Chloride concentrations, loads, and yields in four watersheds along Interstate 95, southeastern Connecticut, 2008-11: factors that affect peak chloride concentrations during winter storms

Abstract

Chloride (Cl - ) concentrations and loads and other water chemistry characteristics were assessed to evaluate potential effects of road-deicer applications on streamwater quality in four watersheds along Interstate 95 (I–95) in southeastern Connecticut from November 1, 2008, through September 30, 2011. Streamflow and water quality were studied in the Four Mile River, Oil Mill Brook, Stony Brook, and Jordan Brook watersheds, where developed land ranged from 9 to 32 percent. Water-quality samples were collected and specific conductance was measured continuously at paired water-quality monitoring sites, upstream and downstream from I–95. Specific conductance values were related to Cl - concentrations to assist in determining the effects of road-deicing operations on the levels of Cl - in the streams. Streamflow and water-quality data were compared with weather data and with the timing, amount, and composition of deicers applied to State highways. Grab samples were collected during winter stormwater-runoff events, such as winter storms or periods of rain or warm temperatures in which melting takes place. Grab samples were also collected periodically during the spring and summer and during base-flow conditions. The estimated Cl - concentrations at the eight water-quality monitoring sites during winter storms peaked as high as 270 milligrams per liter (mg/L) and were well below the U.S. Environmental Protection Agency (EPA) recommended acute chloride toxicity criterion of 860 mg/L and the chronic 4-day average toxicity criterion of 230 mg/L. Cl - concentrations in streams, particularly at sites downstream from I–95, peaked during increased streamflow in the winter and early spring as a result of deicers applied to roads and washed off by stormwater or meltwater. Cl - concentrations during most of the nonwinter seasons decreased during increases in streamflow because storm runoff was more dilute than base flow. However, peaks in specific conductance and estimated chloride concentrations at streams in more urbanized watersheds corresponded to peaks in streamflow well after winter snow or ice events; these delayed peaks in Cl - concentration likely resulted from deicer residue that remained in melting snow piles and on roadsides and (or) that were flushed from soils and shallow groundwater, then discharged downstream. Estimated peak Cl - concentrations varied with the type of winter storm event and were highest during or after winter storms of frozen precipitation and rain, in which the rain or meltwater effectively washed off the deicers. Estimated peak Cl - concentrations correlated positively with the duration of deicer application but generally not with streamflow. Estimated peak Cl - concentrations during the winter season were highest during low streamflow at most sites. Chloride concentrations varied considerably in shallow groundwater as a result of land-use differences. Cl - concentrations were very high (as high as 800 mg/L) in shallow groundwater downstream from I–95 at the Four Mile River site. Chloride/bromide mass concentration ratios and the proximity of a former landfill and sewage lagoon upstream indicate a likely source of Cl - is landfill leachate and possibly sewage leachate. Cl - loads in streams generally were highest in the winter and early spring. The estimated daily Cl - yield for the four monitoring sites downstream from I–95 ranged from 0.0004 ton per day per square mile for one of the least developed watersheds to 0.052 ton per day per square mile for the watershed with the highest percentage of urban development and impervious area. The estimated median contribution of Cl - load from atmospheric deposition was small and ranged from 0.07 percent of Cl - load at the Jordan Brook watershed to 0.57 percent at the Oil Mill Brook watershed. The Cl - loads in streams (outputs) were compared with Cl - load inputs, which include atmospheric deposition and deicer applications; Cl - load inputs were slightly larger than the Cl - load outputs at most of the sites during most years but do not account for the Cl - load in groundwater leaving the watersheds. A multiple linear regression model was developed to describe the variability of the natural log of peak specific conductance, as well as estimated Cl - concentrations. Five significant variables best explained the variability in the natural log of the peak specific conductance after deicing events: (1) snow on ground before deicing event; (2) winter precipitation with rain; (3) specific conductance in base flow; (4) State-operated road lane miles divided by watershed area; and (5) amount of Cl - from deicers applied to State-operated roads per lane mile. In this report, winter precipitation is defined as any type of precipitation, including frozen precipitation and rain, that occurs during the active deicing season, typically November through March. Frozen precipitation is defined here as snow, sleet, freezing rain, or any winter precipitation except rain. The addition of a lane mile in both directions on I–95 would result in an estimate of approximately 2 to 11 percent increase in Cl - input from deicers applied to I–95 and other roads maintained by Connecticut Department of Transportation. The largest estimated increase in Cl - load was in the watersheds with the greatest number miles of I–95 corridor relative to the total lane miles maintained by Connecticut Department of Transportation. On the basis of these estimates and the estimated peak Cl - concentrations during the study period, it is unlikely that the increased use of deicers on the additional lanes would lead to Cl - concentrations that exceed the aquatic habitat criteria.