[D] Geospatial ML for humanitarian drought/flood forecasting: critique my approach / ideas for predictive urgency index

About this article

I'm working on a non-commercial geospatial ML project (AidMap AI) focused on Central Asia/Afghanistan/Syria – predicting "urgency levels" for slow-onset ecological crises (droughts, floods, crop failure, hunger) using open data. Core idea: aggregate multi-source data build a predictive model that outputs a composite "surgency score" (e.g., regression or multi-label classification) for anticipatory humanitarian action. Current rough approach: Data fusion: raster + tabular (e.g., point location...