GEWAS: Garissa early-warning and adaptation system
Which schools, clinics and boreholes flood first?
An automated geospatial pipeline that combines UNOSAT flood extents, Earth Engine rainfall and imagery, OpenStreetMap features and county infrastructure registers to rate how exposed schools, health facilities, boreholes, camps and towns are to flooding. It produces QGIS workspaces, Looker Studio exports, map series and a published story.
The problem
Responders could not see, in one place, which public facilities sat inside likely flood zones.
What we built
- 1
Cleaned and reprojected infrastructure layers, keeping only features inside the county boundary.
- 2
Drew concentric risk bands around historical flood extents and tagged every asset with a risk level and distance to flooding.
- 3
Mined mosques, informal settlements and points of interest from OpenStreetMap to fill gaps in official data.
- 4
Published interactive maps, a static map series and a narrative notebook, with an optional Gemini advisory report.
Highlights
- Risk bands at about 0.5, 1.5, 3.3 and 5.5 km from mapped flood extents
- Styled QGIS workspace generated automatically with PyQGIS
- CSV exports for Google Looker Studio dashboards
- Water-point and WASH survey data published only in anonymised form
- Keyless live weather from Open-Meteo
Built with
- Python
- GeoPandas
- Google Earth Engine
- PyQGIS
- Looker Studio
- Leaflet
- Gemini
Honest status: Risk zones are distance bands around historical flood extents, not hydraulic flood models.
Built by our founder for the County Government of Garissa, before TovuTech was incorporated.
Want something like this?
We design, build, host and secure systems like this end to end, and train your team to run them.
More success stories