ASSIMILATION OF SNOW DEPTH DATE INTO SNOWMELT-RUNOFF MODEL

Abstract

This paper describes an assimilation method to modify underestimation of spatial snowfall distribution. In the Yoneshiro River catchment, underestimation of spatial snowfall distribution causes less river discharge in a runoff model. Error of observed precipitation has a great influence on accuracy of snowfall distribution, especially, on condition of heavier snowfall. In order to reduce the observation error, the snowmelt-runoff model incorporates assimilation strategy of observed snow depth into snow water equivalent (SWE) model at regular intervals. As results, the assimilation makes an improvement in the accuracy of the snowmelt-runoff simulation that the Nash-Sutcliffe coefficient is changed from 0.32 to 0.93.