Glaciated Valley Profiles: An Application of Nonlinear Regression

摘要

The cross valley profile of bedrock valleys that have been extensively glaciated arerngenerally U-shaped; questions of interest include estimation of the actual shape from therndata, ability to discriminate between shapes and comparison of these profiles acrossrnvalleys. Data for such profiles are obtained from digital elevation maps via samplingrntransects across valleys. Current methodology as in Pattyn & Van Heule (1998) employsrn“curve fitting” without the use of any diagnostic measures. We consider improvementsrnthat include the use of random sampling for selection of profiles and the use andrncomparison of various nonlinear regression models; illustrated with an example from thernHimalayas.