hostrada4py

hostrada4Py is an Open Source Python library with access to and evaluation of the HOSTRADA weather data for Germany from the DWD (Deutscher Wetterdienst) and CERRA weather data for the EU from Copernicus.

hostrada4py is developed by the Chair of Building Services Engineering VPT at Berlin University of the Arts.

Installation

$ git clone https://github.com/UdK-VPT/hostrada4py.git
$ cd hostrada4py
$ pip install -e .

The HOSTRADA project of DWD

The high-resolution hourly grid data set (HOSTRADA) for Germany published by DWD is a climatological reference data set. With a spatial resolution of one square kilometer and a temporal resolution of one hour, it provides a wide range of meteorological parameters for the land surfaces of the Federal Republic of Germany since 1995.

For most meteorological parameters, HOSTRADA is based on the interpolation of station data, taking into account satellite and climate model data to calculate a consistent data set. The intensity of the urban heat island (UHI) is included in the data set, allowing for improved representation of temperature distributions in regions with significant orographic features.

The hourly grid data for the HOSTRADA parameters and UHI intensity can be downloaded on a monthly basis from the Open Data section of the DWD Climate Data Center (https://opendata.dwd.de/climate_environment/CDC/grids_germany/hourly/hostrada/). It has a spatial resolution of 1 km x 1 km, covering a total area of 720 km x 938 km, and is available in NetCDF format in EPSG:3034 projection (ETRS89 / LCC Europe) and in UTC format.

The HOSTRADA data set is expanded monthly and contains the following variables:

• Cloud cover,
• Wind speed and direction (at a height of 10 m),
• Air and dew point temperature (at a height of 2 m),
• Relative humidity (at a height of 2 m),
• Air pressure at station height and sea level,
• Water vapor mixing ratio (at a height of 2 m),
• Global radiation,
• Urban Heat Island Intensity (UHI).

The CERRA project

CERRA (Copernicus European Regional ReAnalysis) provides consistent historical reconstructions of atmospheric, surface, and soil conditions across Europe. (https://climate.copernicus.eu/copernicus-regional-reanalysis-europe-cerra). It combines observations with numerical weather models to produce high-resolution data, including a 5.5-km atmospheric dataset with a time resolution of 3 hours, an ensemble dataset for uncertainty estimates, and the CERRA-Land product. Available through the Copernicus Climate Data Store, CERRA supports climate monitoring, research, impact studies, and adaptation planning (https://cds.climate.copernicus.eu/datasets/reanalysis-cerra-single-levels).

The CERRA data set is also expanded monthly and can deliver the same weather data as HOSTRADA except Urban Heat Island Intensity (UHI) since 1984.

To access CERRA weather data, you need an API key, which you receive after registering and must store in a configuration file (see https://cds.climate.copernicus.eu/how-to-api).

The Python library hostrada4py

The folder hostrada4py includes some Python files which simplifies the access to HOSTRADA and CERRA weather data. hostrada.py contains functions for downloading the desired NetCFD files from the DWD server. The functions contained in hostradaPoint.py support the evaluation of weather data at a climate location specified by longitude and latitude, whereas hostradaArea.py provides functions that enable the evaluation of weather data in a grid with a resolution of 1km x 1km, which is defined by a polygon.

For the largest 50 German cities and a couple of regions polygons are predefined in the files hostrada4py/hostradaCities.py and hostrada4py/hostradaRegions.py.

Notebooks

The notebooks

• hostradaPoint.ipynb (time series of weather data for one location),
• hostradaPoint-diffRad.ipynb (calculation of the diffuse radiation based of HOSTRADA values for one location),
• hostradaRoute.ipynb (calculation of HOSTRADA values along a predefined route),
• hostradaArea.ipynb (2D fields of weather data),
• hostradaAreaMean.ipynb (mean values of 2D fields of weather data),
• hostradaHeatPeriods.ipynb (shows the locations with the most intensive heat periods in Germany (HOSTRADA, within 1995 to 2026) and in the EU region (CERRA, 1984 to 2026)),
• hostradaHeatingDegreeDays.ipynb (calculation of heating degree days based of HOSTRADA values for one location) and
• hostradaGenerateWeatherFiles.ipynb (generation of weather data files based of HOSTRADA values for one location for different simulation programs - IDA ICE, Polysun, EnergyPlus, SimStadt and the Modelica library BuildingSystems).

illustrate the use of the Python functions of hostrada4py.

Diffuse horizontal irradiance (DHI)

The HOSTRADA and CERRA datasets contain values for total radiation, but not for the diffuse radiation included in it.

A robust DHI estimate can be calculated directly from HOSTRADA point data with pvlib and the erbs_driesse decomposition model. The function extract_diffuse_radiation_for_point downloads the required HOSTRADA variables for a point, combines them, computes the solar position from time, longitude, and latitude, and returns dhi, dni, and kd.

from hostrada4py import extract_diffuse_radiation_for_point

df = extract_diffuse_radiation_for_point(
    lon=13.405,
    lat=52.52,
    start="2024-01-01T00:00:00",
    end="2024-01-02T23:00:00",
    tz="Europe/Berlin",
)

print(df[["rsds", "dhi", "dni", "kd"]].head())

If additional HOSTRADA variables are available, a conservative weather-based correction can optionally be enabled with apply_weather_correction=True.

Optional NetCDF subsetting cache

For long weather-file exports hostrada4py can reduce the amount of data that is read and kept locally by creating spatial/temporal NetCDF subsets for a point or polygon request. The default behaviour caches the full monthly HOSTRADA files.

Enable the subset mode from Python before starting an export:

import os
os.environ["HOSTRADA_NETCDF_SUBSET_MODE"] = "subset"
os.environ["HOSTRADA_NETCDF_SUBSET_MARGIN_CELLS"] = "1"

or use it directly with the point/area helpers:

from hostrada4py.hostradaPoint import extract_values_for_point

extract_values_for_point(
    "tas",
    lon=13.405,
    lat=52.52,
    start="2023-01-01 00:00",
    end="2023-01-31 23:00",
    cache_strategy="subset",
)

Available modes:

• full: original full-file download and cache (default)
• subset: full download if needed, then write and reuse a small local subset
• http_range: best-effort HTTP byte-range/chunked access, then local subset
• auto: try http_range, then fall back to subset

The DWD OpenData HOSTRADA endpoint is a static file service. A true server-side NetCDF subset is only possible when the remote endpoint and local optional packages support chunked/range access or an OPeNDAP/THREDDS/NCSS-like service. Therefore http_range is intentionally optional and falls back to the robust full download unless HOSTRADA_NETCDF_SUBSET_FALLBACK=0 is set.

If disk space is more important than keeping the original full cache, the full monthly file can be removed after the subset file has been created:

export HOSTRADA_DROP_FULL_AFTER_SUBSET=1