Back to Blog
Technology 9 min read 30 April 2026
How We Trained Our Drought Prediction Model on 20 Years of Ethiopian Satellite Data
Our ML team explains the data pipeline: from NASA MODIS vegetation indices to localised risk scores delivered via USSD to basic mobile phones.
ME
ML Engineering
30 April 2026
Predicting drought in the Horn of Africa is not a problem of data availability — it is a problem of data fusion. This deep-dive walks through how Mobile Vet stitches NASA MODIS NDVI, CHIRPS rainfall estimates, ERA5 surface temperatures and ground-station observations into a single per-woreda risk score, and why every score that ships to a farmer has to be co-signed by a regional veterinary officer.
Enjoyed this post?
Get field reports and updates delivered to your inbox — once or twice a month.
All posts