VolcGIS is a Python package that implements several functions for exposure analyses to volcanic hazards. It is developed by the Volcanic Hazards and Risk Group group at the Earth Observatory of Singapore, Nanyang Technological University. In a nutshell, VolcGIS provides high-level functions to streamline the process of:
- Setting up a GIS to fully exploit the spatial relationship between hazard and exposure datasets
- Pre-process hazard and exposure datasets
- Extract global exposure data and figures for various hazard types and scenarios
- Implement methodologies to estimate some aspects of vulnerability, impact and risk
VolcGIS is available on Github, documentation is available at vharg.github.io/VolcGIS/. Check out the demo notebook to get started.
Create environment and set the channel to conda-forge:
conda config --env --add channels conda-forge
conda config --env --set channel_priority strict
Then:
conda install -c conda-forge pyarrow rioxarray rasterio geopandas bokeh contextily osmnx
conda install -c conda-forge holoviews datashader panel param geoviews
Install these packages with pip:
pip install utm
pip install alive-progress
To install the documentation:
pip install mkdocs-material mkdocstrings livereload mkdocs-jupyter
Anyone who has ever had a critical glance at exposure data derived from global datasets in volcanology or any other natural hazards will have noticed how variable exposure figures from different studies can be. This variability can be explained by various aspects. The first one is the variability of global datasets. For population, different datasets (e.g. GHSL, Landscan) rely on different assumptions (see Freire et al. 2019 for more detail). Similarly global land-cover maps, use different resolutions, classification algorithms and training/validation datasets, which induces a discrepancy in the results. The second one is the manipulation of geospatial data which, through such tasks as re-projection between different coordinate systems and interpolation, add a layer of noise to the data. Whereas it is important to keep these operations under as much control as possible, the same population count achieved using two different Python libraries shows a 3-5% variability in the result.
As a caveat, a critical interpretation of the results is required. Although not systematically quantified, a ±10% variability of the results is a sensible baseline.
Freire, S., Florczyk, A., Pesaresi, M., Sliuzas, R., 2019. An Improved Global Analysis of Population Distribution in Proximity to Active Volcanoes, 1975–2015. ISPRS International Journal of Geo-Information 8, 341. https://doi.org/10.3390/ijgi8080341
VolcGIS is published under a GNU GPL v3 License