Marco Uzielli (Italy) 1,2; Guido Rianna (Italy) 3; Fabio Ciervo (Italy) 3; Paola Mercogliano (Italy) 3,4; Unni K. Eidsvig (Norway) 2; Alfredo Reder (Italy) 3
1 - Georisk Engineering S.r.l.; 2 - NGI (Norwegian Geotechnical Institute); 3 - CMCC Foundation Euromediterranean Center on Climate Change; 4 - CIRA (Centro Italiano Ricerche Aerospaziali)
Pyroclastic slopes in Campania Region (Southern Italy) have been widely affected by landslide in recent times. In particular, the town of Nocera Inferiore facing the Lattari Mountains has been affected by five events (1960, 1961, 1972, 1997, 2005) since World War II. During the 2005 event, the cover reached the highway at the foot, causing three victims. Due to the high exposure of the area, a proper understanding of future variations potentially induced by climate changes in landslide occurrence represents a crucial issue for policy-makers, administrations and infrastructure operators.
This research addresses the temporal variation in landslide hazard on a section of a national motorway in Nocera Inferiore defined as the expected surface area impacted by landslides each year and estimated through anapproach relying on conditional probability.
The study defines hazard as the product of two probabilities: the probability of landslide occurrence (PL) and the reach probability (PR). Occurrence probability is, in turn, given by the product between the probability that a given rainfall scenario determines the triggering of a landslide (PT) and the probability that the same rainfall scenario occurs over a reference time period (PP). A Bayesian probabilistic approach based on rainfall thresholds is developed. Such approach is calibrated through local historical rainfall data and observed events, and returns landslide triggering probability for combinations of 1-day and 59-days cumulative precipitation. Temporal variations in Pp are assessed by exploiting the ensemble of climate projections available by EURO-CORDEX initiative under RCP4.5 and RCP8.5 up to 2100.
Projections arebias-adjusted to remove biases affecting simulations. Descriptive statistics are obtained from ensemble models for the purpose of assessing uncertainties in climate projections. Reach probability parameterizes the likelihood that, once a landslide has been triggered, it reaches a given location within the study area. This probability is calculated through a simulation-based numerical analysis of landslide runout in a GIS environment , and is defined as the frequentist ratio between the number of simulations for which each cell is reached and the total number of runout simulations.
Significant increases in landslide hazard result from the investigations. The temporal increase in landslide hazard brings potentially heavy impacts for the infrastructures located at the toe of the Nocera slopes. The magnitudes of such increases are mainly related to the time horizon and the severity of concentration scenarios. The framework provides useful insights into optimal risk mitigation strategies and decision-making regarding the design of effective protection measures.