ABSTRACT: Neural networks applied to estimating subglacial topography and glacier volume.

CGRG Bibliography of Canadian Geomorphology
Search Results

Author :

Date :

Title :

Publication :

Issue :

Page(s) :

Abstract

Most of Earth's mountain glaciers and ice caps are shrinking rapidly. To predict the rate and consequences of these rapid changes, we need improved regional-scale models of mountain glaciation and better knowledge of the subglacial topography upon which these models must operate. Because geophysical ice thickness mapping of more than 100,000 glaciers is unfeasible, we have developed a neural network approach to estimating glacier thickness that uses calculations performed on a digital elevation model (DEM) and on a mask of the present-day ice cover. To apply the method, we classify ice-covered DEM cells as either gently-sloping, for which the neural net approach is appropriate, and steeply-sloping, for which a simple algebraic estimator is employed. The depth predictions are validated by using a numerical ice dynamics model to reglaciate regions that are currently ice-free, with the aim of generating DEM datasets for which the subglacial topography is perfectly known. In this manner, predictions of ice thickness based on the neural network can be compared to the known ice thickness and the performance of the neural network can be evaluated and improved. From our results, thus far, we find that the neural net depth estimates yield plausible, smoothly varying subglacial topography with a typical r.m.s. elevation error of 50 m.

Bibliography of Canadian Geomorphology