Statistical Downscaling of Snow Cover Fraction in Svalbard

Abstract

The Svalbard archipelago in the Arctic is one of the fastest warming areas on the planet, with important consequences for birds that migrate to the region. Resultant changes in the timing of the spring snow melt have an impact on the breeding success of populations. To study how these populations may be affected in the future, ecologists require high resolution climate data to perform ecological modelling. However, the output from General Circulation Models is too low resolution (roughly 100 km) to be applied directly to Species Distribution Models. As a result, downscaling routines are required to bridge this gap. In this project, a statistical downscaling method is introduced to produce high-resolution (1 km) snow cover fraction data directly from low-resolution model temperature. The resulting product is validated against satellite snow cover fraction and is found to successfully capture both the timing of the annual snow melt and the spatial patterns of snow cover throughout Svalbard. This method is then applied to projections from the EC-Earth climate model to provide estimates of the snow cover under different Shared Socioeconomic Pathways.