From environmental data to early action.
AFYA MAZINGIRA
Hyperlocal Environmental Risk & Early Action Intelligence
AFYA MAZINGIRA transforms environmental measurements into early-action intelligence. It does not provide medical diagnosis, emergency response instructions, or clinical advice.
AFYA MAZINGIRA provides environmental decision support. It does not provide medical diagnosis, emergency-response instructions, or clinical advice.
How It Works
Conduit local measurements
Quality control (station health rules)
Feature engineering
Climate Reflex state classification
+1/+3/+6/+9h WBGT forecasting
Risk / exposure interpretation
Best-Time activity recommendation
Explainable output (EN + SW)
Data Sources & Provenance
Conduit
Local environmental measurements · JHUB Africa / JKUAT
GROUND MEASUREMENT · ~15 min
ERA5
Regional atmospheric reanalysis · ECMWF / Copernicus, through the Open-Meteo archive
REGIONAL MODEL · ~28 km · hourly · about 5 days behind
Rainfall
The station's own gauge 1, read from its daily running total, for the farm water budget; Open-Meteo's regional model for 7- and 30-day context and where the gauge is short of data
GROUND MEASUREMENT + REGIONAL MODEL · daily
Open-Meteo forecast
County outlook on the Risk Map, and plans for tomorrow
REGIONAL FORECAST · hourly
Sentinel-2
Vegetation / NDVI · ESA Copernicus. Only where the server holds Copernicus credentials: without them no NDVI is computed and the Risk Map's vegetation layer stays empty
SATELLITE-DERIVED · 10 m · ~5-day revisit · NEEDS CREDENTIALS
Sentinel-3 SLSTR
Acquisition dates only; land surface temperature is not computed · ESA Copernicus. On the same credentials: without them no acquisition is listed at all
SATELLITE-DERIVED · dates · NEEDS CREDENTIALS
Scientific Limitations
Does not provide medical diagnosis or clinical advice
Does not predict flooding at any scale: the flood page shows rainfall and soil-saturation conditions only
Does not predict malaria, cholera, or asthma attacks
One station does not represent all of Juja
Does not use generative AI as a forecasting engine
Example ScenariosILLUSTRATIVE
Illustrative users and what they could do with the app. Not real case studies.
AFYA MAZINGIRA · Hackathon build · JKUAT/Juja proof of concept · September 2026
Built with JHUB Africa and the JKUAT Conduit station