The problem
Kodagu's hills are steep, its monsoon is intense, and its land is changing — forest thins, slopes are cut for roads and building, and soil saturates fast. The 2018 and 2019 landslides showed how quickly that combination can turn deadly.
The data that could warn people — slope, rainfall, soil moisture, vegetation and forest loss — exists in public satellites, but it sits locked in formats and languages that no resident, farmer, or panchayat member can read in time.
What Nela does
Nela reads Kodagu's land from space and returns it in plain language to the people who live here. It shows a daily landslide-risk map — coloured from low to severe — with a simple Kannada explanation of why each area is rated the way it is.
Four layers sit on one map: landslide risk, forest & canopy, soil & water, and crop & coffee. It works fast, offline, and free, so it's useful in exactly the low-connectivity conditions where warnings matter most.
How it works
Every day, free public satellites (Copernicus, Sentinel, GPM, SMAP, Hansen) record slope, rainfall, soil moisture, vegetation and forest loss. A nightly Google Earth Engine job divides the taluk into H3 hexagon cells and computes numbers for each one — your phone never processes imagery.
A transparent model then rates risk as susceptibility (slope, forest loss, road-cuts) multiplied by trigger (rainfall + soil saturation), backtested against the 2018/2019 disasters. Built on open technology — Google Earth Engine, H3, Supabase/PostGIS, a Next.js PWA, and Leaflet/OpenStreetMap — Nela is open-source and decision-support: it complements official advisories, it does not guarantee outcomes. It currently covers Madikeri taluk, with the rest of Kodagu to follow.
How you can help
- Developers: improve the Earth Engine pipeline, the risk model, and the PWA
- Local validators: ground-truth landslide and land-change signals against what you see on the ground
- Translators & designers: keep the 'why' explanations clear in Kannada and Kodava thakk
- Domain experts: geologists, hydrologists, and disaster-management / forest-department partners to refine the model