Climatology of Ulvebreen, Svalbard, from automatic weather station and regional climate models.

Abstract

The Arctic has warmed nearly four times faster than the global average over the past four decades. With the high Arctic experiencing the strongest human-driven warming, clear from both observations and models. Svalbard holds 6% of the global land ice mass outside the major ice sheets, and is responding strongly to this amplified warming. Its 2024 summer mass loss is comparable to that of the entire Greenland ice sheet (~1% of the archipelago's ice volume). This study uses nine years (2015-2022) of AWS data from Ulvebreen, a glacier on the east coast of Spitsbergen, to analyse local SEB and SSMB. To broaden the perspective, we compare this with RACMO2 and its high-resolution version (HiRes RACMO) across eastern and western glacierised regions, while also considering sea ice concentration trends. SEB climatologies show strong seasonal cycles, with enhanced melt during westerly winds and ice-free periods, beyond just seasonal effects. HiRes RACMO matches observed temperatures well but underestimates incoming radiation by up to 20\% (due to an underestimation of cloud cover), leading to roughly half the observed SSMB loss. Across Spitsbergen, HiRes RACMO shows warming and negative SSMB trends strongest at low elevations and varying spatially. These are driven by shifts in precipitation and enhanced melt and runoff. Sea ice decline is rapid and more pronounced in the west, but its impact on melt is greater in the east. Although sea ice conditions are not directly linked to glacier melt, ice-free periods coincide with enhanced melt driven by higher temperatures, humidity, and incoming radiation. Differences in seasonal sea ice loss between western and eastern Spitsbergen may explain differing temperature trends with elevation, with stronger low-elevation warming in the east associated with greater sea ice decline. While the nine-year AWS Ulvebreen dataset is valuable, given the scarcity of similar datasets and the clear differences between observations and model output. The dataset is too short to detect long-term trends of climate change, highlighting the need for longer observations and higher-resolution models to better understand the impacts of climate and sea ice variability on glaciers.