UNDERSTANDING CHANGING LANDSCAPES BY IMPROVING MULTITEMPORAL LIDAR DATA AND EXPLORING OBJECT-BASED IMAGE ANALYSIS

Landscape changes caused by floods, landslides, and earthquakes pose hazard to communities, but also provide opportunity to enhance scientific understanding of geomorphic process. The increased availability of high-resolution remote sensing data motivates us to improve our ability to quantitatively assess landscape change. Evaluation of patterns and rates of landscape change have enabled the development of landscape-scale sediment budgets, and led to new understanding of erosion, deposition, and crustal deformation. Recent data and methodological advances provide for granular analysis of landscape change and considerable effort is being made to extend these analyses to the finest resolution possible and to 3D. However, these high-resolution analyses can be difficult to classify, analyze, and disseminate. As such, some degree of data generalization may improve understanding of landscape function. Object-based image analysis (OBIA) can be applied to multitemporal geospatial data that capture significant landscape altering events. This developing method provides the structure to segment maps of landscape change into digital objects rather than working at the grid pixel or data point scale. Analysis of these objects using a suite of related geospatial data provides the means to better organize, classify, and extract key information about the effects of geomorphic process on the landscape. We report on application of OBIA to an extreme precipitation event that occurred in northeastern Minnesota in 2012. This event was bracketed by airborne lidar surveys allowing for analysis of flooding, landslides, and infrastructure failure across hundreds of square kilometers. This workflow includes reprocessing and manipulating point clouds using Bayesian methods on flightline and survey trajectory data and applying correction surfaces to overcome survey deficiencies that become apparent when disparate data are compared. From the resulting change mapping, we developed an OBIA ruleset using topographic and imagery data in several derivative forms to establish classified, object-based maps of landscape changes. These results can then be compared to detailed, manually generated, field and remote-sensing based maps of landscape change such as landslide inventories.