Skip to map
← back to the map

Methodology & data sources

Vegvisir is a visualization and decision-support layer. Official warnings always come from IMD, NDMA and the state disaster management authorities — treat their advisories as authoritative.

Composite risk index (v1)

Each district's score (0–100) is a transparent weighted overlay, recomputed every 15 minutes from live rainfall:

risk = 0.40 · rain        (observed 24 h + forecast 48 h, ÷250 mm)
     + 0.25 · landslide   (NRSC atlas susceptibility × rain amplification)
     + 0.25 · flood       (historical proneness × rain amplification)
     + 0.10 · exposure    (population density ÷ 1600 /km²)

amplification = 0.25 + 0.75 · rain-normalized

Terrain factors are rain-amplified because a susceptible slope is chiefly dangerous while it is raining. Severity bands: 0–24 Normal, 25–44 Watch, 45–64 Alert, 65+ Severe. Weights are published so the model can be audited and challenged; calibration against the 2018/2019 Kerala flood record is on the roadmap before any operational use.

Data tiers

  • LIVE — streamed from public feeds through cached API routes: rainfall (Open-Meteo multi-model), river discharge (GloFAS/Copernicus via the Open-Meteo Flood API, sampled at CWC station sites), hazard events (GDACS), earthquakes (USGS), satellite imagery (NASA GIBS / VIIRS).
  • SAMPLE — real facility locations with a curated, illustrative list: hospitals & shelters (pending OSM/KSDMA extraction).

River status thresholds

Each station is coloured by today's modelled discharge relative to its own trailing 31-day median: normal below 1.5×, elevated from 1.5×, and high from 3×. This anomaly approach is self-calibrating per river but is not a substitute for CWC's official warning/danger levels, which are the planned replacement.

Static factors

Landslide susceptibility per district is derived from the NRSC/ISRO Landslide Atlas of India hotspot rankings (the Western Ghats account for ~14.7% of national exposure). Flood proneness reflects the 2018/2019 flood footprints and each district's share of low-lying terrain. Population density is Census-derived (people/km²).

Production integration targets

NDMA SACHET CAP alert feed · IMD official API (api.imd.gov.in) · CWC flood-forecast river levels · INCOIS ocean-state & high-wave alerts · NRSC Bhuvan WMS hazard zonation layers · WorldPop gridded population. The ingestion architecture treats every upstream source as unreliable: adapters validate, normalize and cache, and the UI surfaces per-layer freshness.

Every source

This list and the in-app “Data sources” panel are generated from a single registry, so provenance is described identically everywhere.