Search

Ample weather variability and uncertainties induced by climate change increase the likelihood of underperformance or failure of existing flood mitigation structures, pressing engineers and water resources managers to elaborate alternative strategies to traditional ones. Distributed flood attenuation is a flexible, adaptable and environmentally sustainable approach relying on multiple attenuating features, such as small reservoirs, offline ponds, woody barriers, dense vegetation, and green infrastructure. Features are distributed across a catchment and individual attenuation effects combine to attenuate flooding at the catchment scale. Existing literature has not clarified whether cumulative attenuation is observable at the larger scale., This dissertation presents a methodology for systematically studying the effectiveness of distributed attenuation. I propose a framework for modeling attenuating features at a large number of locations. Two ideas motivate the framework. First, the flexibility of attenuating features allows modelers to locate them in small tributaries and on secondary flowpaths, increasing the number of design options. Second, feature locations determine how attenuating effects combine downstream and eventually the cumulative catchment-scale attenuation. Given a large pool of potential locations, a modeler can select distributions of features and estimate their attenuation power using a hydrologic model., In the second part of the dissertation, I frame feature location siting as a spatial optimization problem in which optimal solutions maximize flow peak reduction at the catchment outlet and minimize feature construction cost. I describe a genetic algorithm to search for and discover a set of solutions/distributions that express the trade-off between flow attenuation and monetary cost., The third and final part of the dissertation investigates the spatial characteristics that make some distributions optimal. I describe a multi-scale optimization method, in which scale guides the optimization process to improve the quality of the search but also to identify spatial patterns among optimal distributions. At a small scale, separate optimization processes occur in distinct study area partitions. Then, results and partitions are aggregated, and the optimization continues at increasingly larger scales up to the scale of the whole study area., I applied the above methods for the siting of small reservoirs in the Hewett Creek watershed, in Northeast Iowa. The main finding is the algorithm’s success in identifying optimal reservoir locations, mainly upstream and midstream, and a broad trade-off function between attenuation and cost. The availability of many potential locations, produced within the framework, and the genetic algorithm’s ability to combine them multiple times were key factors for discovering advantageous and diverse distributions of reservoirs. Multi-scale optimization improved the original optimization, found superior solutions, and highlighted interactions among reservoirs in different drainage areas that increase the cumulative effect of flood attenuation., This research is a foundational step for a novel watershed-scale strategy for flood management and establishes a path for its practical implementation. Future research should further explore the quantity and quality of cumulative attenuation effects within a stream network. Full understanding of these effects will lead to better design of distributed flood attenuation systems, and eventually to their adoption as a viable and sustainable approach.