Spatiotemporal variability of air temperature biases in regional climate models over the Greenland ice sheet

Regional climate models (RCMs) are fundamental tools in understanding and quantifying the contribution of the Greenland ice sheet to sea-level rise. We perform an extensive evaluation of the daily air temperature simulated by two RCMs, MARv3.12 and RACMO2.3p2, and a global atmospheric reanalysis, ERA5, at 35 locations across the ice sheet over the period 1995–2020. We compare model results to weather station data from two climate networks, focusing on the spatial and temporal variability in mean biases (MBs). All three models perform well at low elevations (<1500 m a.s.l.) with an MB of 0.16∘C (MAR), 0.36∘C (RACMO) and 0.41∘C (ERA5), while warm biases (>1.70∘C) are found at high elevations (>1500 m a.s.l.). Temperature biases exhibit a strong seasonality, being more pronounced during winter and much smaller during summer ranging from 0.11∘C to 0.59∘C. No interannual variability is found in the biases of all three datasets. Daily variability within each month is captured well by both climate models and the reanalysis at most locations. Finally, all three models perform overall better in the ablation zone during summer, i.e. where and when considerable melt production occurs