Abstract. The Greenland Ice Sheet is losing mass at a significant rate, primarily driven by increasing surface melt-induced runoff. Because the ice sheet’s surface melt is closely connected to changes in the surface albedo, studying multidecadal changes in the ice sheet’s albedo offers insight into surface melt and associated changes in its surface mass balance. Here, we first analyse the CLARA-A2 SAL satellite-based surface albedo dataset, covering 1982–2015, to obtain decadal albedo trends for each summer month. We also examine the rates of albedo change during the early summer, supported with atmospheric reanalysis data from MERRA-2, to discern changes in the intensity of early summer melt, and their likely drivers. We find that rates of albedo decrease during summer melt have accelerated during the 2000s relative to early 1980s, and that the surface albedos now often decrease to values typical of bare ice at elevations 50–100 m higher on the ice sheet. The southern margins exhibit the opposite behaviour, though, and we suggest this is due to increasing snowfall over the area. We then correct the mass balance estimates observed by the GRACE satellite mission with state-of-the-art ice discharge estimates to obtain observation-based estimates for the surface mass balance. The CLARA albedo changes are regressed with this data to obtain a proxy surface mass balance timeseries for the summer periods 1982–2015. This proxy timeseries is compared with latest regional climate model estimates from the MAR model. We show that the proxy timeseries agrees with MAR through the analyzed period within the associated uncertainties of the data and methods, demonstrating and confirming that surface runoff has dominated the rapid mass loss period between 1990s and 2010s. Finally, we extend the analysis to GrIS basin scale to examine discharge-albedo relationships in order to ascertain if the surface melt contributes to discharge acceleration via basal lubrication. While there is little evidence of surface melt-induced ice flow acceleration at annual timescales, we find time lags between seasonal maximum runoff production and seasonal maximum discharge rate to be in agreement with recent modelling results.