Zur Seitenansicht


Calibrating a glacier ice thickness model from in-situ point measurements
VerfasserThorlaksson, Daniel
GutachterMaussion, Fabien
HochschulschriftInnsbruck, Univ., Masterarb., 2017
Arbeit an der Bibliothek noch nicht eingelangt - Daten nicht geprüft
Datum der AbgabeMai 2017
Schlagwörter (DE)Gletschermodellierung
Schlagwörter (EN)Glacier modeling / Bed inversion modeling / Model calibration
URNurn:nbn:at:at-ubi:1-7259 Persistent Identifier (URN)
 Das Werk ist frei verfügbar
Calibrating a glacier ice thickness model from in-situ point measurements [6.52 mb]
Zusammenfassung (Englisch)

Until recently the total volume estimates of glaciers world wide have been based on volume area scaling which states that the total volume of an ensemble of glaciers is directly related to their surface area. In the last few years distributed modeling of glaciers on a regional or global scale have been made possible by increased and cheaper computational power as well as increased availability of initial input data from remote sensing. The Open Global Glacier Model (OGGM) is capable of modeling glacier flow globally with few input variables and its bed inversion scheme can estimate their volume and the thickness at any given point. Currently, the model is calibrated from the total volume of those glaciers where measurements are available and their total volume has been estimated. In this thesis, I attempted to improve the calibration method of OGGM by using point thickness measurements, newly-published in the Glacier Thickness Database, version 2 (GlaThiDa), to improve the volume estimates of OGGM. This improvement is sought via a parametrization of a single key glacier variable, the temperature dependent A from Glen's flow law. One of the input variables of OGGM is a glacier outline from the Randolf Glacier Inventory (RGI). By cross-referencing RGI and GlaThiDa, RGI shapes were identified for 1060 of the total of 1080 individual GlaThiDa glacier entities. These glaciers were utilized to evaluate and validate the new calibration procedure for the Greater Alpine Area (GAR).

As the measurement points in GlaThiDa were seldom uniformly distributed across each glacier, a elevation distribution-weighted bias was constructed and the weighed bias was then implemented for each of the 98 GAR glaciers. OGGM was used to minimize this bias by comparing the glacier thickness estimate from the OGGM bed inversion function with the actual point measurements in GlaThiDa. Minimizing the bias allowed for the calculation of the scaling multiplier of Glen's $A$, which is a measure of how ice flows. The multiplier was defined to be in the range from 0.1 to 10 and about two thirds of the calibrated glaciers had values which were at these threshold values and the rest well-scattered there between. Of all of the examined glacier variables, no parametrization was found due to the large scatter of the multiplier. I made suggestions for improvements to the bias calculation and the implementation of the point measurements to reduce the scatter and the bed inversion routine of OGGM to allow analysis for glaciers on a regional or global scale. These improvements for the point measurements would be to have a more robust data quality control check of the time discrepancies of the two databases as well a spacial quality control check for the GlaThiDa points, both with regards to their actual locations and with regards to their local location relative to the parabolic bed shape of OGGM. The bed shape inversion routine of OGGM would benefit if the scheme takes into account when a glacier boundary is at glacier divide and thus, not force the boundary of the glacier bed shape to the surface. Likewise it might yield better results if the precipitation gradient would be calibrated from these point observations, either separately or in unison with the multiplier of A.