Determining the effects of vegetation on levee structural integrity on the Green River in King County, Washington

Levee vegetation management has been a challenge in King County due to conflicting federal mandates. Since the early 1990s King County has been incorporating bioengineering techniques into levee repairs. These techniques entailed the use of vegetation to not only provide bank stability but to also improve conditions for salmon as required under the Endangered Species Act. However, the U.S. Army Corps of Engineers required the removal of any vegetation over two inches in diameter to remain eligible for federal emergency funding under Public Law 84-99, hypothesizing that it compromised levee stability as well as hindered inspections. While there have been many studies on vegetation's effect on stability in natural systems, less information is available on the effect of vegetation on the structural stability of levees and revetments.;This research investigates the effect of vegetation on levee structural integrity on the Green River in King County, Washington State. Initially, the investigation set out to compare levees repaired using bioengineering techniques with levees repaired using traditional rock. However, no digital information was available on bioengineered levee projects on the Green River that could be used to develop a study plan. The first project task involved organizing files and documenting the institutional knowledge at King County on the levee projects along the Green River. Fifty-four bioengineered repairs on the Green River were visited and geo-referenced. Based on information gleaned from the data mining and mapping exercises, it became clear that comparisons of stability of bioengineered versus non-bioengineered levees could not be made. As such, a retrospective pilot study was undertaken to directly address the issue of the role of vegetation in levee stability. Using a case control method, 12 documented levee sites that incurred damage during November 2006 flooding were matched with undamaged sites on a one to one basis. Basic land cover type (trees, shrubs, grass, bare ground, impervious surfaces) was delineated through aerial photography taken prior to the flooding and percent cover of each type was calculated from these measurements.;Study results reveal the paired differences between percent cover of trees, impervious surfaces, and bare ground were not statistically significantly different between damaged and control sites. However, of the paired sites where trees were present on both the case (damaged site) and the control, the control generally had more tree cover than the case. Shrub cover was statistically higher on damaged sites than on control sites. but it was not possible to determine if shrub cover was native or non-native.;Post-hoc power analysis indicates that a much larger sample size of at least 54 matched cases and controls would be needed to determine a more scientifically defensible and statistically robust result for the effect of tree cover on levee damage. The steps taken for the pilot study could be duplicated in any larger study. However, it is unlikely that all damages could be satisfactorily matched with bioengineered sites on the Green River given the number of variables to be considered when matching. The results from this study show the complexity of trying to isolate a single factor contributing to levee stability.