Investigating Hydrologic Alteration Under Historical and Future Scenarios in the Mobile River and Perdido River Basins Using the Cubist Algorithm

Study RegionMobile River and Perdido River Basins of Alabama, USA.Study FocusThis study investigates the impacts of human activities and climate variability on hydrologic alterations in the Mobile River and Perdido River Basins of Alabama. The research uses a machine learning approach to quantify and predict changes in flow duration curves under both historical (1980–2009) and future climate scenarios, Representative Concentration Pathways (RCP) 4.5 and 8.5 for two future periods (1980–2069) and 1980–2099). The models incorporate a wide range of covariates, including basin geomorphology, aquifer characteristics, land cover, water storage, environmental factors, solar radiation, and census data. The results demonstrate the efficacy of cubist models in predicting hydrologic alterations and identifying key basin characteristics influencing them. The model performance varied, with a Nash–Sutcliffe Efficiency between 0.91 and 0.94 for testing and between 0.98 and 0.99 for training during the baseline period. Future projections under RCP 4.5 and 8.5 scenarios generally showed statistically significant differences compared to the net change predictions in the baseline period. New Hydrological Insights for the RegionThe findings highlight the importance of considering different flow regimes when assessing the impacts of climate variability on streamflow in the region. This study contributes to the understanding of hydrologic alterations in the Mobile River and Perdido River Basins, providing insights for water resource management and ecological conservation efforts in the region.