Abstract. Daily historical snowfall data were analysed with the objective of determining the stability of their variability at short temporal scales. The data are weakly multifractal over periods shorter than one month, which controls their scaling properties and which can be used to statistically downscale monthly data to shorter-duration values. Although the daily snowfall values appear to be stationary, their multifractality displays much temporal variability, with most sites showing statistically-significant trends. Through use of a physically-based hydrological model, it is demonstrated that the variability of the multiscaling of snowfall can affect the timing and quantity of snow accumulation in catchments where the snowpacks are subject to wind redistribution. Therefore trends in scaling, based on multifractal characteristics, should be taken into account when downscaling climate model scenario outputs.