Transient Simulations of Nitrogen Load for a Coastal Aquifer and

Nov 21, 2007 - Simulated time-dependent nitrogen loads at the coast can be used to correlate with current observed coastal eutrophic effects, to predi...
0 downloads 0 Views 3MB Size
Environ. Sci. Technol. 2008, 42, 207–213

Transient Simulations of Nitrogen Load for a Coastal Aquifer and Embayment, Cape Cod, MA JOHN A. COLMAN* AND JOHN P. MASTERSON United States Geological Survey, 10 Bearfoot Road, Northborough, Massachusetts 01532

Received March 14, 2007. Revised manuscript received September 17, 2007. Accepted September 19, 2007.

A time-varying, multispecies, modular, three-dimensional transport model (MT3DMS) was developed to simulate groundwater transport of nitrogen from increasing sources on land to the shore of Nauset Marsh, a coastal embayment of the Cape Cod National Seashore. Simulated time-dependent nitrogen loads at the coast can be used to correlate with current observed coastal eutrophic effects, to predict current and ultimate effects of development, and to predict loads resulting from source remediation. A time-varying nitrogen load, corrected for subsurface loss, was applied to the land subsurface in the transport model based on five land-use coverages documenting increasing development from 1951 to 1999. Simulated nitrogen loads to Nauset Marsh increased from 230 kg/yr before 1930 to 4390 kg/yr in 2001 to 7130 kg/yr in 2100, assuming future nitrogen sources constant at the 1999 land-use rate. The simulated nitrogen load per area of embayment was 5 times greater for Salt Pond, a eutrophic landward extension of Nauset Marsh, than for other Nauset Marsh areas. Sensitivity analysis indicated that load results were little affected by changes in vertical discretization and annual recharge but much affected by the nitrogen loss rate assumed for a kettle lake downgradient from a landfill.

Introduction Saltwater embayments are common features of the Massachusetts, Rhode Island, and New York coast. These embayments, typically set in highly transmissive glacial outwash sediments and lacking surface water inflow, receive freshwater and nitrogen primarily from groundwater discharge. Nitrogen is often limiting to plant growth in marine waters and is readily transported from sources on land through sand aquifers to the shore (1). Recently, coastal development and associated disposal of wastewater through septic systems have increased loads of nitrogen to coastal embayments (2), and many embayments now exhibit eutrophic effects (1, 2). Increasing eutrophication has led to investigations of nitrogen load, conducted primarily with steady-state, sourceloading models (2, 3) (but see ref 4, where the effect of aquifer travel times on changing loads from septic systems has been simulated by use of mapped travel-time bands). The steadystate nitrogen models simulate a constant nitrogen load to the coast, determined as the sum of nitrogen sources in the watershed area less losses incurred during transport. Bound* Corresponding author phone: (508) 490-5027; fax: (508) 4905068; e-mail: [email protected]. 10.1021/es070638b CCC: $40.75

Published on Web 11/21/2007

aries of the groundwater–watershed are defined for the steady-state nitrogen models from water table maps or from regional groundwater-flow models. The groundwater-flow models, coupled with solutetransport models, also could be used to assess time-varying (transient) loads of nitrogen. The transient simulations allow for time-varying inputs of nitrogen sources, such as those resulting from increasing development, and of recharge, which varies with rainfall. The transient models simulate three-dimensional solute distribution so that time-dependent estuarine loading and nitrogen depth distribution at the coast can be determined. These capabilities are necessary for model verification, correlations to estuary trophic response, predictions of future loads, and determination of the timedependent consequences of remedial actions, including sewering and associated water transfers. Transient loads are especially necessary for the analysis of contributing areas that have experienced recent development. Long aquifer travel times (up to 100 years) mean that increased nitrogen load at the coast may lag behind development of the upland by many decades. Both steady-state and transient models have similar limitations regarding the accuracy of loading data used as inputs, of loss factors for nitrogen in the aquifer, and of groundwater-flow models that relate recharge areas to discharge at the coast. Models are constrained, however, by aquifer recharge and nitrogen source rates. Thus, useful qualitative results, for example, the magnitude and timing of loads increasing or decreasing, are still possible. Here transient simulations are described for nitrogen loads to Nauset Marsh, a coastal embayment on Cape Cod, MA. Descriptions of the site and the computer models are given first followed by field methods and literature sources for data used to determine nitrogen input data for the model, to characterize the aquifer, and to verify the model. Whereas methods for new data collection are given in full, summaries are given for previously published material, with details given in the Supporting Information.

Methods Site Description. The transient models simulated nitrogen in groundwater in the watershed of Nauset Marsh, a coastal embayment on Lower Cape Cod, MA (Figure 1). The watershed is in the Nauset groundwater-flow lens, where freshwater flows over deeper saltwater, from recharge inland to discharge at the coast (5). The groundwater contributing area of Nauset Marsh includes National Seashore land, housing and commercial areas, and five freshwater seepage kettle lakes (lakes formed in glacial outwash sediments by blocks of ice broken from the retreating glacier). Description of the Models. A subregional MODFLOW model (6) (Figure 1) was used to simulate groundwater flow, based on a regional model for Lower Cape Cod developed by Masterson (5). Nitrogen transport was simulated by a time-varying, multispecies, modular, three-dimensional transport model (MT3DMS) model (7) that used flow-data output from the subregional MODFLOW model. The finite-difference grid for both subregional models consists of 180 rows and 164 columns of uniformly spaced model cells 30 m on a side. The grids have 23 layers that extend from the water table to the interface between the freshwater and saltwater flow systems. Detailed descriptions of the models are given in the Supporting Information. Water-Quality Data for Model Inputs. Nitrogen inputs for the model were computed primarily from land-use coverages; however, a nitrogen loading rate from the Visitor

Not subject to U.S. Copyright. Publ. 2008 Am. Chem. Soc.

VOL. 42, NO. 1, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 207

FIGURE 1. Location map showing the Nauset groundwater-flow lens, Nauset Marsh, the location of the subregional solute-transport model domain, and the location of sampling sites used in model verification. Center of the Cape Cod National Seashore was determined from samples collected from the distribution line of the septic leachfield and water-use data from the facility. Computed loading rates for the facility, which is just upgradient from Salt Pond (Figure 1), were used to help evaluate loads from other large septic systems in the groundwater–watershed. The septic system nitrogen load was determined from five samples collected at roughly constant time intervals from June 29, 2001, to May 23, 2002. Samples were sent on ice to the National Water Quality Laboratory (NWQL), United States Geological Survey (USGS), for analysis of total ammonia and organic nitrogen by colorimetry after microkjeldahl digestion (8), at a minimum reporting level (MRL) of 80 µg/L, and for analysis of nitrate and nitrite by colorimetry after cadmium reduction to nitrite and diazotization (9), MRL of 40 µg/L. All concentrations for nitrate and ammonia are reported in this paper as the weight of nitrogen per liter. Monthly water-use data for 2001 were obtained from the Cape Cod National Seashore Visitor Center to compute nitrogen loads. Water-Quality Data for Aquifer Geochemical Characterization and Transport-Model Verification. Two data sets were available for aquifer geochemical characterization and model verification. Samples were collected during 2001–2003 from the two 15-port multilevel samplers (MLSs; at MSL 1 and MLS 2; Figure 1) and during a previous investigation (10, 11) from 1993 to 1994 from pushpoint wells near the shore (at S1 and S2, Figure 1). Samples for nitrogen analyses were collected annually during 2001–2003, by peristaltic pump from ports of the MLS spaced from the water table to 30 m depth. Samples for iron, manganese, and dissolved organic carbon (DOC) analysis were collected in May 2002. Specific conductance (Orion model 126), dissolved oxygen, and pH (flow-through cell with Hydrolab minisonde probes) were determined on-site. Samples were filtered (0.45 µm) inline, preserved with HCl to pH 2.2, and analyzed at the North Atlantic Coastal Laboratory of the Cape Cod National Seashore for nitrate, 208

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 1, 2008

nitrite, ammonia, and in some samples, iron and manganese, or frozen before analysis of dissolved total nitrogen. Dissolved organic nitrogen was computed by the difference between dissolved total nitrogen and inorganic nitrogen species. Nitrogen species analytical methods were from (12), adapted for the Quickchem FIA+ 8000 series flow-injected Lachat Instruments autoanalyzer. The colorimetric chemistry for nitrate and nitrite analysis, MRL of 2 µg/L, was the same as that above, and that for ammonia analysis, MRL of 4 µg/L, was by indophenol formation after treatment with phenol and sodium hypochlorite. Total dissolved nitrogen was analyzed as nitrate, after an alkaline persulfate digestion (13), MRL of 5 µg/L. Analyses of iron and manganese were by flame atomic absorption spectrometry (14), MRL of 100 µg/ L. DOC samples were collected in prebaked amber glass bottles, acidified with H2SO4 to pH