Construction and calibration of numerical ground-water-flow models of the western half of the Milford-Souhegan glacial-drift aquifer, Milford, New Hampshire

Abstract

The Milford-Souhegan glacial-drift (MSGD) aquifer, in southcentral New Hampshire, is an important source of water for industrial, commercial, and domestic use that provides more than 2.7 million gallons of water per day in 1994. A large plume of volatile-organic compounds (approximately 0.5 square miles in area) covers a large part of the western half of the MSGD aquifer and threatens existing groundwater usage. The plume area has been designated a superfund site and named after a former municipal water-supply well (Savage Well) that was discontinued because of contamination. A 3-year study by the U.S. Geological Survey and the U.S. Environmental Protection Agency began in January 1994 to examine the temporal variability of groundwater flow in the contaminant plume and adjacent areas of the western half of the MSGD aquifer. This report summarizes construction and evaluation of threedimensional, steady state and transient numerical groundwater flow models that will be used to help design a remediation plan for containing volatile organic contaminants. The results of the steady-state simulation of average-flow conditions from June 1994 to June 1995 show a close comparison between modelcomputed heads and observed heads at more than 70 wells. Most heads are within 1.5 feet of observed heads. The standard-mean-head error is-0.12 feet, indicating that computed heads are slightly lower than observed heads. The absolutemean-head error is 0.59 feet, and the root-meansquare error is 0.89 feet. The results of the transient model of seasonal-flow conditions also show a close comparison between computed-head fluctuations and observed-head fluctuations for the same wells. The largest difference in computed and observed heads occurs for the fall of 1994 (standard-mean-head error of 0.29 feet). Computed variations in direction and slope of maximum hydraulic gradients compare well with observed variations for most areas of the model. The analysis, however, identified several areas of the model where further testing is suggested. These areas include those (1) near the source area of contaminants, (2) in the northcentral part of the plume, and (3) at the northern leading edge of contaminants near the confluence of two rivers.

Effects of a remedial system and its operation on volatile organic compound-contaminated ground water, Operable Unit 1, Savage Municipal Well Superfund Site, Milford, New Hampshire, 1998-2004

“…Seasonal trends in direct infiltration recharge from precipitation and recharge from river leakage cause directional variations in flow that affect capture of the contaminated ground water by the existing exterior extraction wells as shown by numerical simulations of transient hydrologic conditions. Monitoring data indicate that maximum gradients and recharge from the losing river reach in OU1 are generally during low water-table conditions (Harte, 1999). Therefore, if exterior extraction at EW-1 and EW-2 is turned off during low recharge periods (summer), then less, relatively clean ground water from the river area is captured.…”

“…The amount of aquifer recharge from the losing river reach in OU1 affects ground-water-flow direction outside the barrier. Harte (1999) showed that maximum gradients occur during low water-table conditions in the aquifer because aquifer recharge from the river reach in OU1 is highest during this period. When recharge is highest from the river, more flow occurs from the river area toward EW-1 and EW-2.…”