Effects of selected low-impact-development (LID) techniques on water quality and quantity in the Ipswich River Basin, Massachusetts: Field and modeling studies

During the months of August and September, flows in the Ipswich River, Massachusetts, dramatically decrease largely due to groundwater withdrawals needed to meet increased residential and commercial water demands. In the summer, rates of groundwater recharge are lower than during the rest of the year, and water demands are higher. From 2005 to 2008, the U.S. Geological Survey, in a cooperative funding agreement with the Massachusetts Department of Conservation and Recreation, monitored small-scale installations of low-impact-development (LID) enhancements designed to diminish the effects of storm runoff on the quantity and quality of surface water and groundwater. Funding for the studies also was contributed by the U.S. Environmental Protection Agency’s Targeted Watersheds Grant Program through a financial assistance agreement with Massachusetts Department of Conservation and Recreation. The monitoring studies examined the effects of (1) replacing an impervious parking lot surface with a porous surface on groundwater quality, (2) installing rain gardens and porous pavement in a neighborhood of 3 acres on the quantity and quality of stormwater runoff, and (3) installing a 3,000-square foot (ft2) green roof on the quantity and quality of stormwater runoff. In addition, the effects of broad-scale implementation of LID techniques, reduced water withdrawals, and water-conservation measures on streamflow in large areas of the basin were simulated using the U.S. Geological Survey’s Ipswich River Basin model.

From June 2005 to 2007, groundwater quality was monitored at the Silver Lake town beach parking lot in Wilmington, MA, prior to and following the replacement of the conventional, impervious-asphalt surface with a porous surface consisting primarily of porous asphalt and porous pavers. Changes in the concentrations of the water-quality constituents, phosphorus, nitrogen, cadmium, chromium, copper, lead, nickel, zinc, and total petroleum hydrocarbons, were monitored. Increased infiltration of precipitation did not result in discernible increases in concentrations of these potential groundwater contaminants. Concentrations of dissolved oxygen increased slightly in groundwater profiles following the removal of the impervious asphalt parking lot surface.

In Wilmington, MA, in a 3-acre neighborhood, stormwater runoff volume and quality were monitored to determine the ability of selected LID enhancements (rain gardens and porous paving stones) to reduce flows and loads of the above constituents to Silver Lake. Flow-proportional water-quality samples were analyzed for nutrients, metals, total petroleum hydrocarbons, and total-coliform and Escherichia coli bacteria. In general, when all storms were considered, no substantial decreases were observed in runoff volume as a result of installing LID enhancements. However, the relation between rainfall and runoff did provide some insight into how the LID enhancements affected the effective impervious area for the neighborhood. A decrease in runoff was observed for storms of 0.2 inches (in.) or less of precipitation, which indicated a reduction in effective impervious area from approximately 10 percent to about 4.5 percent for the 3-acre area. Water-quality-monitoring results were inconclusive; there were no statistically significant differences in concentrations or loads when the pre- and post-installation-period samples were compared. Three factors were probably most important in minimizing differences: (1) the small decrease in effective impervious area, (2) the differences in the size of storms sampled for water-quality constituents before and after installation of the infiltration enhancing measures, and (3) small sample sizes.

In a third field study, the characteristics of runoff from a vegetated “green” roof and a conventional, rubber-membrane roof were compared. The amount of precipitation and the length of the antecedent dry period were the two primary factors affecting the green roof’s water-storage capacity. The green roof retained more than 50 percent of the precipitation from storms with 0.04 to 1.0 in. of rain. Approximately 95 percent of the precipitation from one storm of nearly 2 in. was retained by the green roof. On the rubber-membrane roof, only a small, shallow puddle of insubstantial volume ever remained after a storm. Bulk precipitation from 10 storms was monitored for the same constituents (nutrients, metals, and total petroleum hydrocarbons) as the roof runoff, and the results were compared with those for roof-runoff samples. The use of fertilizers to help establish the vegetation during the study probably distorted any effect the plants and growing medium may have had on the retention of target analytes. As a result of the fertilizer and growing medium chemistry, median concentrations of total nitrogen, total phosphorus, cadmium, copper, and nickel in runoff from the green roof were greater than in the runoff from the conventional roof or in bulk precipitation. Concentrations of lead and zinc were greater in runoff from the conventional roof, probably a result of passage through the old, metal drainpipes.

Simulations of the effects of LID on streamflow in the Ipswich River Basin were conducted with a previously calibrated Hydrological Simulation Program-FORTRAN (HSPF) precipitation-runoff model. Simulations were conducted at multiple spatial scales to evaluate the effects of (1) updated water withdrawals for the towns of Reading and Wilmington; (2) potential land-use changes at buildout (potential future development); (3) effective impervious area reductions upstream from the South Middleton streamgage to represent the effects of widespread implementation of LID retrofit techniques; (4) basin-scale water withdrawal reductions scaled up (expanded to the town level) from water-conservation pilot programs conducted by the Massachusetts Department of Conservation and Recreation; and (5) land-use change and LID techniques at a local scale, which is smaller than the HSPF subbasin. Effects on streamflow generally were evaluated by comparing results of two or more related simulations for selected reaches in the basin; thus, relative rather than absolute changes in simulated flow were the focus of the assessment. Simulations indicated that reduced withdrawals for the towns of Reading and Wilmington led to substantially higher medium and low flows in most of the reaches upstream from the South Middleton streamgage. Simulations of water-conservation measures resulted in negligible effects on streamflow.

Overall, simulations indicated that spatial scale is an important factor in determining the effects of land-use change and LID practices on streamflow. Potential land-use changes at buildout had modest (percent differences of less than 20 percent) effects on streamflow in most subbasins because relatively little land in the basin was available for development (about 17 percent); moreover, most of the available open land is zoned for low-density residential development, and this land-use category was simulated to contain relatively little effective impervious area and to be similar hydrologically to the forested land in place prior to development. Results of the simulations conducted to evaluate widespread effective impervious area reductions upstream from the South Middleton streamgage indicated that the percentage of urban land use and associated effective impervious area was too small for a 50-percent reduction of effective impervious area to appreciably affect streamflow (percent differences of less than 20 percent) in most subbasins. In contrast, the results of the hypothetical local-scale simulations indicated that for smaller streams, where the percentage of urban land use and associated effective impervious area in the drainage area may be substantially higher, land-use change, development patterns, and LID practices potentially have much greater effects on streamflow.

Modeling results also indicated that LID was potentially most beneficial for minimizing streamflow alteration when applied to dense urban development, largely because larger tracts of effective impervious area were available for reduction than were available for other land-use categories. For example, commercial-industrial-transportation land use is composed of 37 percent pervious area and 63 percent effective impervious area in the HSPF model, whereas low-density residential area is composed of 97.5 percent pervious area and only 2.5 percent effective impervious area.

Field and modeling studies concurred in the assessment that LID enhancements would likely have the greatest effect on decreasing stormwater runoff when broadly applied to highly impervious urban areas. A measurable effect for small rainfall events (less than 0.25 inch) was determined in the small, highly pervious area that was monitored in this study, but the volume difference was not great.