The PyBathHail repository contains Python code and example files for retrieving the hail size distribution (HSD) and estimating the vertical wind speed from Doppler spectra recorded with DWD's operational C-band radar (vertically pointing) birdbath scan, as described by Gergely et al. (2025).
Distributed under the MIT License. See LICENSE for more information.
Experimental code. To preserve flexibility, download the full PyBathHail repository (latest release), unzip, and run the hail retrieval from this directory. Also needs pre-installed pybathspectra package or downloaded pybathspectra directory copied into working directory.
Exact version numbers of the Python packages used for developing pybathhail modules are listed in setup.py. Other versions, especially much newer versions, can cause compatibility issues, particularly due to the many changes in NumPy and pandas between different version numbers. Therefore, it may be best to create a virtual environment with the exact versions of Python packages listed in 'setup.py' and then run the PyBathHail postprocessing analysis inside this Python environment.
The pybathhail Python package can be installed by downloading the PyBathHail repository and running pip install . (after installing pybathspectra and other requirements), for example. But to preserve flexibility and avoid (some) compatibility issues, it is best to run the retrieval inside the downloaded repository structure, i.e., without prior installation.
All retrieval and analysis methods are collected in the pybathhail directory. The retrieval can be executed by running hail_retrieval.py with the desired retrieval settings selected in hail_config.yaml (and radar signal postprocessing settings in postprocessing_config.yaml, if needed/desired).
The input directory is a collection of birdbath-scan example data for 3 hailstorms, pre-calculated hail radar backscatter cross sections needed for the HSD retrieval, optional ICON atmospheric model outputs for 2 hailstorms at MHP radar (only needed if it is desired to alternatively anchor the slow hail edge to the maximum rain velocity (shift_type: max_rain) instead of the default option (shift_type: min_hail_...) in hail_config.yaml), and the pre-determined hail analysis heights (with the pybathspectra toolkit); the output folder collects all numerical results as well as plots of some results (results both for the hail retrievals of pybathhail and for the radar signal postprocessing steps with pybathspectra).
3 birdbath-scan examples are included as input: 2 at MHP radar in southern Germany, 1 at FLD radar in central Germany (see cited article for details).
Gergely, M., Ockenfuß, P., Seeger, F., Kneifel, S., Frech, M., 2025(?): Retrieval of the hail size distribution and vertical air motion from weather radar Doppler spectra, J. Atmos. Oceanic Technol., submitted March 2025, LINK_GOES_HERE
The work is supported by the German Research Foundation (DFG) 'PROM' priority program SPP-2115 (https://www2.meteo.uni-bonn.de/spp2115) and the German Meteorological Service (Deutscher Wetterdienst, DWD, https://www.dwd.de/DE/Home/home_node.html).