Ice sheet surface mass balance from models and GRACE

Abstract:

Surface mass balance (SMB) represents a significant portion of the overall mass balance of the Greenland and Antarctic ice sheets. Accurate modeling of the SMB processes is key to our understanding of the past, present and future evolution of the ice sheets in response to climate forcings. Characterizing errors and improving the models is necessary to enhance our understanding of the SMB processes driving change, but also to isolate other processes such as dynamic ice loss through the combined reduction of satellite, airborne, in-situ and model data. In this study we quantify the performance of the MERRA-2 SMB model for the Greenland and Antarctic ice sheets using satellite gravity observations from the GRACE mission.

We use two approaches in quantifying the performance and constraining the SMB model. First we compare the spatial and temporal variation of the intrinsic modes determined from both the MERRA-2 SMB model output and the NASA GSFC release 2 iterated high-resolution GRACE mascon solution. We take advantage of the mascon solution’s improved signal to noise and minimization of signal leakage. We use the Ensemble Empirical Mode Decomposition (EEMD) adaptive filter to isolate seasonal, annual, and inter-annual modes in both the GRACE mascons and SMB model data. Spatial and temporal variations are compared and analyzed. Second, we incorporate the SMB model output of spatial and temporal surface mass change into our forward modeling used in the formal reduction of the GRACE inter-satellite range-rate data. The full spatial and temporal resolution of the SMB model is forward modeled. We analyze the change in the inter-satellite range-rate residuals as well as the new estimated mascon solution which has the SMB model output rigorously removed through forward modeling in the GRACE range-rate data reduction. The EEMD analysis is applied to the “delta mascon” solution and compared to the solution without the SMB forward modeled. The results of the EEMD analysis as well as the “delta mascon” solution are presented and interpreted. The techniques developed can be used to quantify the performance and improve other SMB models and to isolate other processes such as dynamic ice loss.