Seasonal snow cover indicators in coastal Greenland from in situ observations, a climate model, and reanalysis

Abstract. Seasonal snow cover has important climatic and ecological implications for the ice-free regions of coastal Greenland. Here we present, for the first time, a dataset of quality-controlled snow depth measurements from nine locations in coastal Greenland with varying periods between 1997 and 2021. Using a simple modelling approach (Δsnow), we estimate snow water equivalent values solely based on the daily time series of snow depth. Snow pit measurements from two locations enable us to evaluate the Δsnow model. As there are very little in situ data available for Greenland, we then test the performance of the regional atmospheric climate model (RACMO2.3p2, 5.5 km spatial resolution) and reanalysis product (CARRA, 2.5 km spatial resolution) at the nine locations with snow observations. Using the combined information from all three data sources, we study spatio-temporal characteristics of the seasonal snow cover in coastal Greenland using the example of six ecologically relevant snow indicators (maximum snow water equivalent, melt onset, melt duration, snow cover duration, snow cover onset, and snow cover end date). In particular, we evaluate the ability of RACMO2.3p2 and CARRA to simulate these snow indicators at the nine different locations, perform a time series analysis of the indicators, and assess their spatial variability. The different locations have considerable spatial and temporal variability in snow cover characteristics, and seasonal maximum snow water equivalent (amount of liquid water stored in the snowpack) values range from less than 50 kg m−2 to greater than 600 kg m−2. The correlation coefficients between maximum snow water equivalent output from Δsnow and CARRA and RACMO are 0.73 and 0.48, respectively. Correlation coefficients are highest for maximum snow water equivalent and snow cover duration, and model and reanalysis output underestimate snow cover onset. We find little evidence of statistically significant (p<0.05) trends at varied periods between 1997 and 2021 except for the earlier onset of snowmelt in Zackenberg (−8 d per decade, p=0.02, based upon RACMO output). While we stress the need for context-specific validation, this study suggests that in most cases snow depth or snow water equivalent output from CARRA can describe spatial and temporal characteristics of seasonal snow cover, particularly changes in melt onset and snow cover end date.