An Integrated Decision Support Framework for Monitoring and Management of Surface Water Quality Influenced by Climate Change

An Integrated Decision Support Framework for Monitoring and Management of Surface Water Quality Influenced by Climate Change

Climate uncertainties and ongoing human pressures on land use and natural resources are creating issues in water quality management. An integrated monitoring and management system is essential to overcome the challenges emerging from various factors impacting the proper management of water quality. This research proposed an integrated decision support framework (DSF) for the monitoring and management of surface water quality which aimed to take account of the dynamic nature of hydrologic variabilities to forecast the likelihood of water quality status and thus help decision-making with a particular focus on three town water supply reservoirs of Toowoomba, Australia. This study first identifies the key water quality parameters most susceptible to changes resulting from extreme rainfall events and a water quality index (WQI) is computed to assess and quantify the overall status of water quality. The ability of four machine learning (ML) and two deep learning (DL) techniques are assessed and employed to predict the WQI using selected water quality parameters as input. The proposed models demonstrated that XGBoost and GRU outperformed the other six models in terms of predictive accuracy. The Modified Mann Kendall Test (MMK) and Innovative Trend Analysis (ITA) are employed to observe the trends of rainfall and water quality and Wavelet Transform Coherence (WTC) and the Generalised Additive Model (GAM) are used to reveal the correlation pattern between water quality and hydrological variables. The WTC identified seasonal temporal patterns of rainfall and water quality, while GAM effectively captured nonlinear and linear relationships between water quality, rainfall, and discharge across three temporal scales: weekly, monthly, and seasonal. Additionally, an optimised machine learning model XGBoost-BO is developed to predict WQI using discharge as the input parameter. The accuracy of the prediction model is validated using multiple performance metrics (R2, RMSE, MAE) to confirm the reliability. Furthermore, the Geographic Information System (GIS) technique is employed to illustrate the spatial and temporal variability of rainfall and the WQI. The integrated DSF developed in this study is expected to be a significant step forward in developing an AI-based datadriven mechanism that can equip policymakers, water resource managers, and other stakeholders with the tools needed to develop more resilient water management strategies to effectively address the uncertainties posed by climate change.

File reproduced in accordance with the copyright policy of the publisher/author.