Atmospheric Pressure Error of GRACE in Antarctic Ice Mass Change

Abstract:

As GRACE has observed time-varying gravity longer than a decade, long-term mass changes have been emerged. In particular, linear trends and accelerated patterns in Antarctica were reported and paid attention for the projection of sea level rise. The cause of accelerated ice mass loss in Antarctica is not known since its amplitude is not significantly larger than ice mass change associated with natural climate variations. In this study, we consider another uncertainty in Antarctic ice mass loss acceleration due to unmodeled atmospheric pressure field. We first compare GRACE AOD product with in-situ atmospheric pressure data from SCAR READER project. GRACE AOD (ECMWF) shows spurious jump near Transantarctic Mountains, which is due to the regular model update of ECMWF. In addition, GRACE AOD shows smaller variations than in-situ observation in coastal area. This is possibly due to the lower resolution of GRACE AOD, and thus relatively stable ocean bottom pressure associated with inverted barometric effect suppresses the variations of atmospheric pressure near coast. On the other hand, GRACE AOD closely depicts in-situ observations far from oceans. This is probably because GRACE AOD model (ECMWF) is assimilated with in-situ observations. However, the in-situ observational sites in interior of Antarctica are sparse, and thus it is still uncertain the reliability of GRACE AOD for most region of Antarctica. To examine this, we cross-validate three different reanalysis; ERA Interim, NCEP DOE and MERRA. Residual atmospheric pressure fields as a measure of atmospheric pressure errors, NCEP DOE – ERA Interim or MERRA – ERA Interim, show long-term changes, and the estimated uncertainty in acceleration of Antarctic ice mass change is about 9 Gton/yr^2 from 2003 to 2012. This result implies that the atmospheric surface pressure error likely hinders the accurate estimate of the ice mass loss acceleration in Antarctica.