Probabilistic assessment of shallow landslides in a changing climate: insight from the rainfall dependence structure

Abstract Climate change is altering not only the marginal characteristics of rainfall variables but also their inter-variable dependence structure, posing significant challenges for assessing future shallow landslide risk. However, existing probabilistic assessments for shallow landslides have primarily focused on marginal changes, while the influence of rainfall dependence under a changing climate remains insufficiently quantified. This study develops a novel probabilistic shallow landslide assessment framework that explicitly accounts for the nonstationary dependence structure of rainfall variables under climate change using a nonparametric copula approach. The statistical dependence between climate-driven rainfall variables (i.e., accumulated event rainfall and rainfall duration) is modeled using a k -nearest neighbors–based conditional resampling copula capable of capturing nonlinear and tail-dependent interactions. The probability of a shallow landslide is estimated via Monte Carlo simulation by evaluating the exceedance probability of copula-simulated rainfall events over the physically based rainfall threshold for shallow landslide initiation. As an illustrative example, the proposed framework is applied to shallow slopes of completely decomposed granite in Sau Mau Ping, Hong Kong. Results show that neglecting rainfall dependence can substantially underestimate landslide probability. The increase in landslide probability under future climate conditions is primarily driven by rainfall margin changes and the existing rainfall dependence structure. This study provides new insights into the role of rainfall dependence in probabilistic landslide assessment under climate change.