[1] Recharge rates of nitrate (NO 3 À ) to groundwater beneath agricultural land commonly are greater than discharge rates of NO 3 À in nearby streams, but local controls of NO 3 À distribution in the subsurface generally are poorly known. Groundwater dating (CFC, 3 H) was combined with chemical (ions and gases) and stable isotope (N, S, and C) analyses to resolve the effects of land use changes, flow patterns, and water-aquifer reactions on the distributions of O 2 , NO 3 À , SO 4 = , and other constituents in a two-dimensional vertical section leading from upland cultivated fields to a riparian wetland and stream in a glacial outwash sand aquifer near Princeton, Minnesota. Within this section a ''plume'' of oxic NO 3 À -rich groundwater was present at shallow depths beneath the fields and part of the wetland but terminated before reaching the stream or the wetland surface. Groundwater dating and hydraulic measurements indicate travel times in the local flow system of 0 to >40 years, with stratified recharge beneath the fields, downward diversion of the shallow NO 3 À -bearing plume by semiconfining organic-rich valley-filling sediments under the wetland and upward discharge across the valley and stream bottom. The concentrations and d 15 N values of NO 3 À and N 2 indicate that the NO 3 À plume section was bounded in three directions by a curvilinear zone of active denitrification that limited its progress; however, when recalculated to remove the effects of denitrification, the data also indicate changes in both the concentrations and d 15 N values of NO 3 À that was recharged in the past. Isotope data and mass balance calculations indicate that FeS 2 and other ferrous Fe phases were the major electron donors for denitrification in at least two settings: (1) within the glacial-fluvial aquifer sediments beneath the recharge and discharge areas and (2) along the bottom of the valley-filling sediments in the discharge area. Combined results indicate that the shape and progress of the oxic NO 3 À plume termination were controlled by a combination of (1) historical and spatial variations in land use practices, (2) contrast in groundwater flow patterns between the agricultural recharge area and riparian wetland discharge area, and (3) distribution and abundance of electron donors in both the sand aquifer and valley-filling sediments. The data are consistent with slow migration of redox zones through the aquifer in response to recharging oxic groundwater during Holocene time, then an order-of-magnitude increase in the flux of electron acceptors as a result of agricultural NO 3 À contamination in the late twentieth century, to which the redox zone configuration still may be adjusting. The importance of denitrification for NO 3 À movement through formerly glaciated terrains should depend on the source areas and depositional environments of the glacial sediments, as well as geomorphology and recent stream-valley sediment history.

Summary Regional ground-water recharge estimates for Minnesota were compared to estimates made on the basis of four local-and basin-scale methods. Three local-scale methods (unsaturated-zone water balance, water-table fluctuations (WTF) using three approaches, and age dating of ground water) yielded point estimates of recharge that represent spatial scales from about 1 to about 1000 m 2 . A fourth method (RORA, a basin-scale analysis of streamflow records using a recession-curve-displacement technique) yielded recharge estimates at a scale of 10-1000s of km 2 . The RORA basin-scale recharge estimates were regionalized to estimate recharge for the entire State of Minnesota on the basis of a regional regression recharge (RRR) model that also incorporated soil and climate data. Recharge rates estimated by the RRR model compared favorably to the local and basin-scale recharge estimates. RRR estimates at study locations were about 41% less on average than the unsaturated-zone water-balance estimates, ranged from 44% greater to 12% less than estimates that were based on the three WTF approaches, were about 4% less than the age dating of ground-water estimates, and were about 5% greater than the RORA estimates. Of the methods used in this study, the WTF method is the simplest and easiest to apply. Recharge estimates made on the basis of the UZWB method were inconsistent with the results from the other methods. Recharge estimates using the RRR model could be a good source of input for regional ground-water flow models; RRR model results currently are being applied for this purpose in USGS studies elsewhere. ª

[1] At many sites contaminated with petroleum hydrocarbons, methanogenesis is a significant degradation pathway. Techniques to estimate CH 4 production, consumption, and transport processes are needed to understand the geochemical system, provide a complete carbon mass balance, and quantify the hydrocarbon degradation rate. Dissolved and vapor-phase gas data collected at a petroleum hydrocarbon contaminated site near Bemidji, Minnesota, demonstrate that naturally occurring nonreactive or relatively inert gases such as Ar and N 2 can be effectively used to better understand and quantify physical and chemical processes related to methanogenic activity in the subsurface. In the vadose zone, regions of Ar and N 2 depletion and enrichment are indicative of methanogenic and methanotrophic zones, and concentration gradients between the regions suggest that reaction-induced advection can be an important gas transport process. In the saturated zone, dissolved Ar and N 2 concentrations are used to quantify degassing driven by methanogenesis and also suggest that attenuation of methane along the flow path, into the downgradient aquifer, is largely controlled by physical processes. Slight but discernable preferential depletion of N 2 over Ar, in both the saturated and unsaturated zones near the free-phase oil, suggests reactivity of N 2 and is consistent with other evidence indicating that nitrogen fixation by microbial activity is taking place at this site.

The fate of hydrocarbons in the subsurface near Bemidji, Minnesota, has been investigated by a multidisciplinary group of scientists for over a quarter century. Research at Bemidji has involved extensive investigations of multiphase flow and transport, volatilization, dissolution, geochemical interactions, microbial populations, and biodegradation with the goal of providing an improved understanding of the natural processes limiting the extent of hydrocarbon contamination. A considerable volume of oil remains in the subsurface today despite 30 years of natural attenuation and 5 years of pump-and-skim remediation. Studies at Bemidji were among the first to document the importance of anaerobic biodegradation processes for hydrocarbon removal and remediation by natural attenuation. Spatial variability of hydraulic properties was observed to influence subsurface oil and water flow, vapor diffusion, and the progression of biodegradation. Pore-scale capillary pressure-saturation hysteresis and the presence of fine-grained sediments impeded oil flow, causing entrapment and relatively large residual oil saturations. Hydrocarbon attenuation and plume extent was a function of groundwater flow, compound-specific volatilization, dissolution and biodegradation rates, and availability of electron acceptors. Simulation of hydrocarbon fate and transport affirmed concepts developed from field observations, and provided estimates of field-scale reaction rates and hydrocarbon mass balance. Long-term field studies at Bemidji have illustrated that the fate of hydrocarbons evolves with time, and a snap-shot study of a hydrocarbon plume may not provide information that is of relevance to the long-term behavior of the plume during natural attenuation.