TerraAlert is designed to use AI to analyze mountain sensor data, alongside weather forecasts and terrain. By connecting rainfall, soil moisture and ground movement, it aims to reveal patterns of landslide and flash-flood risk and help communities make better-informed decisions.
Heavy rain alone does not explain every landslide or flash flood. The challenge is understanding how multiple signals change together, in the context of each mountain and river basin.
Rainfall, soil moisture, ground movement and river levels each reveal a different part of the picture.
The same reading can mean different things depending on local terrain, recent rainfall and ground conditions.
Changes over time and unusual combinations of signals can be more informative than a single measurement.
Bring together rainfall, soil moisture, ground tilt, vibration and river-level measurements. Check missing readings and timestamps before analysis.
Compare changes over time, connect signals across sensors and interpret them alongside weather forecasts and local terrain to assess emerging risks.
Summarize the evidence behind each risk level, highlight uncertainty and help local teams review the findings before issuing warnings.
The proposed analysis combines sensor time series, weather and terrain to identify unusual changes and estimate how risk is evolving. It is designed to explain which signals matter and why, rather than simply flag a single threshold.
Bring sensor streams, weather forecasts and terrain information into a single analysis view, with clear timestamps and source status.
Look beyond individual thresholds to find combinations of rainfall, soil moisture and ground movement that may indicate an emerging risk.
Connect AI-assisted risk levels to locations and their underlying signals, so teams can see where further assessment is needed.
Pre-mapped safe zones and the fastest walking route from your location — works offline with downloaded terrain maps.
Share reviewed warnings with people in affected areas. Critical-alert behavior depends on operating-system support and user permissions.
Photos and reports from residents give local teams additional evidence to compare with sensor readings and AI findings.
Warnings in local languages and dialects, with voice read-out for elderly users and icon-based alerts for children.
After an event, mark yourself safe and instantly see the status of family members inside the affected zone.
Review AI findings, sensor trends and local reports together before deciding on warnings and coordinating a response.
Concept screens bring together sensor readings, AI-assisted risk levels and warning history. All values are illustrative, not live predictions.
Proposed solar-powered LoRa stations provide five signal types: rainfall, soil moisture, ground tilt, vibration and river level.
Track rainfall intensity and accumulation so AI can compare changing rain patterns with ground conditions.
Measure moisture at different depths to help assess how the ground responds to rainfall.
Measure changes in ground angle, adding a movement signal to the AI analysis.
Record ground vibrations for comparison with other signals when investigating unusual activity.
Track upstream water-level changes alongside rainfall to support flash-flood risk assessment.
Solar power and low-energy LoRaWAN connectivity support data collection where grid power is limited.
TerraAlert is a concept for AI-assisted landslide and flash-flood monitoring. We are exploring how sensor data and local expertise can support more informed early warnings.