Adopting a hydrometeorological approach for territorial landslide early warning: insights and effectiveness evaluation from a case study in Campania (Italy)

The Campania region (southern Italy) is characterized by a widespread hydrogeological risk due to the presence of many slopes covered by pyroclastic deposits derived from the activity of the Vesuvius and Campi Flegrei volcanic complexes. Indeed, numerous areas in the region are prone to rainfall-induced landslides—particularly shallow landslides and debris flows—which are often triggered by short-duration, high-intensity precipitation events. Over the years, the region has been affected by severe landslide events, causing loss of lives and economic damage. In Campania, a territorial landslide early warning system (Te-LEWS) has been operational since 2005 managed by the regional department for Civil Protection. For early warning purposes, the regional territory is divided into eight distinct warning zones according to the following factors: hydrography, morphology, climate, geology, land use and administrative boundaries. The predictions of a weather numerical model are used to evaluate the possible occurrence of rainfall-induced landslides within each warning zone. The daily assessment of the criticality is established by comparing the weather forecasts to a set of thresholds associated with rainfall precursors.

In the scientific literature it is widely recognized that rainfall primarily acts as a triggering mechanism, while hydrological variables (e.g., soil moisture) control slope predisposition to failure. Therefore, an evaluation that neglects antecedent hydrological conditions may result in a high number of false alarms, limiting the reliability and credibility of rainfall-only warning models. In recent years, a growing number of weather and hydrological reanalysis products have been produced at fine temporal and spatial resolutions, allowing the potential use of soil moisture data into operational Te-LEWS. This study proposes a hydrometeorological approach integrating meteorological and hydrological information and testing its performance in a landslide-susceptible area of the Campania region, southern Italy. A two-dimensional Bayesian analysis is employed to quantify the conditional probability of landslide occurrence and to derive multiple hydro-meteorological thresholds associated with increasing warning levels. The performances of the warning models are assessed by means of statistical indicators to identify the best-performing combination of hydro-meteorological thresholds. Finally, the potential added value of incorporating soil moisture into territorial landslide warning models is assessed by comparing the hydro-meteorological model developed in this study with the current regional warning system in a real-case scenario.