REAL-TIME ESTIMATION OF GROUND SNOW LOAD USING SNOW LAYER MODEL

Abstract

 We proposed a real-time estimation model based on the method of Takahashi model to understand the snow weight changes during every moment of snow season. This method uses Takahashi model, which is made up from three sub models, snowfall model, compaction model, and snow melting model. In addition, the hourly snow layer model and optimized parameter for each location have been newly introduced to real-time estimation.

 In hourly snow layer model, snow layer is formed by weather observation records obtained hourly from AMeDAS. The density at the hourly layer formation was calculated by the snowfall model. In addition, since it was assumed that the hourly layer was transitioning to the daily snow layer, only the infiltration of snowmelt water and rainfall was considered, while consolidation was not.

 The optimized parameters for each location are average values of several years of pre-calculation result of each location. However, in the analysis using these parameters, when the value is significantly different from the average year, the error might be large. In that case, if overestimation was occurred, consolidation was forcibly performed, and if underestimation occurs, the capture rate was corrected.

 The estimation accuracy was verified using this model. This paper covered Tokamachi, Shinjo, and Sapporo. The snow weight data used to compare calculation result were the observed record in Tokamachi and Shinjo, and for the Sapporo’s case, the value was calculated from the record of the cross-section observation, which was done twice a week. Also, the comparison of snow weight in the verification was based on daily data.

 In Tokamachi and Shinjo, the estimation results were sometimes low in the winter with little snow, they were almost accurate in the winter with a lot of snow. In Sapporo, although they were accurate in some winters, the error of the maximum snow weight was over 30% in winters with a lot of snow.

 In Sapporo in the winter of 2012/13, the obtained value by cross-section observation might be inaccurate and the snow weight might be less than the true value. On the other hand, the increase in snow weight due to rainfall from mid-February might be indication that the estimated layer is overly absorbing precipitation. Therefore, the actual snow conditions of the snow layer from mid-February to March is possibly changing to granular snow earlier than our estimation due to consolidation and precipitation infiltration, rainfall. In result snowmelt water and rainfall may not be retained and flowed out from the bottom of the snow. In addition, the temperature in Sapporo is extremely low in winter, so there are many days when the temperature drops below －2℃, which is set as the minimum snowmelt temperature. At this time, snow melting does not occur in the analysis, but there is a possibility that snow melting might occur due to solar radiation. If snow weight measurements are accurate, one of the factors might lead to overestimation since these mechanisms were not considered in our estimation

 Furthermore, there was a gap between the estimated values and measured values when the snow melted completely in the winter of 2012/13 in Sapporo. Since the gap was caused by the fact that amount of estimated snowmelt in early spring might be smaller than the actual one, it may be necessary to use different parameters depending on the season.