Annual Greenland accumulation derived from airborne radar and comparisons to modeled and in situ data

Mass loss across the Greenland Ice Sheet (GrIS) has accelerated in recent decades and recently a fundamental change in the nature of this mass loss has begun. The dominant GrIS mass-loss process has switched from ice dynamics to surface mass balance (SMB) processes, including melt generation and runoff. This recent shift further emphasizes the need to monitor and constrain SMB, which, across most of the GrIS, is dominated by accumulation. High resolution, near-surface radar data have shown good fidelity at mapping spatial patterns of accumulation to validate model outputs. To better constrain accumulation over the GrIS, we derive annual accumulation rates using NASA Operation IceBridge (OIB) Snow Radar data collected from 2009 through 2012. Accumulation is calculated using the radar-determined depth to an annual layer and the local snow/firn density profile. Up to 30 years of annual stratigraphy is observed in the interior of the ice sheet, near Summit Station, while only the past year is detectable in the ablation zone around the perimeter of the ice sheet. Annual layering is traced using a semi-automatic algorithm and mapped across large areas (tens of thousands of line kilometers). A combined measured and modeled density profile is used to convert the annual stratigraphy into accumulation. Modeled density profiles from the Modèle Atmosphérique Régional (MAR) model are shown to be less than half of in situ observations in the top 1 m of snow/firn and are, therefore, replaced with in situ measurements. Using a compilation of in situ measurements, the mean GrIS snow/firn density is found to be ~340 +/- 40 kg/m3 in the top 1 m. Error in the snow density profile represents the largest error in the radar-derived accumulation. The pattern of radar-derived accumulation rate compares well with MAR estimates, although the latter has a mean bias of 4.6 cm water equivalent, a root mean square error of 16.8 cm water equivalent and a correlation coefficient of 0.6 across the entire GrIS. The accumulation-rate dataset we produced represents the largest Greenland-wide validation dataset for recent annual accumulation and, hence, SMB studies.