Residence times and nitrate transport in ground water discharging to streams in the Chesapeake Bay Watershed

Abstract

One of the major water-quality problems in the Chesapeake Bay is an overabundance of nutrients from the streams and rivers that discharge to the Bay. Some of these nutrients are from nonpoint sources such as atmospheric deposition, agricultural manure and fertilizer, and septic systems. The effects of efforts to control nonpoint sources, however, can be difficult to quantify because of the lag time between changes at the land surface and the response in the base-flow (ground water) component of streams. To help resource managers understand the lag time between implementation of management practices and subsequent response in the nutrient concentrations in the base-flow component of streamflow, a study of ground-water discharge, residence time, and nitrate transport in springs throughout the Chesapeake Bay Watershed and in four smaller watersheds in selected hydrogeomorphic regions (HGMRs) was conducted. The four watersheds were in the Coastal Plain Uplands, Piedmont crystalline, Valley and Ridge carbonate, and Valley and Ridge siliciclastic HGMRs. A study of springs to estimate an apparent age of the ground water was based on analyses for concentrations of chlorofluorocarbons in water samples collected from 48 springs in the Chesapeake Bay Watershed. Results of the analysis indicate that median age for all the samples was 10 years, with the 25th percentile having an age of 7 years and the 75th percentile having an age of 13 years. Although the number of samples collected in each HGMR was limited, there did not appear to be distinct differences in the ages between the HGMRs. The ranges were similar between the major HGMRs above the Fall Line (modern to about 50 years), with only two HGMRs of small geographic extent (Piedmont carbonate and Mesozoic Lowland) having ranges of modern to about 10 years. The median values of all the HGMRs ranged from 7 to 11 years. Not enough samples were collected in the Coastal Plain for comparison. Spring samples showed slightly younger water under wet conditions than under dry conditions. The apparent age of water from wells, springs, and other ground-water discharge points in the four targeted watersheds was modern to 60 years, which was similar to the apparent ages from the spring study. In the Pocomoke River Watershed in the Coastal Plain Uplands HGMR, the apparent age of ground-water samples ranged from 0 to 60 years; the ages in the vicinity of the streams ranged from 0 to 23 years. The apparent ages of ground water in the Polecat Creek Watershed in the Piedmont crystalline HGMR ranged from 2 to 30 years. The apparent ages of water from wells in the Muddy Creek Watershed in the Valley and Ridge carbonate HGMR ranged from 10 to 20 years (except for a single sample that was 45 years). The ages in the East Mahantango Creek Watershed in the Valley and Ridge siliciclastic HGMR ranged from 0 to 50 years. The distribution in apparent age of water from wells in the targeted watersheds, however, generally is older than that for water from the springs. The median age of water from wells in the Muddy Creek Watershed, for example, was 15 years, compared to 11 years for the water from the springs in that watershed, and less than 10 years for water from all springs in the spring study. The similarity in the ranges in apparent age of water from the wells and from the springs shows that the samples from the targeted watersheds and springs have bracketed the range of apparent ages that would be expected in the shallow ground-water-flow systems throughout the Chesapeake Bay Watershed. The apparent age of water from individual wells does not necessarily represent the entire distribution of ages of the discharging ground water, and it is this distribution of ages that affects the response of nutrient concentrations in stream base flow. Nutrient-reduction scenarios were modeled for two watersheds for which the distribution of apparent ground-water ages was available, the East Mahantango Creek Watershed in the Valley and Ridge siliciclastic HGMR and the Locust Grove Watershed in the Coastal Plain Uplands HGMR. A nutrient-reduction scenario was created for East Mahantango Creek, where the average residence time was determined to be approximately 10 years on the basis of the output of particle tracking from a ground-water-flow model. This scenario showed decreases of nearly 50 percent in base-flow concentrations of nitrate in streams within the first year after the reduction in nitrogen input; smaller reductions in nitrate concentration occurred in each subsequent year. A second scenario for that same watershed, in which the same 10-year average residence time was assumed and an exponential model was used for analysis, showed that a 50-percent reduction in base-flow concentrations of nitrate could take up to 5 years. For the Locust Grove Watershed, in which an average residence time of 32 years was assumed, simulation with the exponential model showed that it may take more than 20 years to achieve a 50-percent reduction in base-flow concentra-tions of nitrate. Although it was not possible to construct such scenarios for all watersheds, these examples show the range of possible responses to changes in nutrient inputs in two very different types of watersheds. Findings from this study include information on factors that affect ground-water age, spatial distribution of ages, and nitrogen transport. In the East Mahantango Creek Watershed and the Polecat Creek Watershed, the residence time varied spatially depending on the position of the flow path, and temporally depending on the recharge conditions. Generally, ground water in areas near the stream had short residence times and the water in upland areas had longer residence times. Water traveling through deep layers had longer residence times than water traveling through shallow layers, and residence times were faster under high recharge conditions than low recharge conditions. Ground water in the Pocomoke Watershed exhibits a similar pattern: younger water discharges to small order streams in headwater basins and older water discharges to larger streams near the basin outlet. Factors affecting nitrogen transport in ground water include spatial and temporal variation in input sources, ground-water age, and aquifer processes that lead to denitrification. Spatial and temporal variations in nitrogen sources affect all the watersheds. Tributaries with higher inputs of nitrogen have higher concentrations in stream base flow. Areas where nitrogen application rates have increased over time show an age-nitrate relation in ground-water samples. The age-nitrate relation can be affected by denitrification, which occurs in Pocomoke and East Mahantango Creeks but is not evident in Polecat and Muddy Creeks. In East Mahantango Creek, the level of denitrification is significant in water with residence times greater than 20 years, but because this is a small component of overall ground-water discharge to a stream, it may not remove a significant quantity of nitrogen from the system. Denitrification in Pocomoke Creek is significant and appears to affect mostly older water discharging to streams. Therefore, if most of the nitrogen entering these two streams is associated with the discharge of younger ground water, denitrification may not greatly affect the overall nitrogen delivery to these streams. Other findings of this study show that nitrate in ground water discharging along preferential flow paths may not be affected by natural processes, such as denitrification or uptake by riparian vegetation. Seeps to swales and ditches beneath the north uplands at Polecat Creek indicate a shallow water table and discharge of young ground water whereas the absence of such seeps on the south side indicates a deep water table and a lack of young ground water. Similarly, discharge at the base of the slope and to the valley wetland south of the creek but not north of the creek indicates a different role for the riparian forest on the two sides of the creek. In many of the systems where water discharges at the base of slopes to wetlands, ditches have been dug to drain the valley. Such drainage circumvents possible removal of nitrate by riparian vegetation. Because ground-water residence times do not appear directly related to the HGMRs, the targeting of management practices will achieve the most rapid response in water quality if directed at 1) watersheds with large agricultural sources of nitrate, 2) areas with the shortest ground-water-flow paths and 3) areas not affected by significant denitrification. The fastest response in stream base-flow concentrations of nitrogen to implementation of management practices would be to implement practices in those areas with the highest loads rather than attempt to target practices on the basis of HGMR stratification. Overall findings of the study indicate that 1) ground-water contributions to nitrogen in streamflow are significant, 2) some response to management practices should be evident in base-flow concentrations of nitrogen and loads within 1 to 5 years in watersheds with the shortest average residence times, but response time may be closer to 20 years in watersheds with longer average ground-water residence times, 3) the majority of the response in ground-water discharge to any changes in management practices will be distributed over a 10-year time period even in the watersheds with the fastest response times, and 4) given that half the streamflow is from ground-water discharge and the other half is runoff or soil water, about 90 percent of total water being discharged to a stream will be less than about a decade old; therefore, full implementation of nutrient reductions may result in improved streamwater quality in about a decade. In the more-likely scenario of gradual source reduction, the reduction in concentrations of nitrate in streams and aquifers would take longer than the examples shown here.