Influence of topography and forest characteristics on snow distributions in a forested catchment

Abstract Stored water within snowpack is important for the hydrological balance in many mountainous environments around the world. However, monitoring the spatial and temporal dynamics of snow in such mountainous environments remains rather challenging. We therefore developed a snow depth meter using small temperature loggers. Small temperature loggers were attached to poles at 20 cm intervals from the ground surface. Snow depths were estimated by assessing the daily variations in temperatures. Using this snow depth meter, we continuously observed snow depths at 21 stations in a forested catchment in Japan over three winter seasons. Using correlation analysis, we then analyzed the influence of topography (i.e., elevation and aspect) and forest (i.e., canopy openness) on snow depths. Moreover, we estimated daily snow distributions in the area using multi-regression analysis, thus describing seasonal characteristics of snow distributions. Finally we investigated the relation between number of stations and estimation accuracies of snow distributions using a Monte Carlo sensitivity analysis. We observed that the influence of topographical and forest characteristics changed considerably during the study period, with elevation having a major impact on snow depths. Further, aspect and forest cover had a great influence on the snow depths during the melting period. The regression of elevation slopes was 0.8–2.1 mm/m during rich snow years and 0.5–0.6 mm/m in little snow years. Also, the snow distribution during the melting period was found to be less uniform than during the snow accumulation period using histograms of snow depths. Monte Carlo sensitivity analysis shows that one station per 2.0–2.5 ha is enough to estimate accurate snow distributions. Given the above, we concluded that our proposed approach was quite useful for investigating the influence of topography and forest characteristics on snow accumulation and melting.