Include assign weather_data

CHANGELOG.rst

@@ -137,12 +137,17 @@ Added
 * Extend zensus by a combined table with all cells where
   there's either building, apartment or population data
   `#359 <https://github.com/openego/eGon-data/issues/359>`_
+<<<<<<< HEAD
 * Add example metadata for OSM, VG250 and Zensus VG250.
   Add metadata templates for licences, context and some helper
   functions. Extend docs on how to create metadata for tables.
   `#139 <https://github.com/openego/eGon-data/issues/139>`_
 * Integrate DSM potentials for CTS and industry
   `#259 <https://github.com/openego/eGon-data/issues/259>`_
+=======
+* Assign weather cell id to weather dependant power plants
+  `#330 <https://github.com/openego/eGon-data/issues/330>`_
+>>>>>>> features/#330-assign-weather-cell-id-to-power-plants
 
 .. _PR #159: https://github.com/openego/eGon-data/pull/159
 

src/egon/data/datasets/power_plants/__init__.py

@@ -22,6 +22,8 @@
 from egon.data.datasets.power_plants.pv_rooftop import pv_rooftop_per_mv_grid
 import egon.data.config
 import egon.data.datasets.power_plants.pv_ground_mounted as pv_ground_mounted
+
+import egon.data.datasets.power_plants.assign_weather_data as assign_weather_data
 import egon.data.datasets.power_plants.wind_farms as wind_onshore
 
 Base = declarative_base()
@@ -53,11 +55,10 @@ def __init__(self, dependencies):
             tasks=(
                 create_tables,
                 insert_hydro_biomass,
-                {
-                    wind_onshore.insert,
-                    pv_ground_mounted.insert,
-                    pv_rooftop_per_mv_grid,
-                },
+                wind_onshore.insert,
+                pv_ground_mounted.insert,
+                pv_rooftop_per_mv_grid,
+                assign_weather_data.weather_id,
             ),
         )
 

src/egon/data/datasets/power_plants/assign_weather_data.py

@@ -0,0 +1,101 @@
+import pandas as pd
+import geopandas as gpd
+import numpy as np
+from egon.data import db
+from shapely.geometry import Point
+from pathlib import Path
+import egon.data.config
+
+
+def weather_id():
+    """
+    Assign weather data to the weather dependant generators (wind and solar)
+
+    Parameters
+    ----------
+    *No parameters required
+    """
+
+    # Connect to the data base
+    con = db.engine()
+
+    cfg = egon.data.config.datasets()["power_plants"]
+
+    # Import table with power plants
+    sql = "SELECT * FROM supply.egon_power_plants"
+    power_plants = gpd.GeoDataFrame.from_postgis(
+        sql, con, crs="EPSG:4326", geom_col="geom"
+    )
+
+    # select the power_plants that are weather dependant
+    power_plants = power_plants[
+        (power_plants["carrier"] == "solar")
+        | (power_plants["carrier"] == "wind_onshore")
+        | (power_plants["carrier"] == "wind_offshore")
+    ]
+    power_plants.set_index("id", inplace=True)
+
+    # Import table with weather data for each technology
+    sql = "SELECT * FROM supply.egon_era5_renewable_feedin"
+    weather_data = pd.read_sql_query(sql, con)
+    weather_data.set_index("w_id", inplace=True)
+
+    # Import weather cells with Id to match with the weather data
+    sql = "SELECT * FROM supply.egon_era5_weather_cells"
+    weather_cells = gpd.GeoDataFrame.from_postgis(
+        sql, con, crs="EPSG:4326", geom_col="geom"
+    )
+
+    # import Germany borders to speed up the matching process
+    sql = "SELECT * FROM boundaries.vg250_sta"
+    boundaries = gpd.GeoDataFrame.from_postgis(
+        sql, con, crs="EPSG:4326", geom_col="geometry"
+    )
+
+    # Clip weater data cells using the German boundaries
+    weather_cells = gpd.clip(weather_cells, boundaries)
+
+    for weather_id in weather_cells["w_id"]:
+        df = gpd.clip(
+            power_plants, weather_cells[weather_cells["w_id"] == weather_id]
+        )
+        power_plant_list = df.index.to_list()
+        power_plants.loc[power_plant_list, "weather_cell_id"] = weather_id
+
+    # delete weather dependent power_plants from supply.egon_power_plants
+    db.execute_sql(
+        f""" 
+    DELETE FROM {cfg['target']['schema']}.{cfg['target']['table']} 
+    WHERE carrier IN ('wind_onshore', 'solar', 'wind_offshore') 
+    """
+    )
+
+    # Look for the maximum id in the table egon_power_plants
+    sql = (
+        "SELECT MAX(id) FROM "
+        + cfg["target"]["schema"]
+        + "."
+        + cfg["target"]["table"]
+    )
+    max_id = pd.read_sql(sql, con)
+    max_id = max_id["max"].iat[0]
+    if max_id == None:
+        ini_id = 1
+    else:
+        ini_id = int(max_id + 1)
+
+    # write_table in egon-data database:
+    # Reset index
+    power_plants.index = pd.RangeIndex(
+        start=ini_id, stop=ini_id + len(power_plants), name="id"
+    )
+
+    # Insert into database
+    power_plants.reset_index().to_postgis(
+        cfg["target"]["table"],
+        schema=cfg["target"]["schema"],
+        con=db.engine(),
+        if_exists="append",
+    )
+
+    return 0

src/egon/data/datasets/power_plants/pv_ground_mounted.py

@@ -1062,6 +1062,7 @@ def run_methodology(
             pv_per_distr_100RE["installed capacity in kW"] > 0
         ]
 
+
         return (
             pv_rora,
             pv_agri,
@@ -1145,10 +1146,11 @@ def pv_parks(pv_rora, pv_agri, pv_per_distr, scenario_name):
         insert_pv_parks["scenario"] = scenario_name
 
         # change name and crs of geometry column
+        insert_pv_parks.set_crs(epsg= 3035, allow_override= True, inplace= True)
         insert_pv_parks = (
             insert_pv_parks.rename({"geometry": "geom"}, axis=1)
             .set_geometry("geom")
-            .set_crs(4326)
+            .to_crs(4326)
         )
 
         # reset index

src/egon/data/datasets/renewable_feedin.py

@@ -17,14 +17,14 @@ class RenewableFeedin(Dataset):
     def __init__(self, dependencies):
         super().__init__(
             name="RenewableFeedin",
-            version="0.0.0",
+            version="0.0.1",
             dependencies=dependencies,
-            tasks={wind, pv, solar_thermal},
+            tasks=(wind, pv, solar_thermal),
         )
 
 
 def weather_cells_in_germany(geom_column="geom"):
-    """ Get weather cells which intersect with Germany
+    """Get weather cells which intersect with Germany
 
     Returns
     -------
@@ -36,19 +36,18 @@ def weather_cells_in_germany(geom_column="geom"):
     cfg = egon.data.config.datasets()["renewable_feedin"]["sources"]
 
     return db.select_geodataframe(
-        f"""SELECT a.w_id, a.geom_point, a.geom
+        f"""SELECT w_id, geom_point, geom
         FROM {cfg['weather_cells']['schema']}.
-        {cfg['weather_cells']['table']} a,
-        {cfg['vg250_sta_union']['schema']}.
-        {cfg['vg250_sta_union']['table']} b
-        WHERE ST_Intersects(ST_Transform(b.geometry, 4326), a.geom)""",
+        {cfg['weather_cells']['table']}
+        WHERE ST_Intersects('SRID=4326;
+        POLYGON((5 56, 15.5 56, 15.5 47, 5 47, 5 56))', geom)""",
         geom_col=geom_column,
         index_col="w_id",
     )
 
 
 def federal_states_per_weather_cell():
-    """ Assings a federal state to each weather cell in Germany.
+    """Assings a federal state to each weather cell in Germany.
 
     Sets the federal state to the weather celss using the centroid.
     Weather cells at the borders whoes centroid is not inside Germany
@@ -80,9 +79,11 @@ def federal_states_per_weather_cell():
     ).index_right
 
     # Assign a federal state to each cell inside Germany
-    buffer = 200
+    buffer = 1000
 
-    while len(weather_cells[weather_cells["federal_state"].isnull()]) > 0:
+    while (buffer < 30000) & (
+        len(weather_cells[weather_cells["federal_state"].isnull()]) > 0
+    ):
 
         cells = weather_cells[weather_cells["federal_state"].isnull()]
 
@@ -99,6 +100,8 @@ def federal_states_per_weather_cell():
             .drop_duplicates(subset="w_id", keep="first")
             .set_index("w_id")
         )
+
+        weather_cells = weather_cells.dropna(axis=0, subset=["federal_state"])
 
     return weather_cells.to_crs(4326)
 
@@ -145,7 +148,7 @@ def turbine_per_weather_cell():
 
 
 def feedin_per_turbine():
-    """ Calculate feedin timeseries per turbine type and weather cell
+    """Calculate feedin timeseries per turbine type and weather cell
 
     Returns
     -------
@@ -249,7 +252,7 @@ def feedin_per_turbine():
 
 
 def wind():
-    """ Insert feed-in timeseries for wind onshore turbines to database
+    """Insert feed-in timeseries for wind onshore turbines to database
 
     Returns
     -------
@@ -304,7 +307,7 @@ def wind():
 
 
 def pv():
-    """ Insert feed-in timeseries for pv plants to database
+    """Insert feed-in timeseries for pv plants to database
 
     Returns
     -------
@@ -334,7 +337,7 @@ def pv():
 
 
 def solar_thermal():
-    """ Insert feed-in timeseries for pv plants to database
+    """Insert feed-in timeseries for pv plants to database
 
     Returns
     -------
@@ -364,7 +367,7 @@ def solar_thermal():
 
 
 def insert_feedin(data, carrier, weather_year):
-    """ Insert feedin data into database
+    """Insert feedin data into database
 
     Parameters
     ----------