Sensitivity of HEC-RAS 2D Predictions to Vegetation-Related Manning’s n in Patchy Vegetated Channels

Vegetation can substantially alter flow behavior in streams by increasing hydraulic resistance, modifying momentum exchange, and generating localized flow structures. In two-dimensional flow modeling, representing vegetation effects remains challenging because model setups typically require estimating multiple spatially distributed resistance parameters from vegetation patterns and parameterizing heterogeneous vegetation traits. Among these, the flow-resistance coefficient for the vegetated section is the most sensitive yet uncertain parameter, and it is frequently simplified or assigned empirically. Therefore, this study investigates the sensitivity of two-dimensional model predictions of water-surface elevation and velocity to the flow-resistance coefficient in a vegetated channel.
Numerical simulations were performed using HEC-RAS 2D and calibrated against a large-scale flume experiment conducted in a straight channel with evenly spaced willow patches along the centerline. Channel topography was reconstructed from high-density point-cloud data and resampled into digital elevation model datasets at a 1 mm grid resolution. The vegetated channel was simulated under both high- and low-flow conditions using three vegetation patch configurations: group-dense, single-dense, and single-sparse. Flow resistance within the vegetated areas was represented by spatially distributed Manning’s n values in the HEC-RAS 2D model. Model results obtained using Manning’s n values derived from a momentum-based model were compared with those obtained using manually calibrated Manning’s n values.
The results show that, for the group-dense configuration, applying Manning’s n values estimated by the momentum-based model led to overestimation of both water-surface elevation and flow velocity relative to the large-scale experiment data. Manning’s n values manually calibrated to match the experimental observations were lower than those estimated by the momentum-based model. For the single-patch configurations, both dense and sparse cases consistently underestimated water-surface elevation, whereas flow velocities were overestimated across all tested Manning’s n values. Overall, the study shows that the performance of vegetation-related Manning’s n in two-dimensional hydraulic models varies with vegetation density and patch configuration. The observed differences between group and single-patch vegetation highlight potential limitations in representing vegetation resistance solely through a Manning’s n parameter under spatially heterogeneous conditions.

Acknowledgement: This research was funded by the Korea Environment Industry & Technology Institute (KEITI) through the Smart Water-supply Service Research Program, funded by the Korea Ministry of Climate, Energy, Environment (MCEE)(RS-2022-KE002091).