Accurate prediction of river flows in ungauged basins is critical for effective water resource management and planning. However, the limited availability of observed streamflow data in such basins poses a significant challenge for the precise calibration of hydrologic models. This study presents a methodology that leverages a semi-distributed hydrologic model, specifically the SWAT model, to predict river flows in ungauged basins by regionalizing model parameters.

The proposed method involves deriving probability distribution functions (PDFs) for sensitive model parameters based on data from gauged basins. These PDFs are then related to the physical characteristics of the ungauged basin to extrapolate parameter values. The regionalized parameters are employed to simulate streamflows in ungauged locations. The model was calibrated using observed data from 2000-2004, yielding R² values between 0.70 and 0.86, and Nash-Sutcliffe Efficiency (NSE) values between 0.62 and 0.79. Validation for the period 2005-2008 resulted in slightly lower R² values, ranging from 0.58 to 0.73, and NSE values from 0.34 to 0.64, indicating reasonable performance despite data limitations.

This study demonstrates that the regionalization of model parameters based on their probability distributions can improve confidence in streamflow simulations for ungauged basins. The approach was tested using two watersheds, and the results suggest that the method is a viable option for continuous watershed modeling in ungauged basins. The methodology provides a framework that balances the need for hydrologic predictions in data-scarce environments with the uncertainties inherent in parameter regionalization, contributing to improved water resource planning and management.