Climate-driven intensification of stratification and hypolimnetic deoxygenation in Oberon reservoir: A decadal analysis

Oberon Reservoir (OR), a temperate monomictic water body in Australia, displays strong seasonal stratification that has intensified over the past decade in response to climate change. In this study, a nine-year dataset (2016–2025) from OR reveals pronounced seasonal cycles characterised by stable summer stratification with a thermocline near 10 m, isolating bottom waters and restricting mixing, thereby intensifying hypolimnetic oxygen depletion. Surface water temperatures (WTs) peaked at 20 °C, while hypoxic conditions prevailed in deeper layers. Vertical pH gradients emerge, driven by surface photosynthetic CO₂ uptake, leading to alkalinization, in contrast to bottom-water acidification due to respiratory CO₂ accumulation. These redox conditions promote elevated concentrations of iron (Fe) and manganese (Mn) in anoxic bottom waters through reductive dissolution of metal oxides, concurrently releasing soluble Fe 2+ , Mn 2+ , inorganic phosphorus (P), and ammonium (NH₄ + -N), thereby intensifying internal nutrient loading. Regression analysis confirmed strong correlations between thermocline strength index (TSI) and chemical stratification index (IC-i), indicating tight coupling between thermal and chemical stratification. Although winter mixing temporarily restored homogeneity, it did not counteract the cumulative effects of prolonged stratification. Analysis of the past decade of data shows a steady increase in reservoir surface temperatures accompanied by a consistent decline in bottom dissolved oxygen. Seasonal Autoregressive Integrated Moving Average (SARIMA) forecasting further projects intensified thermal stratification, with surface warming of approximately +2.5 °C by 2030, raising concerns about worsening hypoxia, enhanced nutrient and metal release, and increased acidification. Model performance for WT, DO, pH, Fe, Mn, inorganic P, and NH₄ + -N demonstrated strong predictive accuracy, evidenced by low mean error (ME), root mean square error (RMSE), and mean absolute error (MAE) values, symmetric mean absolute percentage errors (SMAPE) generally below 10% (and < 20% for highly variable constituents), mean absolute scaled errors (MASE)