Stripping Away the Forest; Sweden's Glacially Streamlined Landscape Evaluated through Lidar

The newly available Swedish National Height Model (SNHM) is a 2.0 m horizontal, and 0.1 m vertical resolution digital elevation model (DEM) that is free at the point of use for researchers based at Swedish institutions. With coverage currently at ~80% of the country and due to be completed by 2015 this spatially extensive, high resolution dataset has opened up new avenues of research for Quaternary geology in the country. The work presented here utilises the SNHM to map and evaluate more than 10,000 glacially streamlined landforms in the south-east of Sweden. The subsequently extracted morphological variables of length, width and height are then used to investiagte three areas; to test recent conclusions drawn from the glacially streamlined landscapes of Great Britain and North America/Canada, to assess the impact of different core types on the morphological expression of said features and to attempt to calculate which morphological variable best accounts for the variability seen in the dataset.

It is found that in common with drumlins found in the British Isles, and elsewhere, their characteristics can be described by a log-normal distribution. However the long tail of the features characteristic distributions can cause problems for many of the commonly applied statistical methods of evaluation. Furthermore a re-appraisal of some conclusions drawn by previous works as to the presence of a fundamental scaling law in streamlined feature elongation is necessary due to evidence gathered here. Additionally; based on a limited sample size it has been found that it is not possible to differentiate a streamlined landform’s core type based on their morphological characteristics alone. Larger 'known'-core data sets may be able to do so, based upon the length of a feature for example, however the sample size here was not sufficient to allow significant differences to come to the fore should they exist. And lastly, the extracted variable 'height' was found to account for the vast majoirty of the variance seen in the dataset when subject to a principle component analysis (PCA).