Drought Hazard, Vulnerability and Risk Assessment for Kenya

This dataset provides monthly, ~1 km drought hazard probabilities and risk maps for Kenya, plus a static vulnerability layer. Hazard is modelled using a logistic regression based on monthly anomalies of NDVI, NDII, LST (from MODIS and Sentinel-3) and 3-monthly SPI (from TAMSAT) while training periods (years) where determined using yearly FAOSTAT yield data. The MODIS and Sentinel-3 indices were cross-calibrated to form a consistent time series spanning from 2001 to the present. Predictors were aggregated for cropland, grassland, and shrubland, and applied for the main growing-season months (April to July and November to December - model outputs outside the growing season to be used with care). 
Vulnerability is a relative, spatially explicit index combining the fraction of irrigated land, GDP, population density, a Relative Wealth Index, and grazing-animal density; urban and adjacent high-density pixels are excluded. Risk is computed per month as the product of hazard and vulnerability. All rasters are provided as GeoTIFFs (values 0–1). Due to data availability issues drought vulnerability and risk have to be used with care. 
Key limitations: Products are relative (not absolute), vulnerability is static (updateable periodically), and vegetation anomalies may reflect non-drought stressors; interpret multi-month patterns to assess drought evolution.

The methodology was based on: Schwarz, M., Landmann, T., Cornish, N., Wetzel, K.-F., Siebert, S., & Franke, J. (2020). A Spatially Transferable Drought Hazard and Drought Risk Modeling Approach Based on Remote Sensing Data. Remote Sensing, 12(2), 237. doi.org

You can find additional information and detailed descriptions on the projects website (admkenya.eu).