The Stephens Creek Watershed in southwest Portland, Oregon was chosen by the city as a pilot project for urban stream restoration efforts, and the infiltration of stormwater was identified as a potential restoration strategy. The Stephens Creek Watershed has historically been known to be unstable during high precipitation events (Burns, 1996), and the need to address the response of slope stability to anthropogenically-driven changing groundwater conditions is the focus of this study. Airborne light detection and ranging (LiDAR) and geotechnical data from the City of Portland were employed to create a high resolution (0.84 m2) physics-based probabilistic slope stability model for this watershed, using the map-based probabilistic infinite slope analysis program PISA-m (Haneberg, 2007). Best and worst case models were run using fully dry and fully saturated soil conditions, respectively. Model results indicate that 96.3% of the watershed area had a probability [less than or equal to] 0.25 that the slope factor of safety (FOS) was [less than or equal to] 1 for fully dry conditions, compared to 76.4% for fully saturated conditions. Areas that had a probability [greater than or equal to] 0.25 that the slope factor of safety (FOS) was [less than or equal to] 1 were found to occur mainly along cut/fill slopes as well as within the deeply incised canyons of Stephens Creek and its tributaries. An infiltration avoidance map was derived to define areas that appear to be unsuitable for infiltration. Based on these results, it is recommended that stormwater continues to be directed to existing sewer infrastructure and that the "storm water disconnect" restoration approach not be used by the city.

Particles mobilized by stormwater negatively affect receiving surface waters. Stormwater best management practices (BMPs) can reduce solids along with associated pollutants in runoff but engineers and environmental managers have been long vexed by the problem of choosing the optimal BMP for a given situation. A common BMP process for solids removal is sedimentation. This thesis addresses the question of whether the effectiveness (and thus choice) of a sedimentation device can be estimated (and thus optimized) from the particle size properties of runoff, which, in turn, could be associated with specific runoff zones or land uses. Presented here is a series of experiments to determine the solids-removal capabilities of a manufactured oil-water separator that also removes solids via sedimentation. A statistical model developed from the experimental data shows that, under normal operating conditions, influent particle size can be used to accurately estimate effluent total suspended solids (TSS) for BMPs of this type. Relationships between particle size and particle-bound metal concentrations for Cu, Zn and Pb were then obtained from the literature and incorporated into the model to allow estimates of metal removal efficiencies based on TSS and PSD. The model can be used with an arbitrary particle size distribution (PSD); this allows effluent quality predictions to be made considering that particle sizes entering stormwater BMPs could vary due to anthropogenic, hydraulic or hydrologic factors. To place these experimental and modeling results in the context of an urban environment, samples of deposited stormwater solids were collected from residential areas, commercial areas and an industrial zone in Portland, Oregon, and the PSD of each sample was determined using light obstruction particle sizing. PSDs ranging over sizes from 3μm to 200μm vary among these locations. Areas with high anthropogenic impact were found to have PSDs skewed toward the smallest particle sizes. The statistical model developed here was then used to show that the effluent quality of the BMP tested would differ depending on the locations where solids were collected. The evidence presented in this thesis thus indicates that device performance will correlate with geographic locations or land use zone and validates further investigation into delineating the City of Portland's characteristic runoff zones and using the runoff characteristics of each zone to map it to the most desirable treatment practices.