Integrated Modeling of Water Quality and Meteorological Data for Turbidity Forecasting in the Brisbane River

Effective management of urban water resources requires integrated monitoring systems that link hydrological and meteorological data. In this study, we combined high-frequency water quality observations from the Brisbane River with historical weather data across Australia to investigate the relationship between rainfall patterns and turbidity fluctuations. Through correlation analysis, we observed that same-day rainfall has weak or negative association with turbidity, whereas lagged rainfall—particularly over ۳–۷ days—and cumulative indices such as the Antecedent Precipitation Index (API) show stronger positive correlations. A gradient boosting regression model was developed to predict daily turbidity maxima using a set of engineered meteorological and water quality features. While the model achieved moderate accuracy (RMSE = ۱۴.۲۸ on validation data), feature importance analysis revealed that lagged rainfall, humidity, and water temperature substantially contribute to turbidity prediction. These findings support the integration of weather-derived indicators in short-term water quality forecasting and underscore the role of lagged precipitation effects. This approach contributes to the development of data-driven frameworks for Integrated Water Resources Management (IWRM).

کلیدواژه ها:

Turbidity prediction ، Rainfall lag ، Brisbane River ، Water quality modeling ، Machine learning