Date of Award

5-2021

Degree Type

Report

Degree Name

Master of Science (MS)

Department

Civil and Environmental Engineering

Committee Chair(s)

Bethany Neilson (Committee Chair)

Committee

Bethany Neilson

Committee

David Liddell

Committee

Ruijie Zeng

Abstract

Nearly a quarter of all people rely on karst aquifers for drinking water. In the United States, the Safe Water Drinking Act requires a complete assessment of public water systems' vulnerabilities to contamination. As part of that assessment, watershed boundaries must be delineated, while recharge and supply locations identified. In the context of karst aquifers, surficial karst features, such as sinkholes, can act as a point source of direct recharge to karst aquifers and create vulnerabilities to critical drinking water sources. Historical methods of locating these features are inefficient and depend on basic field investigations, resulting in a clear need for advanced identification methods. To this end, this study focuses on developing more efficient identification methods that use remotely sensed data to locate and map surficial karst features that may require protection. Satellite and unmanned aerial vehicle (UAV) data were used to explore the resolution needed to identify surficial karst feature signatures and the most promising methods for analyzing these data. This study's data included red, green, blue, and near-infrared reflectance rasters, thermal mosaics, and digital surface and terrain models. Spectral and thermal properties were used to filter data that could include karst features. Additionally, digital elevation models were used to explore multiple smoothing methods, image differencing, edge detection, terrain curvature, sink location, and watershed delineation. Findings from the different methods were compared to known karst feature locations. Data with a resolution between 0.5 and 2.5 meters per pixel were found to be ideal for most methods tested. However, vegetation removal, followed by a simple interpolation to fill these areas, created data analysis problems and highlighted the need for other data products, such as LiDAR, that provide accurate elevations of terrain shrouded by vegetation. In the end, it was found that edge detection, mapping curvature, and locating of low points (or sinks) via DEM analyses are all promising methods. It was concluded that by combining multiple methods, detailed digital terrain models could accurately locate many surficial karst features.

Share

COinS