The effect of the snow weighting on the temporal stability of hydrologic model efficiency and parameters

In the current changing climate, the importance of temporal stability of hydrological models is increasing. The objective of this paper is to assess the effects of the changing weight of snow (w(S)) in a multiple objective calibration of a conceptual hydrological model in terms of the temporal stability of runoff and snow model efficiency and hydrologic model parameters. The methodology was tested by a lumped conceptual hydrological model (the TUW model), which was calibrated and validated in 213 Austrian catchments in two climatically different decades (i.e., 1981-1990 and 2001-2010). The results indicate that the runoff and snow model efficiencies are almost insensitive to a large range of wS, i.e., from 0 to 0.9 for the runoff and from 0.75 to 1.0 for the snow. This relationship is similar in the two different decades; however, the calibration runoff model efficiency (i.e., Nash-Sutcliffe efficiency) was about 5 to 10% greater in flatland catchments in the warmer decade (2001-2010). The main impact of wS on the snow model efficiency is the reduction of snow underestimation errors in the flatland catchments. The relative improvement of the snow simulations in the validation periods is greater in the flatland catchments and is in a range of 45-60% compared to 13% and 31% in the alpine catchments. The evaluation of the temporal stability indicates that increasing the weight of the snow increases the temporal stability of the degree-day factor and the parameters of the snowfall and rainfall thresholds. A change in the weight of the snow has the greatest impact on glacerized catchments, which were the most sensitive to changes in the mean annual air temperature between the two decades.