ASSESSING LIDAR CHANGE DETECTION FOR USE AS A DETECTION AND MONITORING TOOL FOR GEOMORPHIC CHANGE ALONG PIPELINE CORRIDORS IN CALIFORNIA

Abstract:

Airborne lidar topography is a ubiquitous and indispensable resource for mapping landforms diagnostic of geologic hazards (e.g., landslides, erosion, and earthquake fault rupture) and performing various hydrologic analyses along major lifelines. Differencing of successive lidar data (change detection) from derivative digital elevation models (DEMs) provides unique observations of landscape evolution that can be used to assess activity, rates of change, and overall risk from geohazards. Geohazard monitoring via remote sensing also has the potential to provide additional coverage (e.g., through-vegetation observation) and a quantitative supplement to traditional methods of infrastructure monitoring such as visual inspection via aerial patrol. In central and northern California, the variability in geographic terrain (steep vegetated slopes to flat deserts) and active tectonics are significant challenges for both planning and maintaining state-wide infrastructure. This study evaluates the application of change detection through DEM differencing for detecting and monitoring landslide and erosion hazards along natural gas pipeline corridors spanning the region. This review assesses data collected within three distinctly different geographic settings; 1) semi-arid, 2) urbanized, and 3) densely vegetated. In all three environments, the quality and specification for lidar data collection and processing are fundamental to producing a useful change detection product. Lidar artifacts (e.g., vertical swath offsets and scan angle artifacts) are the greatest source of noise in unobstructed (minimal vegetation) terrains. In urban settings, topographic change is often associated with anthropogenic activities. The timing of lidar data collection is a significant challenge in forested areas, not only to best capture geomorphic change but also in reducing the noise associated with processing ‘bare earth’ terrain in heavily vegetated riparian environments. Future studies will apply lessons learned from these challenges and incorporate fuzzy inference systems to help noise reduction.