In order to increase coverage, mass balance changes of the world’s ice sheets are increasingly derived from surface elevation changes measured via satellite. Across the percolation zone of the Greenland Ice Sheet, meltwater, percolation and refreezing cause a re-distribution of mass through densification which may result in elevation change with no associated mass loss. Therefore, densification processes need to be quantified, spatially and temporally, and accounted for in mass balance measurements. This thesis investigates the relationships between patterns of elevation change and temporally and spatially variable accumulation and densification processes. In doing so, it provides an important contribution to the validation of the European Space Agency’s CryoSat-2 mission by placing error bars on the accuracy to which changes in satellite-measured ice-mass surface elevation represent real changes in ice mass. Temporal variability in near-surface (<10 m) snowpack and firn density and structure was measured in snowpits, shallow cores and using a neutron probe in the spring and autumn of 2004 at ~1945 m elevation (T05, 69o 51N, 47o 15W) in the percolation zone of the Greenland Ice Sheet. Results show that average snowpack density increased by 26% from spring to autumn, with a 5% (7.6 cm) increase in elevation, and a corresponding 32% increase in mass. Spatial variability was investigated at 11 sites along two transects at spatial scales of 1 m – 10 km. Whilst there was little variability in small scale (1 - 100 m) density changes, ‘seasonal densification’ increased at lower elevations, rising to 47% 10 km closer to the ice sheet margin at 1860 m a.s.l. The spatial variability in seasonal densification was further investigated in spring 2006 at seven sites located at ~10 km intervals along a 57 km transect spanning a 350 m elevation range. Snowpits and shallow cores reveal no significant variation in spring (prior to melt) snowpack density but following summer melt and refreezing cycles, seasonal densification decreased with increasing elevation at 32 kg m-3 per 100 m. Measurements at three sites ranging in elevation from 1860 – 2015 m and spanning three melt-seasons show inter-annual variation in the seasonal densification gradient. In order to obtain a longer time series of mass balance, a 17 m core retrieved in spring 2004 was analysed for stratigraphy, density and ionic and isotopic concentrations to identify annual layers. Unfortunately, the seasonal melt cycle (whereby on average 10% of the snowpack undergoes melt), results in a complex stratigraphy and density and ionic concentrations that cannot be resolved into a seasonal signature. However, the δ18O and δ D isotopes show clear sinusoidal fluctuations, which have been used to derive annual mass balance from 1986 to 2003. These show a mean annual accumulation of 53.7 cm w.e. (s.d. 12.9 cm w.e.) although the accuracy of these measurements is compromised by the percolation of meltwater through more than more year’s snowpack. These findings confirm that estimates of mass balance cannot be calculated solely from observed changes in surface elevation. However, predicting spatial and temporal variations in densification is not straightforward because of the complex inter-annual variations in the processes of accumulation, melt, percolation and refreezing.