Machine Learning: potential for local and regional deep-seated landslide nowcasting

Abstract

<p>Where landslide hazard mitigation is impossible, Early Warning Systems are a valuable alternative to reduce landslide risk. To this extent nowcasting and Early Warning Systems for landslide hazard have been implemented mostly at local scale. Unfortunately, such systems are often difficult to implement at regional scale or in remote areas due to dependency on local sensors. However, in recent years various studies have demonstrated the effective application of Machine Learning for deformation forecasting of slow-moving, deep-seated landslides. Machine Learning, combined with satellite Remote Sensing products offers new opportunities for both local and regional monitoring of deep-seated landslides and associated processes.</p><p>Working from the key variables of the landslide process we selected the available satellite Remote Sensing products, the necessary assumptions for a satellite only application and evaluated the potential benefit of local information. In the absence of continuous, satellite deformation measurements, nowcasting of the system state will provide a short term deformation prediction. We demonstrate the opportunities of Machine Learning on multi-sensor monitored Austrian landslide and anticipate on the integration in an Early Warning System. Furthermore, we highlight the risks and opportunities arising from the limited physics constraints in Machine Learning.</p>