GSA Annual Meeting in Denver, Colorado, USA - 2016

Paper No. 158-4
Presentation Time: 9:00 AM-6:30 PM


KOLLER, Max1, ROBSON, Benjamin Aubrey1 and HÖLBLING, Daniel2, (1)Department of Geography, University of Bergen, Fosswinckelsgate 6, Bergen, 5007, Norway, (2)Department of Geoinfomatics -Z_GIS, University of Salzburg, Schillerstrasse 30, Salzburg, 5020, Austria,

Quantifying erosional processes is important both in terms of understanding catchment dynamics and landscape evolution, and in planning for potential future scenarios. Downscaled climate models predict a 20 – 60 % increase in flood discharge with a 200-year recurrence interval by the year 2100 in western Norway. Thus, the need for improved flood-risk maps and implementation of adaptive and coping strategies is pressing to ensure public safety and the protection of infrastructure. We also strive to present and communicate results in a visually appealing, clear and understandable way to the wider public.

In late October 2014 the western Norway was hit by intense precipitation over three days which resulted in severe flooding in many catchments. In this study, we focus on a meandering section of the river Flåmselvi in the valley of Flåm. The valley is a north-south oriented, glacially eroded valley filled with glacifluvial sediment, and landslide and rock fall deposits on the slopes. The bedrock consists of mainly phyllite and gneisses and the slopes are considered active.

In this study we evaluate the feasibility of using multi-temporal LiDAR data to quantify and visualize erosional processes related to flooding over an area of 3 km2. Three high resolution data sets are available; one from 2008 and two from 2014 – before and after the flood respectively. Initially, three digital terrain models (DTM) are derived from the LiDAR ground returns. The models are co-registered by minimizing slope normalized elevation biases over stable terrain before the difference in elevation is calculated to produce a high resolution (0.5 m) dataset of change in elevation, which is assumed to represent the redistribution of sediment. Uncertainties are estimated based on differences in z values over stable terrain. Furthermore, by integrating an elevation model and its derivatives, as well as orthophotos in an object-based classification environment, major landscape/land use types (arable land, forest, urban areas, etc.) are semi-automatically classified to illustrate how the flood affected different types of area and to identify land use types, and their geomorphological context, that are particularly prone to erosion.