Estimating changes in fine-sediment storage at eddy-sandbars on the Colorado River, Grand Canyon, AZ using oblique imagery from remote cameras

Lima, Ryan Edward (2021) Estimating changes in fine-sediment storage at eddy-sandbars on the Colorado River, Grand Canyon, AZ using oblique imagery from remote cameras. Doctoral thesis, Northern Arizona University.

Text Lima_2021_estimating_changes_fine-sediment_storage_eddy-sandbars_on_co.pdf - Published Version Download (3MB)

Abstract

This research is a fluvial remote sensing study demonstrating methods for sub-annual monitoring via a dataset containing over 1,000,000 ground-based oblique images that capture daily observations at 43 sandbar monitoring sites in the dam-affected Colorado River in Grand Canyon since 1990. Over half of the world’s major rivers are affected by dams. In many dam-affected rivers, sediment limited conditions have led to increased erosion of banks and fine-sediment deposits for hundreds of kilometers downstream. Quantifying the short-term rates of erosion and measuring the effect of dam operations and beach-building high flows on sediment storage at sub-annual scales is critical to managing downstream resources effectively. Sandbars in the Grand Canyon provide relatively flat, vegetation-free substrates utilized by nearly 25,000 river runners annually. Sandbars are also essential components of riverine systems creating habitat for native fish and storing sediment which would otherwise be transported downstream. This study is the most comprehensive attempt at quantitative analysis of this dataset. I present methods for estimating sandbar volume and hypsometry from the remote imagery. I demonstrate a deep learning approach to semantic segmentation, which allowed for detailed image-derived sandbar area analysis of over 13,000 images across 10 years at three sites. Significant variability was observed in the sub-annual area change due to current dam operations. I determined that erosion, deposition, and the resulting mean monthly area at sandbar sites are more closely correlated with antecedent sandbar size than monthly flow metrics. The analysis of time-lapse videos at 41 sites revealed links between daily and seasonal discharge patterns and mass failure rapid erosion events. These insights increase our understanding of the dynamics of fluvial bedforms in dam-affected, canyon-bound rivers and might improve the adaptive management of the Colorado River in the Grand Canyon. These methods could be applied broadly to remote-camera monitoring efforts in many other fluvial and coastal settings for measuring erosion rates and improving, modeling and sediment budgeting efforts.

Item Type:	Thesis (Doctoral)
Publisher’s Statement:	© Copyright is held by the author. Digital access to this material is made possible by the Cline Library, Northern Arizona University. Further transmission, reproduction or presentation of protected items is prohibited except with permission of the author.
Keywords:	Deep Learning; Fluvial Geomorphology; Fluvial Sandbars; Glen Canyon Dam; Grand Canyon; Image Segmentation;Beach formation; Colorado Riveer
Subjects:	G Geography. Anthropology. Recreation > GB Physical geography
NAU Depositing Author Academic Status:	Student
Department/Unit:	Graduate College > Theses and Dissertations College of the Environment, Forestry, and Natural Sciences > School of Earth Sciences and Environmental Sustainability
Date Deposited:	09 Feb 2022 17:49
Last Modified:	26 Aug 2022 08:30
URI:	https://openknowledge.nau.edu/id/eprint/5681

Actions (login required)

IR Staff Record View

Downloads