Development of a Method for Estimation of Volume of Snowmelt from Snowpack

Often snowmelt occurs in fair weather or rising temperatures; however, the Japan Meteorological Agency (JMA) only provides snow warning information per day, not per shorter intervals, as is required for train operation. It is generally considered difficult to obtain hourly quantitative evaluation of the volume of snowmelt from snowpack or due to rainfall. For this reason, the authors are developing a method to estimate the volume of snowmelt from snowpack, in order to secure safe train operation against slope failure during the snowmelt period. A time-series estimation of the volume of snowmelt from snowpack in different theoretical approaches was attempted and four estimation models were developed, based on observations near Oshirakawa Station on the Tadami Line. In relation to measurement values, estimate error was a root-mean-square error (RMSE) of 0.66 mm/h in a physical model and RMSE of 0.173 mm/10 min. in a statistical model. For the daytime snowmelt intensity of 4-6 mm/h observed at the period of greatest snowfall, that translates to around a 10% error in the model estimation.