The use of best management practices or BMPs to treat urban stormwater runoff has been pervasive for many years. Extensive research has been conducted to evaluate the performance of individual BMPs at specific locations; however, little research has been published that seeks to evaluate the impacts of small, distributed BMPs throughout a watershed at the regional level. To address this, a model is developed using EPA SWMM 5.0 for the Duck Pond watershed, which is located in Blacksburg, Virginia and encompasses much of the Virginia Polytechnic and State Institute's campus and much of the town of Blacksburg as well. A variety of BMPs are designed and placed within the model. Several variations of the model are created in order to test different aspects of BMP design and to test the BMP modeling abilities of EPA SWMM 5.0. Simulations are performed using one-hour design storms and yearlong hourly rainfall traces. From these simulations, small water quality benefits are observed at the system level. This is seen as encouraging, given that a relatively small amount of the total drainage area is controlled by BMPs and that the BMPs are not sited in optimal locations. As expected, increasing the number of BMPs in the watershed generally increases the level of treatment. The use of the half-inch rule in determining the required water quality volume is examined and found to provide reasonable results. The design storm approach to designing detention structures is also examined for a two-pond system located within the model. The pond performances are examined under continuous simulation and found to be generally adequate for the simulated rainfall conditions, although they do under-perform somewhat in comparison to the original design criteria. The usefulness of EPA SWMM 5.0 as a BMP modeling tool is called into question. Many useful features are identified, but so are many limitations. Key abilities such as infiltration from nodes or treatment in conduit flow are found to be lacking. Pollutant mass continuity issues are also encountered, making specific removal rates difficult to define. / Master of Science

This study focused on development of a continuous watershed-scale modelling approach capable of evaluating sediment removal performance of vegetative filter strips (VFS). This was done by integrating the single-event hydrologic and sediment transport model AGNPS with the event-based VFS model (VFSMOD) applying the methodology developed by Sebti and Rudra (2010), and also through the development and incorporation of sub-models capable of describing changes in hydrologic conditions between rainfall events into the integrated models. For modeling purposes, the buffer zone is divided to segments called “buffer cells”. The upstream source area corresponding to each buffer cell and the flow-path connecting the area to the stream are identified, and runoff and sediment generated within each area is simulated for each event applying AGNPS. Using VFSMOD, performance analysis of VFS is conducted for each buffer cell. By applying the developed “continuous simulation” sub-models the hydrologic conditions prior to each event were determined.