In regions with heavy snowfall, winter and spring, snowmelt can be major triggers for deep-seated landslides. Because landslides are controlled by hydrological conditions, monitoring subsurface flow and groundwater pressure can help predict them and inform appropriate measures for disaster evacuation and mitigation. Direct monitoring of the entire region is, however, unrealistic; simple prediction methods are required instead. We examined several indices based on hydrological factors, including rainfall intensity, river water level, and reservoir inflow, to assess their correlation with landslide disasters reported from December to June 2003-2013 in Chuetsu region, central Japan. The area, underlain by Neogene mudstone and tuff, is prone to snowmelt-activated landslides; snow depths can reach up to 100 cm a year. Among the indices pro-posed, reservoir inflow and river water level were the more indicative of landslides than rainfall in this region. |