 |
| Titel |
Evaluating the sensitivity of an ice sheet model to changes in bed elevation and inclusion of membrane stresses |
| VerfasserIn |
Andy Aschwanden, Ed Bueler, Constantine Khroulev |
| Konferenz |
EGU General Assembly 2010
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 12 (2010) |
| Datensatznummer |
250036562
|
|
|
|
|
|
| Zusammenfassung |
| To predict Greenland’s contribution to global sea level rise in the next few centuries with
some confidence, an accurate representation of its current state is crucial. Simulations of the
present state of Greenland using the "Parallel Ice Sheet Model" (PISM) capture the essential
flow features but overestimate the current volume by about 30%. Possible sources of error
include (1) limited understanding of physical processes involved, (2) the choice of
approximations made by the numerical model, (3) values of tunable parameters, and (4)
uncertainties in boundary conditions.
The response of an ice sheet model to given forcing contains the above mentioned error
sources, with unknown weights. In this work we focus on a small subset, namely errors
arising from uncertainties in bed elevation and whether or not membrane stresses are included
in the stress balance. CReSIS provides recently updated bedrock maps for Greenland include
high-resolution data for Jacobshavn Isbræ and Petermann Glacier. We present a four-way
comparison between the original BEDMAP, the new CReSIS bedrock data, a non-sliding
shallow ice model, and hybrid model which includes the shallow shelf approximation as a
sliding law. Large gradients possibly found in high-resolution bedrock elevation are expected
to make a hybrid model the more appropriate choice. To elucidate this question, runs are
performed on a unprecedented high spatial resolution of 2km for the whole ice sheet.
Finally, model predictions are evaluated against observed quantities such as surface
velocities, ice thickness, and temperature profiles in bore holes using different metrics. |
|
|
|
|
|
|