Regional-scale mapping of landscape response to extreme precipitation using repeat lidar and object-based image analysis

Abstract

Extreme precipitation events may cause flooding, slope failure, erosion, deposition, and damage to infrastructure over a regional scale, but the impacts of these events are often difficult to fully characterize. Regional-scale landscape change occurred during an extreme rain event in June 2012 in northeastern Minnesota. Landscape change was documented by 8,000 km 2 of airborne lidar data collected before and after the event. Following improved alignment of the lidar point data and reducing error using insight from analysis of extensive stable areas, elevation differences were classified into map objects representing geomorphic change in relation to process and landscape position using object-based image analysis. This remote mapping compares favorably to field and imagery-based mapping and provides the basis for volumetric sediment budgeting. Elevation differences in these objects indicate that 4.5 × 10 6 ± 1.0 × 10 6 m 3 of sediment was eroded in the study area. Of this, 2.5 × 10 6 ± 3.3 × 10 5 m 3 was deposited in deposits on hillslopes and valley floors, and 2.0 × 10 6 ± 4.6 × 10 5 m 3 were removed from watersheds and exported to the Saint Louis River Estuary and Lake Superior. Multivariate logistic regression analysis emphasized that topographic slope and presence of glaciolacustrine clay lithology are the primary control on landslide occurrence, and landslides occur most frequently on slopes within tens of meters of stream channels. These results provide the basis to anticipate the impacts of similar future storm events. Because precipitation events are forecast to continue to increase in frequency and intensity owing to climate change, characterizing and anticipating their effects may support hazard planning.