Allow custom shapes

.gitignore

@@ -1,5 +1,5 @@
 # generated files
-build/
+/build/
 
 # ignore raw data
 data/
@@ -39,7 +39,6 @@ notebooks/
 
 # Python
 *.pyc
-build/
 dist/
 __pycache__
 *.egg-info

config/default.yaml

@@ -68,7 +68,6 @@ scope:
             x_max: -10  # in degrees east
             y_min: 30  # in degrees north
             y_max: 41  # in degrees north
-
 layers:
     continental:
         Austria: nuts0

config/schema.yaml

@@ -111,19 +111,7 @@ properties:
         patternProperties:
             ^.*$:
                 type: object
-                properties:
-                    ^.*$:  # ideally this would be 'oneof' the list of countries above (can this be done in a schema??)
+                propertyNames: {"$ref": "#/properties/scope/properties/countries/items"}
+                patternProperties:
+                    ^.*$:
                         type: string
-                        enum:
-                            - gadm0
-                            - gadm1
-                            - gadm2
-                            - gadm3
-                            - gadm4
-                            - gadm5
-                            - lau2
-                            - nuts0
-                            - nuts1
-                            - nuts2
-                            - nuts3
-

environment.yaml

@@ -6,3 +6,4 @@ dependencies:
   - python=3.6
   - snakemake-minimal=5.4
   - pycountry=20.7.3
+  - pip

envs/default.yaml

@@ -19,4 +19,5 @@ dependencies:
     - pycountry=20.7.3
     - pip=19.1.1
     - pip:
-        - -e ../../lib
+        - -e ../../lib[geo]
+        - pytest-pythonpath

lib/renewablepotentialslib/__init__.py

@@ -1,10 +1,3 @@
 """Top-level package of renewablepotentialslib."""
 
 __version__ = "0.1.0" # Additionally defined in setup.py.
-
-# from https://epsg.io/3035
-EPSG_3035 = "EPSG:3035"
-EPSG_3035_PROJ4 = "+proj=laea +lat_0=52 +lon_0=10 +x_0=4321000 +y_0=3210000 +ellps=GRS80 +units=m +no_defs "
-# from https://epsg.io/4326
-WGS84 = "EPSG:4326"
-WGS84_PROJ4 = "+proj=longlat +datum=WGS84 +no_defs "

lib/renewablepotentialslib/geo/__init__.py

@@ -0,0 +1,6 @@
+# from https://epsg.io/3035
+EPSG_3035 = "EPSG:3035"
+EPSG_3035_PROJ4 = "+proj=laea +lat_0=52 +lon_0=10 +x_0=4321000 +y_0=3210000 +ellps=GRS80 +units=m +no_defs "
+# from https://epsg.io/4326
+WGS84 = "EPSG:4326"
+WGS84_PROJ4 = "+proj=longlat +datum=WGS84 +no_defs "

lib/renewablepotentialslib/conversion.py → lib/renewablepotentialslib/geo/conversion.py

@@ -17,7 +17,7 @@ def _set_proj_lib():
 import shapely.ops
 import pyproj
 
-from renewablepotentialslib import EPSG_3035_PROJ4
+from renewablepotentialslib.geo import EPSG_3035_PROJ4
 
 
 def watt_to_watthours(watt, duration):

lib/renewablepotentialslib/shape_utils.py → lib/renewablepotentialslib/geo/shape_utils.py

@@ -7,8 +7,8 @@
 import pycountry
 import pyproj
 
-from renewablepotentialslib.conversion import transform_bounds, eu_country_code_to_iso3
-from renewablepotentialslib import EPSG_3035, EPSG_3035_PROJ4, WGS84, WGS84_PROJ4
+from renewablepotentialslib.geo.conversion import transform_bounds, eu_country_code_to_iso3
+from renewablepotentialslib.geo import EPSG_3035, EPSG_3035_PROJ4, WGS84, WGS84_PROJ4
 
 
 def determine_pixel_areas(crs, bounds, resolution):
@@ -168,7 +168,8 @@ def update_features(gdf, src):
 def drop_countries(gdf, scope_config):
     countries = [pycountry.countries.lookup(i).alpha_3 for i in scope_config["countries"]]
     _not_in = set(gdf.country_code).difference(countries)
-    print(f"Removing {_not_in} as they are outside of study area.")
+    if _not_in:
+        print(f"Removing {_not_in} as they are outside of study area.")
 
     return gdf[gdf.country_code.isin(countries)]
 

lib/setup.py

@@ -10,14 +10,17 @@
     maintainer_email='[email protected]',
     packages=find_packages(exclude=['tests*']),
     include_package_data=True,
-    install_requires=[
-        "pycountry",
-        "geopandas",
-        "numpy",
-        "pandas",
-        "shapely",
-        "pyproj"
-    ],
+    install_requires=["numpy"],
+    extras_require={
+        'geo': [
+            "pycountry",
+            "geopandas",
+            "numpy",
+            "pandas",
+            "shapely",
+            "pyproj"
+        ],
+    },
     classifiers=[
         'Environment :: Console',
         'Intended Audience :: Science/Research',

lib/test-requirements.yaml

@@ -13,4 +13,4 @@ dependencies:
     - pandas=0.23.2
     - shapely=1.7.1
     - pip:
-        -e .
+        - -e .[geo]

lib/tests/test_conversion.py → lib/tests/test_geo/test_conversion.py

@@ -6,7 +6,7 @@
 import pandas as pd
 from pandas.testing import assert_frame_equal
 
-from renewablepotentialslib.conversion import (
+from renewablepotentialslib.geo.conversion import (
     watt_to_watthours,
     eu_country_code_to_iso3,
     coordinate_string_to_decimal,

lib/tests/test_shape_utils.py → lib/tests/test_geo/test_shape_utils.py

@@ -3,7 +3,7 @@
 import shapely.geometry
 import pycountry
 
-from renewablepotentialslib.shape_utils import (
+from renewablepotentialslib.geo.shape_utils import (
     study_area,
     to_multi_polygon,
     drop_countries,

pytest.ini

@@ -0,0 +1,3 @@
+[pytest]
+norecursedirs = .git .snakemake lib
+python_paths = rules

rules/data-preprocessing.smk

@@ -1,5 +1,6 @@
 """This is a Snakemake file defining rules to retrieve raw data from online sources."""
 import pycountry
+from rule_utils import collect_shape_dirs
 
 RESOLUTION_STUDY = (1 / 3600) * 10 # 10 arcseconds
 RESOLUTION_SLOPE = (1 / 3600) * 3 # 3 arcseconds
@@ -34,7 +35,7 @@ rule raw_gadm_administrative_borders:
     shell: "unzip -o {input} -d build/raw-gadm"
 
 
-rule all_gadm_administrative_borders:
+rule administrative_borders_gadm:
     message: "Merge gadm administrative borders of all countries."
     input:
         ["build/raw-gadm/gadm36_{}.gpkg".format(country_code)
@@ -47,7 +48,7 @@ rule all_gadm_administrative_borders:
     shell: "ogrmerge.py -o {output} -f gpkg -src_layer_field_content '{{LAYER_NAME}}' -t_srs {params.crs} -single {input}"
 
 
-rule raw_nuts_units:
+rule administrative_borders_nuts:
     message: "Download NUTS units as GeoJSON."
     output:
         protected("data/automatic/raw-nuts{}-units.geojson".format(config["parameters"]["nuts-year"]))
@@ -93,9 +94,7 @@ rule administrative_borders:
     message: "Normalise all administrative borders."
     input:
         src = script_dir + "administrative_borders.py",
-        nuts_geojson = rules.raw_nuts_units.output[0],
-        gadm_gpkg = rules.all_gadm_administrative_borders.output[0],
-        lau_gpkg = rules.administrative_borders_lau.output[0]
+        **collect_shape_dirs(config, rules)
     params:
         crs = config["crs"],
         scope = config["scope"]

rules/rule_utils.py

@@ -0,0 +1,10 @@
+def collect_shape_dirs(config, rules):
+    STRIP = "0123456789"
+    shapes = set(shape.rstrip(STRIP) for layer in config["layers"].values() for shape in layer.values())
+    inputs = {}
+    for shape in shapes:
+        if shape in ["nuts", "gadm", "lau"]:
+            inputs[f"SHAPEINPUT_{shape}"] = getattr(rules, f"administrative_borders_{shape}").output[0]
+        else:
+            inputs[f"SHAPEINPUT_{shape}"] = config["data-sources"][shape]
+    return inputs

scripts/administrative_borders.py

@@ -1,6 +1,6 @@
 import geopandas as gpd
 
-from renewablepotentialslib.shape_utils import (
+from renewablepotentialslib.geo.shape_utils import (
     buffer_if_necessary,
     to_multi_polygon,
     drop_countries,
@@ -22,12 +22,10 @@
 }
 
 
-def normalise_admin_borders(path_to_nuts, path_to_gadm, path_to_lau, crs, scope_config, path_to_output):
+def normalise_admin_borders(crs, scope_config, path_to_output, **shape_dirs):
     """Normalises raw administrative boundary data and places it in one, layered geodatapackage."""
 
-    for _src, _path in {
-        'nuts': path_to_nuts, 'gadm': path_to_gadm, 'lau': path_to_lau
-    }.items():
+    for _src, _path in shape_dirs.items():
         gdf = gpd.read_file(_path)
         gdf = gdf.to_crs(crs)
         gdf.geometry = gdf.geometry.map(buffer_if_necessary).map(to_multi_polygon)
@@ -47,11 +45,13 @@ def normalise_admin_borders(path_to_nuts, path_to_gadm, path_to_lau, crs, scope_
 
 
 if __name__ == "__main__":
+    shape_dirs = {
+        source.replace("SHAPEINPUTS_", ""): source_dir
+        for source, source_dir in snakemake.input.items() if source.startswith("SHAPEINPUTS")
+    }
     normalise_admin_borders(
-        path_to_nuts=snakemake.input.nuts_geojson,
-        path_to_gadm=snakemake.input.gadm_gpkg,
-        path_to_lau=snakemake.input.lau_gpkg,
         crs=snakemake.params.crs,
         scope_config=snakemake.params.scope,
-        path_to_output=snakemake.output[0]
+        path_to_output=snakemake.output[0],
+        **shape_dirs
     )

scripts/built_up_area.py

@@ -4,7 +4,7 @@
 from rasterstats import zonal_stats
 import pandas as pd
 
-from renewablepotentialslib.shape_utils import determine_pixel_areas
+from renewablepotentialslib.geo.shape_utils import determine_pixel_areas
 
 
 def built_up_areas(path_to_built_up_share, path_to_units, path_to_result):

scripts/capacityfactors/id_map.py

@@ -12,7 +12,7 @@
 NO_DATA_VALUE = 64001
 INDEX_EPSILON = 10e-3
 
-from renewablepotentialslib import EPSG_3035, EPSG_3035_PROJ4, WGS84, WGS84_PROJ4
+from renewablepotentialslib.geo import EPSG_3035, EPSG_3035_PROJ4, WGS84, WGS84_PROJ4
 
 
 def id_map(path_to_timeseries, path_to_map, resolution_km):

scripts/capacityfactors/ninja_input_pv.py

@@ -2,8 +2,8 @@
 import pandas as pd
 import geopandas as gpd
 
-from renewablepotentialslib.shape_utils import point_raster_on_shapes
-from renewablepotentialslib.conversion import area_to_capacity, orientation_to_azimuth
+from renewablepotentialslib.geo.shape_utils import point_raster_on_shapes
+from renewablepotentialslib.geo.conversion import area_to_capacity, orientation_to_azimuth
 
 
 def pv_simulation_parameters(path_to_shapes_of_land_surface, path_to_roof_categories,

scripts/capacityfactors/ninja_input_wind.py

@@ -2,7 +2,7 @@
 import pandas as pd
 import geopandas as gpd
 
-from renewablepotentialslib.shape_utils import point_raster_on_shapes
+from renewablepotentialslib.geo.shape_utils import point_raster_on_shapes
 
 
 def wind(path_to_shapes_of_land_surface, path_to_shapes_of_water_surface, bounds, ninja,

scripts/estimate_protected_shapes.py

@@ -9,7 +9,7 @@
 import geopandas as gpd
 import pycountry
 
-from renewablepotentialslib import EPSG_3035_PROJ4
+from renewablepotentialslib.geo import EPSG_3035_PROJ4
 
 
 def estimate_shapes(path_to_input, scope_config, path_to_output):

scripts/lau.py

@@ -2,7 +2,7 @@
 import geopandas as gpd
 import pandas as pd
 
-from renewablepotentialslib.shape_utils import to_multi_polygon
+from renewablepotentialslib.geo.shape_utils import to_multi_polygon
 
 OUTPUT_DRIVER = "GeoJSON"
 KOSOVO_MUNICIPALITIES = [f"RS{x:02d}" for x in range(1, 38)]

scripts/shared_coast.py

@@ -8,7 +8,7 @@
 import geopandas as gpd
 from shapely.prepared import prep
 
-from renewablepotentialslib.shape_utils import buffer_if_necessary
+from renewablepotentialslib.geo.shape_utils import buffer_if_necessary
 
 DRIVER = "GeoJSON"
 

scripts/swiss_building_footprints.py

@@ -6,7 +6,7 @@
 import geopandas as gpd
 
 from renewablepotentialslib.eligibility import Eligibility
-from renewablepotentialslib.conversion import area_in_squaremeters
+from renewablepotentialslib.geo.conversion import area_in_squaremeters
 
 
 def swiss_building_footprint(path_to_building_footprint, path_to_eligibility,

scripts/technically_eligible_area.py

@@ -2,7 +2,7 @@
 import numpy as np
 import rasterio
 
-from renewablepotentialslib.shape_utils import determine_pixel_areas
+from renewablepotentialslib.geo.shape_utils import determine_pixel_areas
 from renewablepotentialslib.eligibility import Eligibility
 
 DATATYPE = np.float32

scripts/technically_eligible_electricity_yield.py

@@ -4,7 +4,7 @@
 import numpy as np
 import rasterio
 
-from renewablepotentialslib.conversion import watt_to_watthours
+from renewablepotentialslib.geo.conversion import watt_to_watthours
 
 
 def determine_electricity_yield(path_to_eligibility_categories, path_to_capacities_pv_prio,

tests/test_administrative_borders.py → tests/build/test_administrative_borders.py

@@ -12,6 +12,7 @@
 
 PATH_TO_BORDERS = ROOT_DIR / "build" / "administrative-borders.gpkg"
 
+
 @pytest.mark.skipif(not PATH_TO_BORDERS.exists(), reason="Consolidated administrative border shapefile not available.")
 def test_administrative_border_layers():
     layers = fiona.listlayers(str(PATH_TO_BORDERS))