Combined inversion of GRACE-FO and ICESat-2 data for Antarctic Ice Sheet mass balance estimation

The Gravity Recovery and Climate Experiment and its follow-on mission (GRACE and GRACE-FO) have provided us with the most direct and near-continuous observation of the mass variation of the Earth’s surface since 2002. Combined with a glacial isostatic adjustment model, we can resolve the temporal gravity field signal related to the Antarctic Ice Sheet (AIS) to estimate the corresponding mass changes. However, the native spatial resolution of the GRACE/GRACE-FO is limited to ~300 x 300 km, hindering our ability to resolve the origin of smaller-scale mass changes. Satellite altimetry can resolve mass changes of the AIS with much finer spatial resolution. However, conversion from altimeter-measured surface height to mass anomalies required additional corrections for firn air content changes and ice density profile, introducing extra uncertainty to the altimeter mass balance estimate. The finer spatial resolution also means that satellite altimeters take longer to achieve fully coverage of observations of the Earth. Here, we combine the strengths of these two techniques to obtain a GRACE-FO solution that can better resolve the origin of mass change while still having a monthly time step. This method combined the ICESat-2 dataset with the GRACE-FO inversion process at the normal equation level. Instead of using regular gridded interpolated data, we employ the along-track ATL11 data set. This approach allows versatile mass balance estimates that easily cope with different temporal steps and spatial patterns.

In this talk, we will present results of inversions with simulated ICESat-2 and GRACE-FO data. The data were simulated using a “truth” temporal gravity field, the actual orbital data of the ICESat-2 and GRACE-FO and noise to represent uncertainty in firn and GIA corrections. Results show that our method can better resolve the simulated “truth” temporal gravity field. However, how much the combined inversion can improve the accuracy of the solution depends on the level of uncertainty related to the ICESat-2 data. Other than the original uncertainty of the ICESat-2 dataset, firn air content is another major source of uncertainty in coastal regions. Moreover, the coverage of the ICESat-2 dataset over a single mascon is important when calculating the mean height anomaly of a mascon as sampled by ICESat-2. When the height changes within a mascon have high spatial variability, significant mass changes occurring in a small area can alter even the sign of the mean height anomalies estimates if they were not sampled by the ICESat-2 groundtracks. Therefore, a scaling factor related to the intra-mascon variability of each mascon is required to account for this in the inversion process.