An efficient Bayesian method for estimating runout distance of region-specific landslides using sparse data

The runout distance of landslides is a critical factor that influences landslide risk quantification and mitigation designs. Nevertheless, empirical correlation models developed in a region may be biased when it is applied to landslides in another region, partly due to different geological formation process undergone in different regions. Besides, the development of a reliable correlation model for estimating runout distance generally requires a relatively large number of data points, collection of which is difficult and time-consuming. This study proposes a Bayesian method for efficiently developing region-specific correlation models to estimate runout distance. The method integrates systematically sparse data collected in a specific region and prior knowledge embedded in existing correlation models in other regions. The method is illustrated using both real and numerical examples. The results show that the proposed method is efficient and robust in developing region-specific correlation models for estimating runout distance.