EEZ: Update EEZ to v12, auto-download and remove from databundle (#1188)

* eez: update to version 12, autodownload, remove pycountry

* eez: do not simplify as it distorts topology

* remove missed merge conflicts

doc/conf.py

@@ -53,7 +53,6 @@ extensions = [
 autodoc_mock_imports = [
     "atlite",
     "snakemake",
-    "pycountry",
     "rioxarray",
     "country_converter",
     "tabula",

doc/requirements.txt

@@ -17,7 +17,6 @@ tabula-py
 
 # cartopy
 scikit-learn
-pycountry
 pyyaml
 seaborn
 memory_profiler

envs/environment.yaml

@@ -18,7 +18,6 @@ dependencies:
   # Dependencies of the workflow itself
 - xlrd
 - openpyxl!=3.1.1
-- pycountry
 - seaborn
 - snakemake-minimal>=8.14
 - memory_profiler

rules/build_electricity.smk

@@ -92,7 +92,7 @@ rule build_shapes:
         countries=config_provider("countries"),
     input:
         naturalearth=ancient("data/naturalearth/ne_10m_admin_0_countries_deu.shp"),
-        eez=ancient("data/bundle/eez/World_EEZ_v8_2014.shp"),
+        eez=ancient("data/eez/World_EEZ_v12_20231025_gpkg/eez_v12.gpkg"),
         nuts3=ancient("data/bundle/NUTS_2013_60M_SH/data/NUTS_RG_60M_2013.shp"),
         nuts3pop=ancient("data/bundle/nama_10r_3popgdp.tsv.gz"),
         nuts3gdp=ancient("data/bundle/nama_10r_3gdp.tsv.gz"),

rules/retrieve.smk

@@ -16,7 +16,6 @@ if config["enable"]["retrieve"] is False:
 if config["enable"]["retrieve"] and config["enable"].get("retrieve_databundle", True):
     datafiles = [
         "je-e-21.03.02.xls",
-        "eez/World_EEZ_v8_2014.shp",
         "NUTS_2013_60M_SH/data/NUTS_RG_60M_2013.shp",
         "nama_10r_3popgdp.tsv.gz",
         "nama_10r_3gdp.tsv.gz",
@@ -215,6 +214,41 @@ if config["enable"]["retrieve"]:
             move(input[0], output[0])
 
 
+if config["enable"]["retrieve"]:
+
+    rule retrieve_eez:
+        params:
+            zip="data/eez/World_EEZ_v12_20231025_gpkg.zip",
+        output:
+            gpkg="data/eez/World_EEZ_v12_20231025_gpkg/eez_v12.gpkg",
+        run:
+            import os
+            import requests
+            from uuid import uuid4
+
+            name = str(uuid4())[:8]
+            org = str(uuid4())[:8]
+
+            response = requests.post(
+                "https://www.marineregions.org/download_file.php",
+                params={"name": "World_EEZ_v12_20231025_gpkg.zip"},
+                data={
+                    "name": name,
+                    "organisation": org,
+                    "email": f"{name}@{org}.org",
+                    "country": "Germany",
+                    "user_category": "academia",
+                    "purpose_category": "Research",
+                    "agree": "1",
+                },
+            )
+
+            with open(params["zip"], "wb") as f:
+                f.write(response.content)
+            output_folder = Path(params["zip"]).parent
+            unpack_archive(params["zip"], output_folder)
+            os.remove(params["zip"])
+
 if config["enable"]["retrieve"]:
 
     # Download directly from naciscdn.org which is a redirect from naturalearth.com

scripts/build_shapes.py

@@ -26,7 +26,7 @@ Inputs
     .. image:: img/countries.png
         :scale: 33 %
 
-- ``data/bundle/eez/World_EEZ_v8_2014.shp``: World `exclusive economic zones <https://en.wikipedia.org/wiki/Exclusive_economic_zone>`_ (EEZ)
+- ``data/eez/World_EEZ_v12_20231025_gpkg/eez_v12.gpkg   ``: World `exclusive economic zones <https://en.wikipedia.org/wiki/Exclusive_economic_zone>`_ (EEZ)
 
     .. image:: img/eez.png
         :scale: 33 %
@@ -76,19 +76,13 @@ from operator import attrgetter
 import geopandas as gpd
 import numpy as np
 import pandas as pd
-import pycountry as pyc
+import country_converter as coco
 from _helpers import configure_logging, set_scenario_config
 from shapely.geometry import MultiPolygon, Polygon
 
 logger = logging.getLogger(__name__)
 
-
-def _get_country(target, **keys):
-    assert len(keys) == 1
-    try:
-        return getattr(pyc.countries.get(**keys), target)
-    except (KeyError, AttributeError):
-        return np.nan
+cc = coco.CountryConverter()
 
 
 def _simplify_polys(polys, minarea=0.1, tolerance=None, filterremote=True):
@@ -135,22 +129,15 @@ def countries(naturalearth, country_list):
     return s
 
 
-def eez(country_shapes, eez, country_list):
+def eez(eez, country_list):
     df = gpd.read_file(eez)
-    df = df.loc[
-        df["ISO_3digit"].isin(
-            [_get_country("alpha_3", alpha_2=c) for c in country_list]
-        )
-    ]
-    df["name"] = df["ISO_3digit"].map(lambda c: _get_country("alpha_2", alpha_3=c))
+    iso3_list = cc.convert(country_list, src="ISO2", to="ISO3")
+    df = df.query("ISO_TER1 in @iso3_list and POL_TYPE == '200NM'").copy()
+    df["name"] = cc.convert(df.ISO_TER1, src="ISO3", to="ISO2")
     s = df.set_index("name").geometry.map(
         lambda s: _simplify_polys(s, filterremote=False)
     )
-    s = gpd.GeoSeries(
-        {k: v for k, v in s.items() if v.distance(country_shapes[k]) < 1e-3},
-        crs=df.crs,
-    )
-    s = s.to_frame("geometry")
+    s = s.to_frame("geometry").set_crs(df.crs)
     s.index.name = "name"
     return s
 
@@ -262,9 +249,7 @@ if __name__ == "__main__":
     country_shapes = countries(snakemake.input.naturalearth, snakemake.params.countries)
     country_shapes.reset_index().to_file(snakemake.output.country_shapes)
 
-    offshore_shapes = eez(
-        country_shapes, snakemake.input.eez, snakemake.params.countries
-    )
+    offshore_shapes = eez(snakemake.input.eez, snakemake.params.countries)
     offshore_shapes.reset_index().to_file(snakemake.output.offshore_shapes)
 
     europe_shape = gpd.GeoDataFrame(