry.land_and_foreshore.geometry, drop=False
+                    )
+                    .notnull()
+                )  # Awkward as clip of a bool xarray doesn't work as expected
+            else:
+                no_value_mask = xarray.zeros_like(self._dem.z)
             if not no_value_mask.any():
                 logging.info(
                     f"No land areas greater than the cell buffer {buffer_cells}"

src/geofabrics/geometry.py

@@ -70,28 +70,35 @@ def _set_up(self):
         # Clip and remove any sub pixel regions
         self._land = self._catchment.clip(self._full_land, keep_geom_type=True)
         self._land = self._land[self._land.area > self.resolution * self.resolution]
+        if self._land.area.sum() > 0:
+            self._foreshore_and_offshore = self.catchment.overlay(
+                self.land, how="difference"
+            )
+            # Buffer and clip and remove any sub pixel regions
+            self._land_and_foreshore = geopandas.GeoDataFrame(
+                {"geometry": self.land.buffer(self.resolution * self.foreshore_buffer)},
+                crs=self.crs["horizontal"],
+            )
+            self._land_and_foreshore = self._catchment.clip(
+                self._land_and_foreshore, keep_geom_type=True
+            )
+            self._land_and_foreshore = self._land_and_foreshore[
+                self._land_and_foreshore.area > self.resolution * self.resolution
+            ]
 
-        self._foreshore_and_offshore = self.catchment.overlay(
-            self.land, how="difference"
-        )
+            self._foreshore = self._land_and_foreshore.overlay(self.land, how="difference")
 
-        # Buffer and clip and remove any sub pixel regions
-        self._land_and_foreshore = geopandas.GeoDataFrame(
-            {"geometry": self.land.buffer(self.resolution * self.foreshore_buffer)},
-            crs=self.crs["horizontal"],
-        )
-        self._land_and_foreshore = self._catchment.clip(
-            self._land_and_foreshore, keep_geom_type=True
-        )
-        self._land_and_foreshore = self._land_and_foreshore[
-            self._land_and_foreshore.area > self.resolution * self.resolution
-        ]
+            self._offshore = self.catchment.overlay(
+                self.land_and_foreshore, how="difference"
+            )
+        else:
+            # Assume no foreshore
+            self._foreshore_and_offshore = self.catchment
+            self._foreshore = self._land
+            self._land_and_foreshore = self._land
+            self._offshore = self.catchment
 
-        self._foreshore = self._land_and_foreshore.overlay(self.land, how="difference")
-
-        self._offshore = self.catchment.overlay(
-            self.land_and_foreshore, how="difference"
-        )
+
 
         assert len(self._catchment) == 1, (
             "The catchment is made of multiple separate polygons it must be a single "
@@ -166,12 +173,15 @@ def land_and_foreshore_without_lidar(self, dense_extents: geopandas.GeoDataFrame
 
         self._assert_land_set()
 
-        if dense_extents is None:
+        if dense_extents is None or dense_extents.is_empty.all():
             logging.warning(
                 "In CatchmentGeometry dense extents are `None` so "
                 " `land_and_foreshore_without_lidar` is returning `land_and_foreshore`"
             )
             return self.land_and_foreshore
+        elif self.land_and_foreshore.area.sum() == 0:
+            # There is no land and foreshore region - return an empty dataframe
+            return self.land_and_foreshore
         # Clip to remove any offshore regions before doing a difference overlay. Drop
         # any sub-pixel polygons.
         land_and_foreshore_with_lidar = dense_extents.clip(
@@ -191,7 +201,7 @@ def offshore_without_lidar(self, dense_extents: geopandas.GeoDataFrame):
 
         self._assert_land_set()
 
-        if dense_extents is None:
+        if dense_extents is None or dense_extents.is_empty.all():
             logging.warning(
                 "In CatchmentGeometry dense extents are `None` so "
                 "`offshore_without_lidar` is returning `offshore`"
@@ -219,12 +229,16 @@ def offshore_dense_data_edge(self, dense_extents: geopandas.GeoDataFrame):
 
         self._assert_land_set()
 
-        if dense_extents is None:
+        if dense_extents is None or dense_extents.is_empty.all():
             logging.warning(
                 "In CatchmentGeometry dense extents are `None` so "
                 "`offshore_dense_data_edge` is returning `None`"
             )
-            return None
+            return dense_extents
+        elif self.foreshore.area.sum():
+            # If no foreshore just return the foreshore
+            return self.foreshore
+
         # the foreshore and whatever lidar extents are offshore
         offshore_foreshore_dense_data_extents = geopandas.GeoDataFrame(
             {

src/geofabrics/processor.py

@@ -895,6 +895,8 @@ def run(self):
         with cluster, distributed.Client(cluster) as client:
             print("Dask client:", client)
             print("Dask dashboard:", client.dashboard_link)
+            logging.info("Dask client:", client)
+            logging.info("Dask dashboard:", client.dashboard_link)
 
             # Load in LiDAR tiles
             self.raw_dem.add_lidar(
@@ -1106,6 +1108,8 @@ def run(self):
         with cluster, distributed.Client(cluster) as client:
             print("Dask client:", client)
             print("Dask dashboard:", client.dashboard_link)
+            logging.info("Dask client:", client)
+            logging.info("Dask dashboard:", client.dashboard_link)
 
             # setup the hydrologically conditioned DEM generator
             self.hydrologic_dem = dem.HydrologicallyConditionedDem(