Firn aquifers in the percolation facies of the Greenland Ice Sheet (GrIS) form when summer meltwater percolates into the pores of the firn and liquid water does not freeze in the following winter. It occurs in areas with strong surface melting and high snow accumulation and has a significant impact on the stability and mass balance of the ice sheet. At present, most of researches on firn aquifers come from field investigation, airborne radar detection and model simulation. But the detection range and time of field investigation and airborne radar detection are limited. The model simulation is greatly affected by the physical process description and the quality of input parameters. The regional climate model has great uncertainty especially in the simulation of ice sheet melt. The microwave scatterometer has wide coverage and short revisit period that can map Greenland information rapidly. In this study, the C-band Advanced Scatterometer (ASCAT) data is utilized to investigate the dynamics in firn aquifers over the GrIS because of its excellent spatial and temporal coverages, as well as its sensitivity to the changes in subsurface liquid water. Firn aquifers are mapped based on a Random Forest Classifier with the aid of ground truth measurements from the NASA Operation IceBridge (OIB) program. The satellite-derived melt days, melt intensity and mean winter backscatter values are used as input vectors. We obtain the annual firn aquifers extents from 2008 to 2020, and the total firn aquifers area from 2008 to 2020 is 81844 km², which is 14.11% of the percolation facies area and 4.55% of GrIS area. The overall accuracy between 2010-2017 results and the OIB results is 98.09% and the Kappa coefficient is 0.59. We find firn aquifers in the southeastern, southern, and northwestern regions of Greenland. And the extent of firn aquifers reached their maximum in 2008 and minimum in 2018.

Greenland Ice Sheet,Firn aquifers,Microwave scatterometer,Melt intensity,Random Forest Classifier