This study adopted the Shiono-Knight model (SKM) to estimate the lateral distribution of the depth-averaged velocity within rectangular and trapezoidal compound channels with emergent vegetation in floodplains. To implement the SKM, it was required to estimate the eddy viscosity coefficient, friction coefficient, and secondary flow coefficient. The present study estimated the friction coefficient using the Colebrook–White equation modified by Rameshwaran and Shiono for vegetated beds. An analysis of eddy viscosity models across compound channels indicated that the model was not sensitive to the eddy viscosity coefficient; thus, the eddy viscosity coefficient could be assumed constant across the channel. However, the negligence of the secondary flow in the model would lead to a significant error, and it was required to calibrate the secondary flow coefficient. Thus, this study used a genetic algorithm (GA) to develop equations for the secondary flow coefficient for different sections of the compound channel under two different approaches: (1) the approach of Abril and Knight (2004), who proposed constant values for the main channel and floodplains, and (2) the equations of Devi and Khatua (2017), which related the secondary flow coefficient to the relative depth and width ratio. It was found that the secondary flow coefficient was dependent on the relative depth and width ratio. As a result, the equation optimized based on the Devi-Khatua approach outperformed the Rameshwaran-Shiono technique in estimating the lateral distribution of the velocity, with a 10.2% lower error. This paper employed SKM to estimate the depth-averaged velocity within three compound channels of rectangular and trapezoidal cross-sections with a vegetated floodplain at small and large scales. To solve the SKM, it was required to calculate the friction coefficient, eddy viscosity coefficient, and secondary flow coefficient. The friction coefficient was calculated using the modified Colebrook–White equation. Several eddy viscosity models were adopted to implement the SKM. It was found that the eddy viscosity coefficient had no significant effect on the performance of SKM. The present study focused on calibrating the secondary flow coefficient as it played a key role in the flow simulation of compound channels using SKM. Two approaches were adopted to calibrate the secondary flow coefficient: (1) the approach of Abril and Knight (2004) and (2) the approach of Devi and Khatua (2017). The latter defines the secondary flow coefficient as a function of the relative depth and width ratio. The optimal secondary flow coefficient was obtained using a GA and experimental data for different geometric and hydraulic conditions. A comparison of the predicted and observed velocities demonstrated that the Devi-Khatua calibration method improved the predictive accuracy of SKM by nearly 10.2%. The secondary flow coefficient was found to be dependent on the relative depth and width ratio. It was calculated to be positive in both the main channel and floodplain, suggesting clockwise secondary flows. The difference between the observed and predicted velocities was larger in the floodplain than in the main channel, which could have arisen from flow complexities around vegetation.