Historic comparison PV and wind (Germany)¶

In this example we are examining the ranalysed power feed-ins calculated by atlite. Based on data for capacity distributions and weather during the year 2013, we try to match historical statistics.

Importing packages¶

[1]:

import atlite
import xarray as xr
import pandas as pd
import scipy.sparse as sp
import numpy as np

import pgeocode
from collections import OrderedDict

[2]:

import matplotlib.pyplot as plt
%matplotlib inline

import seaborn as sns
sns.set_style('whitegrid')

Download Necessary Data Files¶

We need to download the locations of all the PV installations in Germany to later tell atlite where to setup the PV panels and with which capacity. This information is available in Germany (thanks to the EEG feed-in-tariffs in the so-called “Anlagenregister”).
We also download a reference time-series to compare our results against later. We retrieve the data from https://open-power-system-data.org which in return gets it from ENTSO-E.
Finally we also download a cutout of weather data from the ERA5 dataset containing Germany and the year we want to examine (2012).

[3]:

import requests
import os
import zipfile

def download_file(url, local_filename):
    # variant of http://stackoverflow.com/a/16696317
    if not os.path.exists(local_filename):
        r = requests.get(url, stream=True)
        with open(local_filename, 'wb') as f:
            for chunk in r.iter_content(chunk_size=1024):
                if chunk:
                    f.write(chunk)
    return local_filename

Reference time-series¶

The reference is downloaded from Open Power System Data (OPSD) and contains data reported by the TSOs and DSOs.

[4]:

opsd_fn = download_file('https://data.open-power-system-data.org/index.php?package=time_series&version=2019-06-05&action=customDownload&resource=3&filter%5B_contentfilter_cet_cest_timestamp%5D%5Bfrom%5D=2012-01-01&filter%5B_contentfilter_cet_cest_timestamp%5D%5Bto%5D=2013-05-01&filter%5BRegion%5D%5B%5D=DE&filter%5BVariable%5D%5B%5D=solar_generation_actual&filter%5BVariable%5D%5B%5D=wind_generation_actual&downloadCSV=Download+CSV',
                        'time_series_60min_singleindex_filtered.csv')

[5]:

opsd = pd.read_csv(opsd_fn, parse_dates=True, index_col=0)

# we later use the (in current version) timezone unaware datetime64
# to work together with this format, we have to remove the timezone
# timezone information. We are working with UTC everywhere.

opsd.index = opsd.index.tz_convert(None)

# We are only interested in the 2012 data
opsd = opsd[("2011" < opsd.index) & (opsd.index < "2013")]

PV locations (“Anlagenregister”)¶

Download and unzip the archive containing all reported PV installations in Germany in 2011 from energymap.info.

[6]:

eeg_fn = download_file('http://www.energymap.info/download/eeg_anlagenregister_2015.08.utf8.csv.zip',
                        'eeg_anlagenregister_2015.08.utf8.csv.zip')

with zipfile.ZipFile(eeg_fn, "r") as zip_ref:
    zip_ref.extract("eeg_anlagenregister_2015.08.utf8.csv")

Create a Cutout from ERA5¶

Load the country shape for Germany and determine its geographic bounds for downloading the appropriate cutout from ECMWF’s ERA5 data set.

[7]:

import cartopy.io.shapereader as shpreader
import geopandas as gpd
shp = shpreader.Reader(shpreader.natural_earth(resolution='10m', category='cultural', name='admin_0_countries'))
de_record = list(filter(lambda c: c.attributes['ISO_A2'] == 'DE', shp.records()))[0]
de = gpd.GeoSeries({**de_record.attributes, 'geometry':de_record.geometry})
x1, y1, x2, y2 = de['geometry'].bounds

/home/fabian/.miniconda3/lib/python3.7/site-packages/ipykernel_launcher.py:5: FutureWarning:     You are passing non-geometry data to the GeoSeries constructor. Currently,
    it falls back to returning a pandas Series. But in the future, we will start
    to raise a TypeError instead.
  """

[8]:

cutout = atlite.Cutout('germany-2012',
                       module='era5',
                       x=slice(x1-.2,x2+.2), y=slice(y1-.2, y2+.2),
                       chunks={'time':100},
                       time="2012")

/home/fabian/vres/py/atlite/atlite/cutout.py:160: UserWarning: Arguments module, x, y, time are ignored, since cutout is already built.
  warn(f'Arguments {", ".join(cutoutparams)} are ignored, since '

[9]:

cutout.prepare()

[9]:

<Cutout "germany-2012">
 x = 5.75 ⟷ 15.00, dx = 0.25
 y = 47.25 ⟷ 55.25, dy = 0.25
 time = 2012-01-01 ⟷ 2012-12-31, dt = H
 module = era5
 prepared_features = ['height', 'wind', 'influx', 'temperature', 'runoff']

Downloading the cutout can take a few seconds or even an hour, depending on your internet connection and whether the dataset was recently requested from the data set provider (and is thus cached on their premise). For us this took ~2 minutes the first time. Preparing it again (a second time) is snappy (for whatever reason you would want to download the same cutout twice).