An estimated 60% of the presently observed sea level rise is attributed to melting glaciers, ice caps and ice sheets. The contribution of Arctic glaciers and especially the Greenland ice sheet (GrIS) has more than doubled over the period 1961-2008, and is likely to increase further when global temperature increases over the coming century. A first order effect of increasing temperatures on the glacier mass budget is an increase in surface melt. However, not all melt contributes to mass loss via runoff, because part of the meltwater refreezes in the cold snow. For the GrIS, an estimated 30-50% of all meltwater refreezes. Infiltration of water and subsequent refreezing is not well included in existing mass balance models, and estimates of refreezing are not well validated due to a lack of observations. Given the importance of refreezing for the mass budget of Arctic glaciers, in this proposal we focus on improving refreezing estimates for the GrIS and therewith other Arctic ice masses, by using a combined observational and modelling approach. Existing satellite and in-situ observations, complemented with detailed snow temperature observations to be carried out in western Greenland (K-transect), will be used to evaluate the mass balance terms and refreezing modelled with the regional climate model RACMO2, which includes a multi-layer snow model. After improvements to the snow model have been implemented and evaluated, RACMO2 will be run for the period 1958-present-2100 in order to study the role of refreezing in the mass balance of the GrIS in a changing climate.