Accurately estimating river streamflow necessitates reliable uncertainty quantification in stage-discharge rating curves (SDRCs). This study investigates three methods (Quasi-Newton + covariance matrix (QN + CM), Generalized Likelihood Uncertainty Estimation (GLUE), and Sequential Uncertainty Fitting Ver-2 (SUFI-2)) to analyze uncertainty in isovel contour-based SDRCs and evaluate their performance. By leveraging hydraulic parameters and velocities derived from isovel contours, SDRCs can be established. The model was calibrated using observational data, its performance assessed during the validation period, and its accuracy verified in the testing period. The optimal number of observation data was determined based on the model’s performance in the validation phase. Correlation coefficient (CC) and mutual information (MI) were employed to identify influential hydraulic parameters. The Trent River, characterized by a complex cross-sectional geometry including a floodplain, was selected as a case study. Results indicated that GLUE outperformed SUFI-2 in both the validation and testing periods, achieving superior streamflow estimation (NSE = 0.97, R2 = 0.97, PBIAS=-0.14, and RSR = 0.17). Based on the P/R criterion for assessing uncertainty analysis accuracy, GLUE (P/R = 2.50) and SUFI-2 (P/R = 2.57) significantly outperformed QN + CM. Consequently, calibrating the model using GLUE or SUFI-2 enhances both estimation and uncertainty analysis accuracy. This research also contributes to identifying outliers in streamflow time series.

Keywords