An early warning criterion for the prediction of snowmelt-induced soil slope failures in seasonally cold regions

In Hokkaido, Japan, soil slope failures occur frequently during the snow melting season. These slope failures are triggered by the excess amount of water derived from snowmelt and rainfall. For the prediction of snowmelt-induced soil slope failures in seasonally cold regions, an early warning criterion is required. The existing Japanese early warning criteria for sediment disasters, i.e., the relationship between the 60-min cumulative rainfall and the Soil Water Index (SWI), the effective rainfall index etc., consider the influence of rainfall and the time-dependent random moisture of the soil. However, these criteria do not consider the soil moisture contributed by the snowmelt water. In this study, therefore, the applicability of the existing early warning criteria to predict snowmelt-induced soil slope failures is examined. An empirical method to quantify the amount of snowmelt water is presented. Various scenarios of conceptual soil slope failures are studied using numerical simulations under different magnitudes of rainfall and snowmelt water. As a result, a revision is introduced for the SWI and the effective rainfall index, adding the amount of snowmelt water to that of rainfall, and slope failure scenarios are studied. Based on the results, a new early warning criterion, the Effective Precipitation (EP) index, is introduced. It is found that the new failure criterion performs well for the prediction of snowmelt-induced soil slope failures.