Regional-Scale Landscape Change: Data Preparation and Two Examples Using Object-Based Image Analysis and Windowed Iterative Closest Point Algorithm to Interrogate Results
Abstract
As lidar and other high-resolution topographic data availability expands, and is increasingly collected repeatedly to monitor landscape change, methodological improvement is required to overcome several common challenges. Lidar flightline misalignment and positioning errors are magnified when multitemporal data are compared, and complex coordinate system and elevation model transformations are often required to compare data from different sources. Furthermore, even when errors are minimized, analysis and extraction of landscape change of interest at regional scale is challenging because results are often in the form of millions of individual measurement pixels. We present results from two studies that provide examples of optimizing regional-scale topographic change detection. First, we report on a study of widespread landscape change over nearly 8000 km2 in northeastern Minnesota that resulted from an extreme storm that dropped over 20 cm of rain in 24 hours, causing infrastructure damage, widespread slope failure, and fluvial system change. We demonstrate steps to minimize flightline error using Bayesian methods and methods to reduce positioning error using corrections from stable landscape areas. These approaches improve detection of real versus spurious landscape change, which are then analyzed and classified using Object-Based Image Analysis, providing classified vector objects. These approaches show promise as a way forward to better interrogate complex and noisy regional-scale landscape change results. Second, we report on improvements to, and use of, windowed iterative closest point change detection to measure decimeter-scale fault slip over >100 kilometers of creeping faults in central California over a decadal timescale. We propose methods that allow for detection of fault offsets at magnitudes similar to the inherent error in the input lidar data sets, advancing the sensitivity of this emergent regional-scale change detection method, and better illuminating plate-boundary process and active faulting.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2020
- Bibcode:
- 2020AGUFMNH0090002D
- Keywords:
-
- 4302 Geological;
- NATURAL HAZARDS;
- 4306 Multihazards;
- NATURAL HAZARDS;
- 4328 Risk;
- NATURAL HAZARDS