A channel stability assessment and logistic regression model for a reach of muddy creek below Wolford Mountain Reservoir, in north-central Colorado

Abstract

Water resource managers face increasing pressure to meet community water needs while responsibly managing resource infrastructure and preserving aquatic and riparian ecosystems. Management of many western rivers involves multiple uses for multiple stakeholders, especially rivers downstream from dams impounding water-supply reservoirs. These resource management issues arise in a reach of Muddy Creek near Kremmling, Colorado, which serves multiple functions including: (1) wetland mitigation, (2) private, recreational fishing, (3) and agriculture. Alteration of streamflow from reservoir operations and loss of upstream sediment supply, in conjunction with legacy management effects, have resulted in channel instability and increased streambank erosion in Muddy Creek below Wolford Mountain Reservoir. A typical response to channel instability on managed rivers includes installation of erosion-control structures. However, installation of these structures to protect property and infrastructure is expensive and can have unintended consequences at adjacent locations, highlighting the need for resource managers to better understand the underlying geomorphic processes controlling channel adjustment within the reach. To address these issues, field reconnaissance and channel surveying completed during base-flow conditions were used to (1) determine the dominant erosive and resistive processes within the reach that contribute to channel stability and response, and (2) assess the validity of using logistic regression techniques as an analytical framework and to estimate the probability or risk of localized streambank erosion. These findings can be used in conjunction with local management objectives to evaluate or gage acceptable risk to current infrastructure and to target and prioritize where monitoring or remediation should be conducted. Understanding the geomorphic processes and reach characteristics driving streambank erosion can be used to guide management and operational decisions within the reach to minimize impacts. A map of probability of erosion, for each streambank, is presented which shows risk (as a probability of streambank erosion, ranging from < 3 to 80 percent) based on significant explanatory variables from the logistic regression model. The study found that stream-induced scour and undercutting have differing effects within the reach due to changes in the erosive power of the stream and relative difference in streambank and riparian characteristics. Areas most susceptible to streambank erosion occurred in wider cross sections where fluvial energy was oriented into the streambanks, not necessarily in areas with the greatest fluvial energy and potential erosive power (i.e., areas with the steepest bed slope). This suggests additional, localized conditions within the reach need to be considered. Differences in streambank and riparian characteristics were shown to have varying levels of resistance to streambank erosion within the study reach. Larger streambank heights increased the probability of streambank erosion when these streambanks were not supported by bedrock outcrops of Pierre Shale or alluvial fans and talus slopes. Erosion-control structures decreased the probability of streambank erosion where structures retained original positions relative to flow. Where changes to flow orientation occurred, the probability of streambank erosion around these structures increased substantially. Riparian vegetation type also influenced streambank erosion as well as channel top-width. Streambanks covered in willows were found to decrease the risk of streambank erosion, whereas areas dominated by grasses increased streambank erosion potential as well as increased channel top-widths relative to areas dominated by willows. Additional effects from reach-scale characteristics were evaluated. Areas of greater sinuosity and wider valley widths show increased probability of streambank erosion, as well as in areas located downstream of an irrigation diversion structure. This may be due to a combination of effects from further confinement of the reach as valley widths decrease, increased streamflow from tributaries, and/or the occurrence of increased seepage along areas near an irrigation ditch. Streambanks with observed saturated soils were also found to have between 20-44 percent increased odds of streambank erosion. A linkage between proximity of streambank seeps and unlined irrigation ditches and irrigation turn-outs was highly significant, with potential effects extending to distances up to 250 m from the irrigation water sources.