Semester of Graduation

Summer 2026

Degree Type

Thesis

Degree Name

Masters in Civil Engineering

Department

Department of Civil and Environmental Engineering

Committee Chair/First Advisor

Dr. Tien Yee, Assistant Chair, Department of Civil and Environmental Engineering

Second Advisor

Dr. Sunanda Dissanayake, Chair, Department of Civil and Environmental Engineering

Third Advisor

Dr. Da Hu, Assistant Professor, Department of Civil and Environmental Engineering

Abstract

Flash floods are highly unpredictable and typically strike small urban streams. While critical for monitoring extreme weather impacts, traditional stream monitoring in the United States is often restricted by high deployment costs, spatial limitations, and maintenance hazards during extreme events. To address this gap, this research develops a low-cost, non-intrusive Large-Scale Particle Image Velocimetry (LSPIV) system that safely detects and measures high-flow scenarios. The framework combines highresolution camera hardware, open-source computational processing, and depth-varying log-law correction to compute depth-averaged velocities and total channel discharge.

The system was initially validated under controlled laboratory conditions at Kennesaw State University, achieving a relative error of -2.4%. Field validation was conducted at two active urban sites, Sewell Mill Creek and Noonday Creek, where LSPIV flow estimates were compared against U.S. Geological Survey (USGS) stream gauge data. Results showed errors of roughly -8% at Sewell Mill Creek and -1.8% during highflow conditions at Noonday Creek.

Transect velocity distributions were also cross-validated against independent twodimensional hydrodynamic simulations built in the Sedimentation and River Hydraulics – two-dimensional (SRH-2D) model. Noonday Creek’s LSPIV estimates aligned well spatially with the simulation, but the same comparison was not possible for Sewell Mill Creek, where the surveyed cross-section geometry differed significantly from the 1-m digital elevation used in the numerical model, indicating substantial bathymetry change over time that made direct comparison inappropriate. These results confirm that the v proposed low-cost LSPIV system, paired with current physical bathymetry, offers a reliable, cost-effective alternative for flash flood monitoring.

Comments

USGS 104B Award (ID: GA_2024_YEE)

Available for download on Tuesday, July 24, 2029

Share

COinS