Comparison of GRACE/GFO- and RACMO-based seasonal mass anomalies over the coastal zone of Greenland

Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GFO) satellite missions, altogether referred to as GRACE hereafter, are a powerful tool to provide Greenland monthly mass anomaly estimates over a time span of more than two decades. Here we present a comparison between GRACE seasonal mass anomalies and surface mass balance (SMB) estimates from the Regional Atmospheric Climate Model RACMO2.3p2. The latter provides estimates of SMB and its individual components over Greenland with a daily temporal sampling and high spatial resolution, i.e. 5.5 km in the tundra area and 1 km over glaciated areas, using statistical downscaling. Our goals are to provide frameworks to independently evaluate the model and to estimate buffered water storage (BWS), i.e. temporal storage of meltwater in the deeper ice sheet system subject to discharge into the ocean during the melt season or soon thereafter.

The comparison spans the time interval between February 2003 and July 2022. Long-term (slow) trends are removed to exclude signals related to ice discharge variability and glacial isostatic adjustment (assuming the seasonal variability of these signals is negligible). The GRACE- and RACMO-based estimates are inverted into mean mass anomalies per calendar month to extract their typical seasonal patterns. The study area is limited to the coastal zone of Greenland, including the ice sheet ablation zone where signals related to BWS and under- or over-estimation of runoff are concentrated. The coastal zone is further divided into 12 areas for a detailed analysis. We compare the time series of seasonal patterns in terms of RMS differences, Pearson correlation coefficient, and Nash-Sutcliffe Efficiency (NSE). Over the entire coastal zone, the RMS difference is 13.7 Gt or 1.3 cm in terms of mean equivalent water height (about 14% of the total signal); the correlation coefficient and NSE are 0.992 and 0.980, respectively, indicating a sufficient match between the overall seasonal patterns of the two datasets. A few individual coastal zones also show a good agreement between GRACE- and RACMO-based estimates, with the RMS differences below 2 cm. In two northwestern coastal areas, however, GRACE- and RACMO-based mass anomalies show large discrepancies (e.g. NSE lower than 0.7), potentially due to an overestimation of modelled runoff. In two southwestern coastal areas, a mismatch is observed as well, with maximum differences occurring in July, in concert with the timing of BWS documented in earlier studies.

While previous studies already attempted to quantify BWS in Greenland, they did not account for the scaling of modelled runoff, introducing a biased estimation of BWS. In light of our study, a more comprehensive approach can be adopted for BWS quantification. Such an approach can also benefit from using other regional climate models for BWS estimation.