Scale Effect Geomorphometric Parameters of Spatial Pattern of Snow Depth

Promotion of scale informaion quantity can improve the prediction of snow parameters. There are limited studies about the interaction on in the pizel size. The aim of this study is investigation on the effect of spatial resolution on predicting snow depth through empirical test of the relationship between some digital elevation models and snow depth modeling using multi variate regression medel. First using Latin Hypercube Sampling (LHS) technique 100 snow depth data and 195 random data were collected. Then a base DEM with 10m resolution was selefcted and 25 terrain parameters were extracted from it as the ANN input. 9 DEMs with different pixel sizes were resampled from the base DEM. Finally effective parameters on sonw depth were estracted from 10 DEMs and their relationship between measured data was calculated usnig a multiple linear regression. The models were compared by RMSE, NMSE, MSE and MAE and the results showed that the DEM with 150m resolution was the best DEM for snow depth simulation. Thus this result can reduce costs and increase the accuracy of estimation of snow depth.