Graduation Year

2023

Document Type

Thesis

Degree

M.S.E.V.

Degree Name

MS in Environmental Engr. (M.S.E.V.)

Degree Granting Department

Civil and Environmental Engineering

Major Professor

Qiong Zhang, Ph.D.

Committee Member

James Mihelcic, Ph.D.

Committee Member

Hui Wang, Ph.D.

Keywords

Climate Change, Forecasting, Landuse, Machine Learning, Water Utility

Abstract

Water, a crucial resource for sustaining life, covers approximately 70% of the earth's surface. Nonetheless, the quality of water is deteriorating rapidly due to the rapid growth of urban areas and industries, which is a worrying trend causing harm to human health and the ecosystem. Water quality forecasting has a key role in water resources management by enabling effective pollution control, ecosystem monitoring, and decision-making.

Previously, traditional statistical models were used to forecast water quality, but they were unable to examine the non-linear relationships between water quality parameters, and they assumed that all datasets were distributed normally. This study uses Random Forest (RF) and Artificial Neural Networks (ANN) to predict and forecast water quality for multiple water quality parameters for different water sources using ambient temperature, rainfall, and land use as predictor variables for Tampa Bay Water.

The result from this study indicates that distance to the Alafia River was the major influencing factor for groundwater quality models with a feature importance value of 0.58, season with a feature importance of 0.9 was the highest significant parameter that impacted seawater quality models, and land use having a feature importance of 0.8 contributed highly to surface water quality models. The results of the comparison between RF and ANN in forecasting water quality indicate that RF performed better than ANN in most cases, with R2 values of 0.95 and 0.56 being the highest for groundwater and seawater, respectively. However, for some surface water quality models, ANN outperformed RF with an R2 value of 0.28.

Overall, this research highlights the efficacy of machine learning techniques in water quality prediction, with RF performing slightly better than ANN. Forecasted water quality results for July 2023 to December 2024 showed that groundwater quality remains relatively stable, and seawater and surface water quality were significantly influenced by changes in ambient temperature, land use, and rainfall. The findings emphasize the importance of considering these variables in water resource management and decision-making, particularly for seawater and surface water sources, while emphasizing the possibility of utilizing machine learning for prediction and forecasting water quality.

Download

Share

COinS