Natural Flood Management (NFM) through catchment roughness rehabilitation offers a nature-based approach to reducing flood peaks by desynchronising tributary and trunk stream flow contributions. This study investigates tributary roughness manipulation to achieve flood peak desynchronisation and/or attenuation in the Hunter River catchment, New South Wales, Australia. A two-dimensional (2D) rain-on-grid hydraulic model was developed in HEC-RAS 7.0 using observed gridded rainfall from the March 2022 major flood event, which recorded an observed peak water level of 13.16 m at the Singleton gauge. The model was calibrated to replicate peak flood stage, commonly used for flood warning and flood hazard assessment.
The calibrated model reproduced the observed peak with a 7.1% error (simulated peak 12.23 m) and achieved a Nash-Sutcliffe Efficiency of 0.611. Two NFM scenarios were simulated by modifying riparian vegetation roughness: (1) increasing Manning’s n across all tributaries from the calibrated value of 0.060 to 0.100, representing a uniform riparian revegetation strategy; and (2) increasing roughness only in tributaries upstream of Denman to 0.180, representing denser, targeted revegetation.
The targeted scenario achieved a 7.9% reduction in peak water level at the Wollombi Brook–Hunter River confluence, compared with 3.9% for the uniform scenario. The uniform scenario also delayed peak timing by 3 hours, indicating flood peak desynchronisation. These findings demonstrate that the spatial positioning and density of NFM interventions strongly influence downstream flood attenuation, with targeted upper-catchment roughness rehabilitation providing greater benefits per unit of intervention.