Multi-objective optimisation of natural flood management for community-level property damage reduction using a physically based distributed hydrological model

Asid Ur Rehman, Elizabeth Lewis, Ben Smith, Steve Birkinshaw

Natural flood management (NFM) at the catchment scale requires cost-effective spatial configurations that maximise flood risk reduction while minimising implementation cost. Existing studies often assess predefined scenarios at catchment outlets, limiting systematic portfolio evaluation and community-scale impact assessment. This study couples the fully distributed, physically based SHETRAN hydrological model with the NSGA-II multi-objective optimisation algorithm to minimise total NFM area and community-level property damage across the Upper Irwell catchment in the UK. Four NFM types, runoff attenuation features, floodplain reconnection, woodland, and soil improvement, were represented at the grid cell scale, with volume damage avoided curves developed for 47 communities. SHETRAN achieved a Nash-Sutcliffe Efficiency of 0.78 against observed daily discharge over 13 years. Individual NFM types reduced peak flow by up to 10.4%, while a combined scenario reduced it by 15.3%. Optimisation using a 10-day event identified 100 Pareto-optimal solutions, including one that achieved the same damage avoidance as maximum NFM implementation while requiring 39.6% less intervention area. Benefit-cost ratio filtering identified 30 high-performing solutions, with damage avoidance of up to £33 million when applied across the full 35-year simulation. Efficient sub-catchments were mainly concentrated upstream. The framework enables systematic identification of cost-effective NFM portfolios across large spatial solution spaces, supporting community-scale flood risk reduction beyond scenario-based or semi-distributed hydrological approaches.