Model Inputs

In numerical modelling (ice sheet, ocean or atmospheric), the quality of model output is dependent on the quality of model input. The most important data used as boundary and forcing conditions in an ice-sheet model are subglacial topography and surface mass balance. For present-day ice sheets, the quality of these model inputs is dependent on field data which, for some areas, are noticeably absent. Airborne radar surveying of large ice sheets provides the only viable method of acquiring information on the icesheet base at a continental scale. Such surveys are organized as a series of gridded flightlines. Although the technique is sound, there are currently two limitations to datasets collected by radar surveying. The first is that information between flightlines is absent. Interpolation software is used to infer the topography in such regions. The second limitation, which is acute in Antarctica, is that large portions of some ice sheets remain to be surveyed. Thus, subglacial topography in these regions has to be estimated, often from just a few seismic measurements, and is subject to a high level of potential error. To date, Antarctic Ice Sheet models have been run over a topographic DEM based on an incomplete coverage of bed elevation data.

There have been several recent advances to alleviate this problem. Since the 1970s, when radar data from over 400,000 km worth of flightlines were collected over a wide area of Antarctica (albeit at a coarse line spacing), there have been several surveys over much smaller regions with close flightline spacing. The most recent depiction of Antarctic topography was assembled by the BEDMAP consortium (Lythe et al., 2000). While this database quantifies topography well in many places, it remains restricted in two ways. First, there are several large ''data gaps'' where very little is known about sub-ice topography. Second, even in regions where coverage is good, radar transects are often separated by several kilometres (often tens of kilometres) across which interpolation of data remains necessary. Comparison of the topography measured by recently acquired radar data with the interpolated information from BEDMAP reveals that, in some places, the errors in BEDMAP are up to the order of hundreds of metres (Welch and Jacobel, 2003).

Mass balance of large ice masses is also very difficult to establish accurately over a wide area. Although it is possible to measure annual mass balance at specific sites, there is no quick and easy method by which large-scale data can be obtained. Satellite observations, used in conjunction with direct measurements, provide a means of estimating mass balance across a wide area, and have been used to assess Antarctic mass balance (Vaughan et al., 1999; Arthern et al., 2006). Such analysis only provides an estimate of the current situation, however, which is of less use when trying to model past or future scenarios. As this is a real problem for Antarctic Ice Sheet models, many are coupled with models of surface mass balance. The accuracy of ice-sheet model results will then be dependent on the quality of the surface balance results. Similar problems exist when accounting for the processes of iceberg calving and ice-shelf basal melting in ice-sheet models.

0 0

Post a comment