Accurate calibration of hydrological models with evolutionary computation multi-method ensembles

Climate change significantly impacts the hydrological cycle, posing challenges for regional water resource management and Sustainable Development Goals. Hydrological modelling is essential for planning and adaptation, being necessary to dispose different mathematical computing tools to be able to calibrate them properly. This study introduces the Dynamic Probabilistic Coral Reefs Optimization algorithm with Substrate Layer (DPCRO-SL), a multi-method ensemble approach, to enhance hydrological model calibration. The algorithm was applied to the $abcd$ hydrological model in two Spanish river basins, using lumped and semi-distributed structures to test adaptability. Results were compared with the SCE-UA algorithm, a benchmark in hydrology, using metrics such as Nash–Sutcliffe Efficiency (NSE), Mean Squared Error (MSE), Kling–Gupta Efficiency (KGE), and Percent Bias (Pbias). For instance, in THRB basin during the test period, the proposed DPCRO-SL algorithm achieved NSE = 0.765, KGE = 0.875, MSE = 171.9, and Pbias = 2.6, whereas the reference SCE-UA algorithm obtained NSE = 0.689, KGE = 0.748, MSE = 227.8, and Pbias = –14.8. DPCRO-SL consistently outperformed SCE-UA, especially in test scenarios reflecting forward projections. These findings underscore the potential of DPCRO-SL as a robust tool for hydrological modelling and climate adaptation, offering improved accuracy and reliability in model calibration.