Greenland ice sheet surface mass balance variability (1988-2004) from calibrated polar MM5 output

Regional climate model runs using the fifth-generation Pennsylvania State University-National Center for Atmospheric Research Mesocale Model modified for use in polar regions (Polar MM5), calibrated by independent in situ observations, demonstrate coherent regional patterns of Greenland ice sheet surface mass balance (SMB) change over a 17-yr period characterized by warming (1988-2004). Both accumulation and melt rates increased, partly counteracting each other for an overall negligible SMB trend. However, a 30% increase in meltwater runoff over this period suggests that the overall ice sheet mass balance has been increasingly negative, given observed meltwater-induced flow acceleration. SMB temporal variability of the whole ice sheet is best represented by ablation zone variability, suggesting that increased melting dominates over increased accumulation in a warming scenario. The melt season grew in duration over nearly the entire ablation zone by up to 40 days, 10 days on average. Accumulation area ratio decreased by 3%. Albedo reductions are apparent in five years of the Moderate Resolution Imaging Spectroradiometer (MODIS) derived data (2000-04). The Advanced Very High Resolution Radiometer (AVHRR)-derived albedo changes (1988-99) were less consistent spatially. A conservative assumption as to glacier discharge and basal melting suggests an ice sheet mass loss over this period greater than 100 km(3) yr(-1), framing the Greenland ice sheet as the largest single glacial contributor to recent global sea level rise. Surface mass balance uncertainty, quantified from residual random error between model and independent observations, suggests two things: 1) changes smaller than approximately 200 km(3) yr(-1) would not satisfy conservative statistical significance thresholds (i.e., two standard deviations) and 2) although natural variability and model uncertainty were separated in this analysis, the magnitude of each were roughly equivalent. Therefore, improvements in model accuracy and analysis of longer periods (assuming larger changes) are both needed for definitive mass balance change assessments.