India-wide landslide intelligence

SlopeSense reads the slope before it breaks.

A research prototype. The console shows timestamped weather-model snapshots from a public provider, labeled with source and age. There are no validated landslide risk scores and no official warnings.

Risk console
Loading provider snapshots
Real provider values
Weather snapshots

Loading map

Select a point from the map or the feed.

Points reporting
0

Representative points, not full district coverage

Newest snapshot
-

Time since the latest provider fetch

Highest 72 h rain
-

No data

Validated warnings
0

No validated risk model, no official warnings

How it works

From a satellite pass to a phone in a village

The intended six-hour pipeline has five steps. Pick a step to see what happens inside it.

Step 1 of 5
Data ingestion

Proposed public data inputs

No new hardware on the hill

Rainfall, soil moisture, slope and forecast signals already exist in public data. Today the site stores and shows one provider's weather model values. Fusing the satellite feeds into a validated score is planned, not done.

6feeds

Free satellite and model feeds

Planned inputs: GPM, SMAP, Copernicus DEM, Sentinel-2, GFS and static terrain layers. Today the site shows only Open-Meteo model values.

1km²

Planned grid resolution

Design target: score cell by cell, then roll up to blocks so the answer matches an administrative unit. Not running yet.

1h

Snapshot cycle

Provider snapshots are fetched about hourly and shown with their fetch time. Check system status for freshness.

Evidence, not claims

Illustrative event replays, in the open

Retrospective · Wayanad 2024

Synthetic replay, not a real warning

The bundled Wayanad example uses synthetic inputs and illustrative scores. It does not establish predictive accuracy or a warning delivered before the event.

A villager reading an alert on a phone
Architecture

CAP v1.2 native

Alerts are published as CAP XML and JSON through a REST API, so any system that speaks the Common Alerting Protocol can consume them.

Model FPI search

Check any coordinate

Request the model index for a coordinate. Unavailable real inputs may return no score. FPI is not a calibrated probability.