Snow processes are a key component of the water cycle in mountainous areas as well as in many areas of the mid and high latitudes of the Earth. The complexity of these processes, coupled with the limited data available on them, has led to the development of different modelling approaches aimed at improving our understanding of these processes and supporting decision-making and management practices. Physically based approaches, such as the energy balance method, provide the best representation of snow processes, but limitations in data availability in many situations constrain their applicability in favour of more straightforward approaches. Indeed, the comparatively simple temperature-index method has become the most widely used modelling approach for representing snowpack processes in rainfall-runoff modelling, with different variants of this method implemented across many models. Nevertheless, the decisions on the most suitable degree of detail of the model are in many cases not adequately assessed for a given application.

In this study we assessed the suitability of a number of formulations of different components of the simple temperature-index method for rainfall-runoff modelling in mountainous areas of central Europe by using the Hydrologiska Byråns Vattenbalansavdelning (HBV) bucket-type model. To this end, we reviewed the most widely used formulations of different components of temperature-based snow routines from different rainfall-runoff models and proposed a series of modifications to the default structure of the HBV model. We narrowed the choice of alternative formulations to those that provide a simple conceptualisation of the described processes in order to constrain parameter and model uncertainty. We analysed a total of 64 alternative snow routine structures over 54 catchments using a split-sample test. Overall, the most valuable modifications to the standard structure of the HBV snow routine were (a) using an exponential snowmelt function coupled with no refreezing and (b) computing melt rates with a seasonally variable degree-day factor. Our results also demonstrated that increasing the degree of detail of the temperature-based snow routines in rainfall-runoff models did not necessarily lead to an improved model performance per se. Instead, performing an analysis on which processes are to be included, and to which degree of detail, for a given model and application is a better approach to obtain more reliable and robust results.