Modelling reference evapotranspiration (ET0) is an essential aspect of water resources planning and management, particularly in relation to drought and flood hazard assessments. ET0 serves as a critical component of the hydrological cycle and plays a significant role in the monitoring of drought conditions. Nevertheless, biases and uncertainties brought on by nonlinear processes, model parameterization, and inaccurate weather forecasts frequently hinder it. The conventional machine learning (ML) models often result in higher error rates as well as lower predictive accuracies due to manual tuning. In order to address this limitation, the current study has adopted an innovative, nature-inspired metaheuristic algorithm, which is referred to as the Snake Optimization Algorithm (SOA). This approach aims to improve the predictive accuracy and tuning efficiency levels of Support Vector Machines (SVM) by addressing the limitations of conventional ET0 prediction models that rely on manual or single-objective calibration at Rangpur station, operated by the Bangladesh Meteorological Department (BMD). Four statistical model evaluation metrics including Normalized Root Mean Squared Error (NRMSE), Kling-Gupta Efficiency (KGE), Mean Absolute Error (MAE), and Coefficient of Determination (R2) are used to assess the performance of each model at both training and testing stages. Among the developed models, the hybrid SCSO-SOA-SVM-II model has outperformed all other models (NRMSE = 0.021, KGE = 0.984, MAE = 0.551 and R2 = 0.944) at the testing stage. The outcomes of the current study conclusively proves that the innovative metaheuristic optimization is highly effective for the enhanced prediction of ET0 that can support effective drought monitoring and management.