<p>A reliable parameter identification approach in hydrogeological models can reduce prediction uncertainty while enhancing model robustness and reliability. Parameter ranges could directly influence uncertainty, and a reasonable reduction of the parameter space could reduce the number of local optima and avoid unnecessary computational effort. In this paper, we introduced a strategy that employed the Self-Organizing Map (SOM) to intelligently navigate the parameter space of the MODFLOW model. Our approach constructed a hypothetical model in the Taoerhe area, Jilin Province, China. First, we defined a set of hydrogeological parameters across six zones and conducted the forward run of the model to obtain the observed head within nine wells. Then, the SOM method was employed for range optimization over two rounds, constructing Particle Swarm Optimization (PSO) and SOM-PSO schemes for comparison. Finally, we evaluated the parameter estimates, groundwater head simulations, and uncertainty. Our results showed that the average relative error of the best estimates decreased from 24.27% to 14.89% after optimizing the parameter range. Additionally, SOM-PSO outperformed PSO in both Nash-Sutcliffe Efficiency (NSE) and Mean Absolute Error (MAE), demonstrating a better alignment with the observed data trends and lower error across nine wells. Furthermore, the number of unidentified parameters in the behavioral parameter sets was reduced, and SOM-PSO demonstrated better performance in parameter estimation. Moreover, output uncertainty gradually decreased as the parameter range was continuously optimized. The proposed method had the potential to serve as an alternative parameter identification approach for hydrogeological modeling.</p>