Annual cycles and multi-year summer simulations of Greenland ice sheet surfaceclimate are made with the 0.5 degree horizontal resolution regional climate modelHIRHAM. The model results are compared with meteorological and energy balanceobservations from 15 Greenland Climate Network (GC-Net) automatic weatherstations. The model reprouces the surface climate, often within model andobservational uncertainty. However, the presence of numerous systematic modelbiases have been identified. Improving the representation of the model orographywould constitute a major improvement of the simulation of the Greenland ice sheetsurface climate, for biases in parameters such as temperature and pressure arelinked with elevation errors, not physical model error. The modeled near surface airtemperatures are warmer compared to observations, resulting primarily from thepoor resolution of the near surface layer and near surface temperature inversion.A more advanced planetary boundary layer scheme for Arctic conditions yieldedpromising results in a sensitivity run. Additional model warm bias is linked withnegative albedo bias which leads to a positive net shortwave bias. Negative albedobias is caused by a insufficient ice sheet elevation representation (-190 m averageorography bias) that results in warmer temperatures. Surface sensible heat fluxes are overestimated,likely due to model warm bias and too great wind speed. Latent heat fluxes in the model arealso larger than calculated from GC-Net measurements, indicating more evaporation than observationssuggest. Furthermore, the bulk model for evaporation neglects the majority ofwater vapour deposition, particularly at high elevations. Annual maps and total massflux of precipitation and evaporation are analyzed in context of results in literature.Based on the results of a multi parameter comparison, numerous recommendationsfor further model development are made.