||Recent surface mass balance changes in space and time over the polar ice sheets need to be better constrained in order to estimate the ice-sheet contribution to sea-level rise. The mass balance of any ice body is obtained by subtracting mass losses from mass gains. In response to climate changes of the recent decades, ice-sheet mass losses have increased, making ice-sheet mass balance negative and raising sea level. In this work, I better quantify the mass gained by snowfall across the polar ice sheets; I target specific regions over both Greenland and West Antarctica where snow accumulation changes are occurring due to rising air temperature. Southeast Greenland receives 30% of the total snow accumulation of the Greenland ice sheet. In this work, I combine internal layers observed in ice-penetrating radar data with firn cores to derive the last 30 years of accumulation and to measure the spatial pattern of accumulation toward the southeast coastline. Below 1800 m elevation, in the percolation zone, significant surface melt is observed in the summer, which challenges both firn-core dating and internal-layer tracing. While firn-core drilling at 1500 m elevation, liquid water was found at ~20-m depth in a firn aquifer that persisted over the winter. The presence of this water filling deeper pore space in the firn was unexpected, and has a significant impact on the ice sheet thermal state and the estimate of mass balance made using satellite altimeters. Using a 400-MHz ice-penetrating radar, the extent of this widespread aquifer was mapped on the ground, and also more extensively from the air with a 750-MHz airborne radar as part of the NASA Operation IceBridge mission. Over three IceBridge flight campaigns (2011-2013), based on radar data, the firn aquifer is estimated to cover ~30,000 km2 area within the wet-snow zone of the ice sheet. I use repeated flightlines to understand the temporal variability of the water trapped in the firn aquifer and to simulate its lateral flow, following the gentle surface slope (< 1) and undulated topography of the ice sheet surface toward the ablation zone of the ice sheet. The fate of this water is currently unknown; water drainage into crevasses and at least partial runoff is inferred based on the analysis of radar profiles from different years. I also present results from a field expedition in West Antarctica, where data collection combined high-frequency (2-18 GHz) radar data and shallow (< 20 m) firn cores from Central West Antarctica, crossing the ice divide toward the Amundsen Sea. The radar-derived accumulation rates show a 75% increase (+0.20 m w.eq. y-1) of net snow accumulation from the ice divide, toward the Amundsen Sea for a 70-km transect, assuming annual isochrones being detected in the radar profile. On the Ross Sea side of the divide, with accumulation rates less than 0.25 m w.eq. y-1 and significant wind redistribution, only a multi-annual stratigraphy is detected in the radar profile. Using radar, I investigated the small-scale variability within a radius of ~1.5 km of one firn-core site, and I find that the averaged variation in accumulation-rate in this area is 0.1 m w.eq. y-1 in the upper 25-m of the firn column, which is 20% of the average accumulation rate.