Landslide Early Warning
Improving landslide early warning with soil moisture data
Experience
Anas, the manager of over 32 thousand kilometres of roads and motorways in Italy, launched a call for innovative solutions within the Smart Road Belluno Cortina project, the first Smart Road in Italy and a key access route to the 2026 Winter Olympics. The initiative focused on identifying and monitoring factors beyond the roadway itself which, although less controlled, have a significant impact on traffic flow and safety.
Within this project, Finapp carried out a three year proof of concept (2022 to 2025) along the SS51 Alemagna road, installing CRNS probes that use cosmic rays to measure soil moisture on a large scale, in depth and in real time along critical sections of the road.
Traditional landslide early warning systems rely mainly on rainfall thresholds. However, rain alone does not fully explain slope instability: soil conditions play a critical role, and rainfall based models often generate false alarms or miss critical events.
The purpose of this collaboration was to understand whether integrating measured soil moisture into prediction models could enable more accurate forecasts, site specific early warning for known landslide areas and a better allocation of operational resources.
Use
As a first step, Finapp analysed a historical dataset of 17 soil landslides, comparing predictive models based on rainfall only with models combining rainfall and simulated soil moisture. Adding soil moisture increased prediction accuracy by about 15%, confirming that soil moisture is a key variable in landslides involving soil.
The models were then validated in the field in the Fadalto area (Belluno), characterised by debris type landslides. Finapp probes were installed and monitored over time, and during the observation period 4 landslides occurred.
Compared to models using simulated soil moisture, the model combining rainfall and Finapp soil moisture measurements delivered a 4.6% improvement in predictive performance and a 48% reduction in false alarms, a critical result for operational management.
Compared to rainfall only models, adding Finapp soil moisture improved prediction by about 5.5% and increased the true detection rate by 25%. This means more real events are correctly identified, with fewer missed landslides.
These findings directly informed the development of Fin4Earth, Finapp’s web based platform for the early warning of landslides, wildfires and floods. Fin4Earth integrates real soil moisture data from Finapp probes into conventional models, defines custom risk thresholds based on each site’s characteristics and historical events, and, by combining local weather forecasts with proprietary AI algorithms, projects risk up to 72 hours in advance.
For Anas, this opens the possibility to optimise the deployment of crews and equipment, increase infrastructure safety and move towards proactive risk management.
“The work carried out by Finapp has been of high quality and demonstrated how measuring soil moisture with their probes can effectively improve early warning systems. We are now evaluating the extension of this solution to other areas across Italy.”
Benefits
- Improved landslide predictability, with about 15% higher accuracy in soil landslides
- 48% fewer false alarms compared to models based on simulated soil moisture
- 25% higher true detection rate compared to rainfall only models
- Continuous soil moisture monitoring on a large scale, in depth and in real time
- Site specific early warning tailored to each landslide area
- Risk forecasts up to 72 hours in advance with Fin4Earth
- Better allocation of crews and equipment
- Increased road safety and proactive risk management
