AI Overview
AI-generated explanation of the current landslide risk
Area Search
Draw or upload an area to analyse what it covers
Map Search
Find a road, Taman or building on the map
Search imported OpenStreetMap data. Tailscale connection required.
Risk Prediction (next 6 hours)
Predicted landslide risk level over time
Slopes at this station
Risk score calculation
Each slope starts from a baseline set by its factor of safety (FoS). Forecast rainfall over the last three days then raises it. The station score is the average of its slopes.
| Step | Formula |
|---|---|
| 1. Slope baseline | B = clamp(95 − 45 × FoS, 5, 95) |
| 2. Weighted forecast rainfall | W = 0.6 × FD−2 + 0.8 × FD−1 + 1.0 × FD |
| 3. Rainfall uplift | U = clamp((W − 10) / 12, 0, 1) × 60 |
| 4. Slope risk | R = min(B + U, 98) |
| 5. Station risk score | Score = mean R of the station's slopes / 100 |
FoS: factor of safety of each slope (slope.csv). Below 1.0 the slope is unstable, so a lower FoS gives a higher baseline.
Fd: the model's forecast rainfall for the station on day d (mm), as shown in Rainfall Forecast. D is the selected date; earlier days count less.
U: no uplift while W is below 10 mm; it rises by 5 points per mm and reaches the full 60 points at 22 mm. A station without forecast data gets no uplift.
Levels: 0.70 and above is High, 0.45–0.69 Moderate, below 0.45 Low.
Population Impact
Estimated age-group population in the selected DUN
Population breakdown
Click a DUN on the map to view its population impact.
Rainfall Observation
Select a station on the map
Nodes 5 of 5
Loading rainfall readings…
Anomaly · Hover to highlight a node. Click a reading to inspect it. Missing readings are gaps.
Anomaly details
Date: —Node readings, anomaly reasons and confidence scores are for demonstration. Nearby median compares the other nodes at this station.
Rainfall Forecast
| Measure | Selected date |
|---|---|
| Forecast rainfall | — |
| Actual rainfall | — |
| Forecast error | — |
Summary:
Forecast error = forecast − actual. Positive: forecast was higher than observed; negative: forecast was lower than observed.
Mean absolute error = average of |forecast − actual| over the month.