Regional Mountain Conference

Quantitative climate information is paramount for good decision-making. However, its usage in complex mountain terrain is often challenging because of the scale mismatch between provided meteorological variables and needs of input models. Sometimes, this scale gap is overcome using statistical methods. Here we explore different topics related to mostly statistical downscaling and bias adjustment, for example why we could, why we should, or maybe should not do it. 

1) Using in-situ observations of snow depth and snowfall, we examine spatial and temporal variability. We show that even variables with high spatial and interannual variability like snow can share similar characteristics under long-term climate forcing. 

2) We assess the role of multivariate bias adjustment and downscaling techniques with an ensemble of climate models to drive hydrological models that simulate snow water equivalent and runoff. This highlights the benefits and pitfalls of statistical post-processing of climate model output. 

3) We present a novel downscaling technique based on principal components analysis that is able to merge information from different sources that are not in temporal synchrony. The contrasting results for temperature and precipitation should make us cautious when applying statistical methods that do not align with the processes resolved in climate models.

Consequently, we present conclusions from statistical analyses of climate variables and some statistical tools applied in post-processing of climate projections or (seasonal) weather forecasts, which are helpful to derive more meaningful and accurate climate and weather information for society. One such example of how these approaches can be used to create useful information is the Frame3S project that aims to provide seasonal forecasts of snow cover for the Tyrol-SouthTyrol-Trentino region.