Abstract

Improving models of Antarctic ice shelf tides through data assimilation

Satellite-based measurements of ice shelf elevations include a significant contribution from ocean tides. An accurate prediction of this tidal signal must be removed from each measurement so that all elevations are referenced to the same datum (mean sea level) to expose any low frequency (seasonal and climatic) variability. Our goal is to build a tidal model that is sufficiently accurate to fully exploit the capabilities of the Geoscience Laser Altimeter System (GLAS), to be launched on the Ice, Cloud and land Elevation Satellite (ICESat) in December 2001. GLAS is capable of measuring surface elevation to ~5 cm accuracy over distances of tens of km, however existing tide models cannot predict to this level of accuracy. An accurate model is also required for tide removal prior to estimating three-dimensional shelf ice motion using synthetic aperture radar (SAR). Our approach to this problem is to use formal data assimilation methods in a model for the seas surrounding Antarctica, including the cavities beneath the floating ice shelves. We exploit a variety of data sources: TOPEX/Poseidon and ERS satellite altimetry over the ice-free ocean; benthic and coastal tide gauges; and GPS, gravimeter and ERS satellite altimetry on ice shelves. Additional constraining data will be obtained from GLAS, especially during the 8-day repeat Verification Phase at the start of the mission.

The greatest tidal displacements of ice shelves are found at the back of the Filchner-Ronne and Larsen Ice Shelves in the Weddell Sea, where spring tidal ranges can exceed 5 m. There are, however, few or no validation data in these regions, and so we focus on the better-sampled Amery Ice Shelf (AIS), where tidal ranges are typically 1-2.5 m. Comparison of our assimilation model with independent GPS data from the AIS demonstrates that, with sufficient measurements, the ~5 cm target for model accuracy is within reach. A significant caveat to this statement is that atmospheric pressure must also be known to <5 mb accuracy, since the ice shelf and underlying ocean also respond to the inverse barometer effect in which the ocean surface is depressed by ~1 cm for each mb of air pressure.

ERS altimeter measurements over ice shelves are noisy compared with our target accuracy, and the ERS orbit does not reach the southern portions of the Ross and Filchner-Ronne Shelves. Furthermore, neither ERS nor GLAS has an orbit that is well suited to tide signal retrieval. Thus, while these altimeter data can still be used in assimilation schemes, improvement in model skill for the far-southern shelves is best achieved with additional in situ tide measurements such as GPS and gravimeter records.