Spatial Variability of Groundwater Recharge and its Effect on Shallow Groundwater Quality in Southern New Jersey

ity in the Glassboro, NJ area (Fig. 1). The Darcian method was used to estimate groundwater recharge Point estimates of groundwater recharge at 48 sediment-coring from water-retention parameters and unsaturated hylocations vary substantially ( 18.5–1840 cm yr 1) in a 930-km2 area of southern New Jersey. Darcian estimates of steady, long-term recharge draulic conductivity on the basis of sediment texture made at depth in the unsaturated zone were estimated using pedoand moisture content data obtained near the water table transfer functions of soil texture and interpolated (mapped) with at 48 locations in the study area. The recharge estimates nonparametric methods to assess aquifer vulnerability in the area. were geostatistically analyzed to evaluate the spatial The probability of exceeding the median recharge (29.1 cm yr 1) is variability of measured sediment properties, to map relow in the southwestern and northeastern portions of the study area charge with respect to land use, and to derive statistical and high in the eastern and southeastern portions. Estimated recharge distributions of recharge at specific locations in the study is inversely related to measured percentage clay and positively related area. The recharge estimates were compared with soils to the percentage of well-drained soils near wells. Spatial patterns of and topographic data to determine whether recharge recharge estimates, exceedance probabilities, and clay content indicate could be accurately predicted from landscape characterthat sediment texture controls recharge in the study area. Relations with land elevation and a topographic wetness index were statistically istics. Finally, recharge estimates were compared with insignificant. Nitrate concentration and atrazine (6-chloro-N2-ethylconcentrations of NO3 and atrazine to evaluate potential N4-isopropyl-1,3,5-triazine-2,4-diamine) percentage detection in sameffects on the quality of shallow, recently recharged ples of shallow groundwater (typically 10 m) are higher for low groundwater. In this study, “shallow groundwater” rerecharge sites ( 29.1 cm yr 1) than for high recharge sites ( 29.1 cm fers to depths 10 m. The depth of the screened interval yr 1) in agricultural and urban areas. Differences between high and below water in observation wells in the area is about low recharge sites in these areas are highly significant for NO3 concen3 m. The objectives of the study were to tration, but not for atrazine concentration. • evaluate the spatial variability of point estimates of ground-water recharge, • map recharge with respect to land use, and N contamination is considered the • compare recharge estimates with NO3 and atrazine single greatest threat to water quality (Corwin et concentrations in shallow groundwater. al., 1997). Preventing contamination of groundwater is crucial in areas where it is a major source of public and domestic supply. Knowing where an aquifer is vulneraMATERIALS AND METHODS ble to surface-derived contaminants would help managDescription of Study Area ers prioritize scarce resources for alternative management practices, monitoring, and cleanup. The study area (Fig. 1) comprises about 930 km within the Coastal Plain Physiographic province of southern New Jersey. Aquifer vulnerability studies at large spatial scales Population in the area has increased from about 50 000 people have used index methods, such as DRASTIC and SEEPin 1940 to about 250 000 in 2000. Groundwater withdrawals AGE (Navulur and Engel, 1996), or overlays made with from the surficial, Kirkwood-Cohansey aquifer system, which geographic information systems (GISs) (Nolan et al., consists of highly permeable unconsolidated sands and gravels, 1997). Both DRASTIC and SEEPAGE underestimated have recently increased to meet the growing demand for drinkcontamination potential by describing areas with high ing water. As of 1986, the Glassboro area comprised 21% NO3 concentration as low risk (Navulur and Engel, urban land, 26% agricultural land, and 39% undeveloped land 1996). Index and overlay methods provide only limited (Stackelberg et al., 1997). understanding of processes controlling the transport of The outcrop of the Kirkwood Formation, a confining unit water and chemicals in the unsaturated zone. Alternaabout 30 m thick, underlies the aquifer and forms the northwest boundary of the study area. Aquifer thickness increases tively, deterministic models can simulate water and to about 75 m at the southeastern boundary (Zapecza, 1989). chemical fluxes, but the spatial variability of sediment Unsaturated zone sediment in the study area consists mainly properties at field scales and above limits accuracy and of the Cohansey Sand, which was deposited during the Mioimposes large uncertainty on model predictions. cene Age on inner shelf, nearshore, and beach areas during In the current study, we used a combined determinisslow retreat of the sea. Sediments in the Cohansey Sand genertic–geostatistical approach to assess aquifer vulnerabilally are coarser at shallower depths, which is consistent with similarly deposited formations in the New Jersey Coastal Plain B.T. Nolan, U.S. Geological Survey, 413 National Center, Reston, (Zapecza, 1989). The Bridgeton Formation overlies the CoVA 20192; A.L. Baehr and L.J. Kauffman, U.S. Geological Survey, West Trenton, NJ. Received 27 Nov. 2002. Original Research Paper. Abbreviations: CCDF, conditional cumulative distribution function; *Corresponding author ([email protected]). DEM, digital elevation model; DO, dissolved oxygen; GIS, geographic information system; IGF, Indicative Goodness of Fit; IK, indicator Published in Vadose Zone Journal 2:677–691 (2003).  Soil Science Society of America kriging; KED, Kriging with external drift; MLR, multiple linear regression; PTF, pedotransfer function. 677 S. Segoe Rd., Madison, WI 53711 USA