HYDROLOGICALLY CONNECTED FLOODPLAINS AND WOOD-RICH STREAMS PROMOTE GREATER ATTENUATION OF FLASH FLOODS

Abstract

Flash floods are a significant threat, destroying infrastructure and claiming lives. Management of river corridors within the past ~200 years has mainly involved building in areas which frequently flood and then attempting to control floods through local infrastructure projects. We have accelerated floodwater through channels (increasing conveyance) in populated areas and stored flood water behind dams to reduce flooding. Increasing floodwater conveyance is achieved by two main methods, often done in tandem. The first method, termed “snagging,” removes obstructions to flow including wood, riparian vegetation, and boulders. The second method of increasing conveyance is channelization, wherein the river channel is deepened and widened. These practices come at a heavy cost monetarily and environmentally. Global efforts have been made to explore alternatives to these historical flood-control practices. One of the leading ideas is termed “Natural Flood Management,” which involves strategically restoring the natural ability of rivers and floodplains to attenuate floods with highly obstructed wood-rich channels and well-connected floodplains. The rapid floods with high intensity but a relatively low total volume experienced by these systems make them prime candidates for this style of management. I quantified the flood control benefits provided by wood-rich streams with connected floodplains in Colorado headwater systems. To fully understand the ability of healthy rivers in Colorado to attenuate flash floods, I studied them at multiple scales. Firstly, I created a highly physics-based two-dimensional model of three reaches of streams to understand which processes are key to adequately represent the ability of a reach to attenuate flow. I then created novel simplified models which encapsulated the important processes I identified with the 2-D hydrodynamic models while being easily applied at large scales. Finally, I used this newly developed model to represent a full watershed in northern Colorado to test the impacts of different management strategies on the propagation of flash floods throughout the system. I quantified flood attenuation at the reach scale utilizing two-dimensional hydrodynamic models simulating flash floods in three stream reaches located in the Colorado Rocky Mountains of the United States. I quantified the (a) magnitude of attenuation, (b) total accessible floodplain volume, (c) volume of floodwater stored in floodplain depressions, (d) variability of flow path travel times, (e) floodplain heterogeneity, and (f) relative importance of these mechanisms in flood attenuation. I found unprecedently high discharge attenuation with an average 13.8% reduction in peak flow per kilometer reach length and continued attenuation up to the 100-year recurrence interval flood. For the studied sites, the strongest correlations were between attenuation and storage in floodplain depressions. Flow path diversity metrics correlated best with attenuation for floods with a time-to-peak greater than 1 hr. My findings also indicated that maintenance of high floodplain roughness and accessibility are effective strategies for bolstering attenuation of flash floods at the reach scale in mountain systems. To extend these findings to larger scales, I developed five numerical models varying in detail and physical realism for these same streams to identify a simple, effective, and efficient method of routing floods. These models included spatially uniform 1D hydrodynamic models and four variations of Muskingum-Cunge hydrologic routing models. For each model I compared output hydrographs using Kling Gupta Efficiency, Nash Sutcliffe Efficiency, difference in peak flow attenuation, and differences in predicted total volume retained within the models. I found that standard Muskingum-Cunge routing methods are insufficient in reproducing flood hydrographs because they underpredict water retention within the floodplain and overpredict peak outlet discharge. The model that most effectively reproduced flood routing when compared to 2D hydrodynamic models was a new method I developed to account for floodplain storage of water, which I refer to as the Muskingum-Cunge Augmented Floodplain (MCAF) method. This method scales the volume of water retained on the floodplain by the overall inundated floodplain volume. My results indicate that hydrologic routing algorithms, such as the classic Muskingum-Cunge approach, need better methods to account for mass storage of water in floodplains to accurately predict the attenuation of flash floods in mountain streams, and my Muskingum-Cunge Augmented Floodplain routing scheme improves predictions of flood routing. Finally, I applied the MCAF method in a full watershed hydrologic model of the Elkhorn Creek watershed in northern Colorado. The studied watershed has recently been the site of reach- scale wood additions. I installed hydrologic instruments throughout the watershed and observed two flash floods driven by convective storms over two years. I used the larger of the floods to calibrate a Monte-Carlo event-based watershed model and validated model performance on the smaller observed flood. This produced six behavioral models which I used to simulate floods with different river corridor management scenarios by changing routing of flood waters within the river corridor parameters. I simulated four routing scenarios: the existing conditions, snagged conditions, conditions before the recent restoration efforts, and complete restoration of in-channel wood. The simulated snagging had drastic impacts on the system, accelerating the flood wave propagation through the system and increasing the peak discharge across the watershed. The largest impacts of snagging were at the outlet of the watershed, with a mean 34.9% increase in peak discharge and a 1.44-hour reduction in time to peak. Adding more obstructions to the channel minimally impacted flood propagation, with a small, inconsistent delay in peak timing and no meaningful reduction in peak magnitude. The recent wood additions contributed to the attenuation of the site, with an average 5.2% reduction in peak flows in the headwater reaches after implementation. In certain flood scenarios the recent wood additions reduced peak discharges by up to 35%. My results indicate that maintaining wood supply in mountain streams may reduce the risk of deadly flash floods through reductions in peak discharge magnitude and delay flood propagation.