Evaluating the Influence of Climate on Water Quality Using Machine Learning Models

Doddivenuka, Sai Kiran Goud (2025) Evaluating the Influence of Climate on Water Quality Using Machine Learning Models. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (748kB) | Preview
Preview	PDF (Configuration Manual) Download (480kB) | Preview

Abstract

This research presents an enhanced machine learning and deep learning framework for predicting Dissolved Oxygen (DO) levels in surface water, incorporating both climate and temporal features. While traditional DO models often rely solely on water chemistry, this study integrates dynamic variables such as air temperature, seasonal cycles, and derived indices like thermal stratification and temperature-salinity interactions.

The project explores multiple supervised learning techniques including Random Forest, XGBoost, Logistic Regression, and a Multi-Layer Perceptron (MLP) neural network, benchmarking them across regression and classification tasks. A new risk-based DO classification system (Safe, Moderate, Critical) was also introduced and balanced using the Synthetic Minority Oversampling Technique (SMOTE) for fair model training. SHAP (SHapley Additive exPlanations) explainability methods highlighted that features like DO lag, water temperature, and climate interactions significantly influence DO predictions. The Extra Trees classifier achieved the best classification accuracy (84.6%) in binary mode, while the MLP achieved 77% accuracy in the multi-class DO risk task after SMOTE.

This work contributes a comprehensive, interpretable, and scalable pipeline that supports real-time DO monitoring under the lens of climate change. It strengthens the foundation for smart environmental systems and sustainable decision-making in water resource management.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Haque, Rejwanul UNSPECIFIED
Uncontrolled Keywords:	Dissolved Oxygen prediction; Climate-aware AI; SHAP explainability; Machine Learning; Water Quality; Deep Learning; SMOTE
Subjects:	G Geography. Anthropology. Recreation > GE Environmental Sciences Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	25 Aug 2026 13:29
Last Modified:	25 Aug 2026 13:29
URI:	https://norma.ncirl.ie/id/eprint/9626

Actions (login required)

View Item