Add locate_bus to power model

scripts/_helpers.py

@@ -1232,6 +1232,7 @@ def locate_bus(
     path_to_gadm=None,
     gadm_clustering=False,
     dropnull=True,
+    col_out=None,
 ):
     """
     Function to locate the points of the dataframe df into the GADM shapefile.
@@ -1244,15 +1245,18 @@ def locate_bus(
         List of target countries
     gadm_level: int
         GADM level to be used
-    path_to_gadm: str
+    path_to_gadm: str (default None)
         Path to the GADM shapefile
-    gadm_clustering: bool
+    gadm_clustering: bool (default False)
         True if gadm clustering is adopted
-    dropnull: bool
+    dropnull: bool (default True)
         True if the rows with null values should be dropped
+    col_out: str (default gadm_{gadm_level})
+        Name of the output column
     """
+    if col_out is None:
+        col_out = "gadm_{}".format(gadm_level)
     df = df[df.country.isin(countries)]
-    col_out = "gadm_{}".format(gadm_level)
     df[col_out] = None
     for co in countries:
         gdf_shape, col = _get_shape_col_gdf(

scripts/build_powerplants.py

@@ -111,6 +111,7 @@
 from _helpers import (
     configure_logging,
     create_logger,
+    locate_bus,
     read_csv_nafix,
     to_csv_nafix,
     two_digits_2_name_country,
@@ -371,28 +372,13 @@ def replace_natural_gas_technology(df: pd.DataFrame):
         country_list = snakemake.params.countries
         geo_crs = snakemake.params.geo_crs
 
-        gdf = gpd.read_file(snakemake.input.gadm_shapes)
-
-        def locate_bus(coords, co):
-            gdf_co = gdf[gdf["GADM_ID"].str.contains(co)]
-
-            point = Point(coords["lon"], coords["lat"])
-
-            try:
-                return gdf_co[gdf_co.contains(point)][
-                    "GADM_ID"
-                ].item()  # filter gdf_co which contains point and returns the bus
-
-            except ValueError:
-                return gdf_co[
-                    gdf_co.geometry == min(gdf_co.geometry, key=(point.distance))
-                ][
-                    "GADM_ID"
-                ].item()  # looks for closest one shape=node
-                # fixing https://github.com/pypsa-meets-earth/pypsa-earth/pull/670
-
-        ppl["region_id"] = ppl[["lon", "lat", "Country"]].apply(
-            lambda pp: locate_bus(pp[["lon", "lat"]], pp["Country"]), axis=1
-        )
+        ppl = locate_bus(
+            ppl.rename(columns={"lon": "x", "lat": "y", "Country": "country"}),
+            country_list,
+            gadm_layer_id,
+            snakemake.input.gadm_shapes,
+            snakemake.params.alternative_clustering,
+            col_out="region_id",
+        ).rename(columns={"x": "lon", "y": "lat", "country": "Country"})
 
     ppl.to_csv(snakemake.output.powerplants)

scripts/cluster_network.py

@@ -134,6 +134,7 @@
     configure_logging,
     create_logger,
     get_aggregation_strategies,
+    locate_bus,
     update_p_nom_max,
 )
 from add_electricity import load_costs
@@ -379,23 +380,13 @@ def n_bounds(model, *n_id):
 
 
 def busmap_for_gadm_clusters(inputs, n, gadm_level, geo_crs, country_list):
-    gdf = gpd.read_file(inputs.gadm_shapes)
 
-    def locate_bus(coords, co):
-        gdf_co = gdf[gdf["GADM_ID"].str.contains(co)]
-        point = Point(coords["x"], coords["y"])
-
-        try:
-            return gdf_co[gdf_co.contains(point)]["GADM_ID"].item()
-
-        except ValueError:
-            return gdf_co[
-                gdf_co.geometry == min(gdf_co.geometry, key=(point.distance))
-            ]["GADM_ID"].item()
-
-    buses = n.buses
-    buses["gadm_{}".format(gadm_level)] = buses[["x", "y", "country"]].apply(
-        lambda bus: locate_bus(bus[["x", "y"]], bus["country"]), axis=1
+    buses = locate_bus(
+        n.buses,
+        country_list,
+        gadm_level,
+        inputs.gadm_shapes,
+        gadm_clustering=True,
     )
 
     buses["gadm_subnetwork"] = (
@@ -657,7 +648,10 @@ def cluster_regions(busmaps, inputs, output):
         from _helpers import mock_snakemake
 
         snakemake = mock_snakemake(
-            "cluster_network", network="elec", simpl="", clusters="min"
+            "cluster_network",
+            network="elec",
+            simpl="",
+            clusters="min",
         )
     configure_logging(snakemake)