Award Date

12-15-2019

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Civil and Environmental Engineering and Construction

First Committee Member

David James

Second Committee Member

Donald Hayes

Third Committee Member

Daniel Gerrity

Fourth Committee Member

Erica Marti

Fifth Committee Member

Jaeyun Moon

Number of Pages

260

Abstract

Lakes and freshwater reservoirs often serve as the primary drinking and irrigation water sources for surrounding communities. They provide recreational and tourism opportunities, thereby promoting the prosperity of neighboring communities. Reliable estimates of water quality in lakes and reservoirs can improve management practices to protect water resources.

Seasonal water temperature and solar shortwave radiation variations, and their subsequent interactions with water column aquatic life, combined with seasonal variations of mixing intensity throughout the water column, result in variations of water quality constituents with depth during the annual cycle. The complexity of these variations entails the use of advanced water quality modeling approaches to evaluate the trends of water quality variations over time.

The current study presents two different modeling approaches for water quality modeling in lakes and reservoirs.

In the first approach, a three-dimensional process-based model (AEM3D, HydroNumerics Pty Ltd.) was used for hydrodynamic modeling of Lake Arrowhead, California. The model was calibrated based on in-situ measured meteorological and water quality data. The calibrated process-based model was able to simulate water temperature and salinity profiles in the lake at different depths from May 2018 to April 2019, with mean relative errors of less than 6.1% and 4.2%, respectively. The model was also used to evaluate the mixing intensities at different depths during the study period.

The second approach employed two separate data-driven models incorporating wavelet transform and artificial neural networks for water quality modeling of Boulder Basin, Lake Mead. The first data-driven model proposed a cost-effective method for estimating water quality profiles based on environmental data measured at the water surface. The model could estimate water temperature, dissolved oxygen, and electrical conductivity profiles from May 2011 to January 2015 with mean relative errors of 0.52%, 0.62%, and 0.22%, respectively.

The second data-driven model was designed to forecast future water quality variations at different depths in Boulder Basin, Lake Mead. This model used a time step of 6 hours based on the availability of water quality data, and forecasted up to 960 step-ahead (240 days) water quality profiles in the basin. The data-driven model was able to successfully forecast 180-day ahead water temperature, dissolved oxygen, and electrical conductivity profiles in the basin with relative errors of less than 7.5%, 15.5%, and 4.7%, respectively.

Results of this study can benefit water management practices to evaluate different water quality modeling approaches and select appropriate methods based on their needs and budget to simulate water quality variations of their lakes and reservoirs.

Keywords

Artificial neural networks; Hydrodynamic modeling; Lake water quality modeling; Surface mixed layer; Thermal stratification; Wavelet transform

Disciplines

Civil Engineering | Environmental Engineering | Water Resource Management

File Format

pdf

File Size

13.2 MB

Degree Grantor

University of Nevada, Las Vegas

Language

English

Rights

IN COPYRIGHT. For more information about this rights statement, please visit http://rightsstatements.org/vocab/InC/1.0/


Share

COinS