Demand downscaling

[giacfalk]

I suggest considering additional factors (=data) to downscale national demands to multi-note demand.

E.g.

• urbanisation level -> https://ghsl.jrc.ec.europa.eu/ghs_smod2019.php
• remoteness -> https://malariaatlas.org/research-project/accessibility-to-healthcare/
• presence of roads -> https://www.globio.info/download-grip-dataset
• healthcare facilities -> https://healthsites.io/
• building data -> https://sites.research.google/open-buildings/
• ...

Happy to contribute to the downscaling coding work.

Cheers, Giacomo

[pz-max]

Great could be used also as learning input data of the subnational GEGIS approach.

1. machine learning approach on country level
2. statistical downscaling

We need to put raster data into national shapes. Here is the script to build national shapes:

• https://github.com/pypsa-meets-africa/pypsa-africa/blob/new-main/scripts/build_shapes.py

[giacfalk]

For downscaling (usually done for climate data), we can follow this existing methods:

https://github.com/pangeo-data/scikit-downscale https://openresearchsoftware.metajnl.com/articles/10.5334/jors.197/

[giacfalk]

Reccomended efficient library to extract raster data into polygons:

https://pythonhosted.org/rasterstats/

[giacfalk]

This feature is in active development and updates will soon follow.