Use of High-Resolution Multispectral Imagery to Assess the Role of Sand and Gravel Shoals for Flood-Infrastructure Interaction in the Yellowstone River, Montana

Gravel and sand shoals may play a significant role for infrastructure impacts during flood events. This study uses high-resolution (30-cm panchromatic resolution at nadir) pan sharpened multispectral satellite images to map sand and gravel shoals near the Cinnabar Basin Bridge in Gardiner, Montana. While the bridge was not damaged during the flood event, erosion of the riverbanks and scour were observed with significant differences noticeable between both riverbanks, which may be related to the presence or evolution of shoals during the flooding. Toward analyzing the role of the shoals on local erosion and scour patterns, the goal of this study is to classify sediments based on soil type (sand, gravel, and cobbles) and color, using unsupervised K-means clustering and spectral information divergence (SID) with images taken in October 2022 and January 2023. The soil classifications from the K-means clustering and SID analysis are then compared to the pan sharped RGB images, drone images, and on-site observations and photos to assess the results from the classifications. The resulting classifications are compared for changes in the composition of the shoals over time from the October image to the January image, but also take advantage of receding water levels over the winter. Both the K-means clustering and SID analysis yield sediment classifications matching the site photos and drone images. Results show that unsupervised K-means clustering and SID can be used to classify sediments into different groups based on soil type and color, which may enable a deeper investigation of the impacts of sand and gravel shoals on infrastructure.