Modeling the hydrologic and particulate loadings from paved urban surfaces

This dissertation details the results of recent investigations concerning non-colloidal particulate matter found in urban rainfall-runoff and urban snowmelt. Three forms of the power law model (PLM) were used to model particle number density. It was found that significant model error could be introduced into PLMs when the median diameter particle was used to estimate the number of particles per interval, but the SSE of continuous PLMs could be significantly reduced through the use of correction factors when the sieve interval was large (P\u3c0.05). Additionally, a multiple PLM analysis may be more appropriate than a single PLM analysis when changes in particle population occur within a single gradation. PLMs were also used to model the relationship between cumulative granulometric mass and cumulative particulate-bound metal mass. The use of such calibrated PLMs provided considerable cost reduction when compared to conventional direct measurements of particulate-bound metal mass. A first flush analysis indicated that only events with average volumetric flow rates approaching 1 L/min/m of drainage width exhibit a rapid depletion of particulate matter consistent with the concentration-based definition of the first flush. An analysis of particle separation mechanisms indicated that 90% of these particles, by mass, could be removed from the discharge with detention times of 30 min and 120 min for snowmelt and rainfall-runoff, respectively, in a typical roadside drainage system. It was also found that the kinematic wave model could accurately model significant aspects of rainfall-runoff events in traffic-impacted watersheds provided that abstractions were incorporated into the modeling process. For this watershed the source of abstraction was attributable to vehicular traffic. For high-intensity events with more than 10 vehicles per runoff volume (VPV) the runoff coefficient varied between 0.6 and 0.9 while for low-intensity events with fewer than 10 VPV the coefficient ranged between 0.2 and 0.4