Assessing effects of the choice of meteorological forcing datasets and downscaling methods on distributed snow simulations in a mountainous catchment