Abstract:Prediction of rainfall-induced landslides is often entrusted to the definition of empirical thresholds (usually expressed in terms of rainfall intensity and duration), linking the precipitation to the triggering of landslides. However, rainfall intensity-duration thresholds do not exploit the knowledge of the hydrological processes developing in the slope, so they tend to generate false and missed alarms. Rainfall-induced shallow landslides usually occur in initially unsaturated soil covers following an increase of pore water pressure, due to the increase of soil moisture, caused by large and persistent rainfall. Clearly, it should be possible to use soil moisture for landslide prediction. Recently, Bogaard & Greco (2018) proposed the cause-trigger conceptual framework to develop hydro-meteorological thresholds that combine the antecedent causal factors and the actual trigger connected with landslide initiation. In fact, in some regions where rainfall-induced shallow landslides are particularly dangerous and pose a serious risk to people and infrastructures, the antecedent saturation is the predisposing factor, while the actual landslide triggering is associated with the hydrologic response to the recent and incoming precipitation. In fact, numerous studies already tried to introduce, directly or with models, the effects of antecedent soil moisture content in the empirical thresholds for improving landslide forecasting. Soil moisture can be measured locally, by a range of on-site measurement techniques, or remotely, from satellites or airborne. On-site measurements have proved promising in improving the performance of thresholds for landslide early warning. On-site data are accurate but sparse, so there is an increasing interest on the possible use of remotely sensed data. And in fact, recent research has shown that they can provide useful information for landslide prediction at regional scale, despite their coarse resolution and inherent uncertainty.However, while remote sensing techniques provide near-surface (5cm depth) soil moisture estimate, the depth involved in shallow landslide is typically 1-2m below the surface. This depth, overlapping with the root penetration zone, is influenced by antecedent precipitation, soil texture, vegetation and, so, it is very difficult to find a clear relationship with near-surface soil moisture. Many studies have been conducted to provide root-zone soil moisture through physically-based approaches and data driven methods, data assimilation schemes, and satellite information.In this study, the question if soil moisture information derived from current or future satellite products can improve landslide hazard prediction, and to what extent, is investigated. Hereto, real-world landslide and hydrology information, from two sites of Southern Italy characterized by frequent shallow landslides (Peloritani mountains, in Sicily, and Partenio mountains, in Campania), is analyzed. To get datasets long enough to carry out statistical analyses, synthetic time series of rainfall and soil cover response have been generated, with the application of a stochastic rainfall model and a physically based infiltration model, for both the sites. Near-surface and root-zone soil moisture have been tested, accounting also for effects of uncertainty and of coarse spatial and temporal resolution of measurements. The obtained results show that, in all cases, soil moisture information allows building hydro-meteorological thresholds for landslide prediction, significantly outperforming the currently adopted purely meteorological thresholds.&#160;&#160;

Independent demonstration of a deep-learning system for rainfall-induced landslide forecasting in Italy

Deep learning forecast of rainfall-induced shallow landslides

Spatio-temporal probability prediction of rainfall-induced landslides based on deep learning under climate change

Landslide displacement forecasting using deep learning and monitoring data across selected sites

Optimization of rainfall thresholds for landslide early warning through false alarm reduction and a multi-source validation

Towards landslide space-time forecasting through machine learning: the influence of rainfall parameters and model setting

Instrumental Monitoring of a Slow-Moving Landslide in Piedmont (Northwest Italy) for the Definition of Rainfall Thresholds

Deep learning models to predict flood events in fast-flowing watersheds

A data-driven approach to assess the role of the groundwater conditions in triggering shallow landslides initiating with frictional failure

Estimating the Rainfall Threshold of a Deep-Seated Landslide by Integrating Models for Predicting the Groundwater Level and Stability Analysis of the Slope

Landslide susceptibility mapping using ensemble machine learning methods: a case study in Lombardy, Northern Italy

Definition and performance of a threshold-based regional early warning model for rainfall-induced landslides

Regional-scale spatiotemporal landslide probability assessment through machine learning and potential applications for operational warning systems: a case study in Kvam (Norway)

Hydro-meteorological thresholds based on synthetic dataset for improved prediction of rainfall-induced shallow landslides.

Shallow landslides predisposing and triggering factors in developing a regional early warning system

Extraction of Spatiotemporal Information of Rainfall-Induced Landslides from Remote Sensing

Image analysis and LSTM methods for forecasting surficial displacements of a landslide triggered by snowfall and rainfall

Comparative analysis of conventional and machine learning techniques for rainfall threshold evaluation under complex geological conditions

Landslide detection by deep learning of non-nadiral and crowdsourced optical images

Integrating multiple information sources for landslide hazard assessment: the case of Italy

A New Deep-Learning-Based Approach for Earthquake-Triggered Landslide Detection From Single-Temporal RapidEye Satellite Imagery