1Yefang Jiang (yfjiang@gov.pe.ca), 1George Somers, 2Daniel Paradis, 2Martine M. Savard, 2Shawna Liao , 3Eric van Bochove
1 Prince Edward Island Department of Environment, Energy and Forestry, 11 Kent St., Charlottetown, PE, C1A 7N8, Canada
2 Geological Survey of Canada, 490 de la Couronne, Quebec, Canada
3 Agriculture and Agri-Food Canada, 2560 Hochelaga, Quebec, Canada
Abstract: 3D Groundwater flow and mass transport models are developed to assess nitrate contamination in the Wilmot River Watershed on Prince Edward Island (PEI), Canada. Vertically the flow model delineates multiple flow systems within the active flow zone and the systems roughly correspond to the stream network. Significant transverse anisotropy of the aquifer dominates the flow pattern and governs the distribution and migration of nitrate in the aquifer. The nitrate transport model shows that surface water, dominantly fed by shallow groundwater, responds relatively rapidly to nitrate leaching from the crop lands with the nitrate plume in the shallow flow systems reaching quasi-steady state within ~5 years and the nitrate plume front progressively moving into the deeper portions of the aquifer.
Key words: Nitrate modeling; groundwater; Wilmot River Watershed
Introduction
Prince Edward Island is located on the eastern coast of Canada (Fig.1). The island covers an area of 5,750 km2 and has a population of 138,100 (2005). Agriculture, fisheries and tourism are the major industries in the province.
Groundwater and surface water in some heavily farming watersheds on PEI are suffering from nitrate contamination. The island is underlain by a terrestrial “red-bed” sandstone sequence. The uppermost portion of the bedrock formations forms an unconfined/semi-confined fractured-porous aquifer. The bedrock is overlain by a thin veneer of permeable till. Potatoes are widely grown across the island. Residual fertilizer applied to the crop land leaches through the till to the water table and contaminates local groundwater. Base flow contributes significantly to stream flow and as a consequence, the contaminated groundwater discharges into the streams, resulting in surface water contamination. In many areas of the island, nitrate (N3O-N) concentrations in groundwater exceed background levels, and in some cases, the health threshold of 10 mg/l (Somers et al., 1998; Young et al., 2002). Surface water nitrate levels showed an increasing trend over time and excessive nitrate loading contributes to eutrophication in estuarine environments (Somers et al., 1998; Young et al., 2002).
Groundwater is the sole source of potable water as well as for the vast majority of the industrial water use on the island. Elevated nitrate concentrations in groundwater and a trend of increased concentrations in surface water in some heavily farming watersheds is concern both for drinking purpose and aquatic ecosystem protection. In response to this concern, a three-year collaborative project, involving the Geological Survey of Canada, PEI Department of Environment, Energy and Forestry, and Agriculture, Agri-Food Canada, was implemented in 2003 to examine nitrate sources, fate and transport in the Wilmot River Watershed, which has been selected as representative of the hydrogeological conditions on PEI and of those watersheds at risks due to agricultural activities on the island (Savard et al., 2004). An important aspect of the project is to assess land use impacts on the water environment using modeling approaches and provide input for the development of a nutrient reduction strategy on PEI . Some preliminary results from the modeling exercise are summarized here.
Hydrogeology of the Wilmot River Watershed
The Wilmot River Watershed is situated in the central west portion of the island and bears similar hydrogeology to the other parts of the island (Fig. 1). The watershed is underlain by terrestrial sandstone formations with a total thickness of > 850 m. The formation consists of a sequence of Permo-Carboniferous red beds ranging in age from Carboniferous to Middle Early Permian (van de Poll, 1983). Sandstone is the dominant rock type with a texture ranging from very fine to very coarse. Regionally, the bedrock is either flat lying or dipping gently to the east, northeast or north. There has been little structural deformation of these sediementary rocks; however, steeply dipping joints in excess 75° are common (van de Poll, 1983). The bedrock is overlain by a thin veneer of glacial deposits (1-5 m). This overburden is primarily basal till of local origin, and covers most areas of the watershed.
Fig.1 Location of the study areas (modified from Google Earth2005)
The uppermost portion of the bedrock formations forms a fractured-porous aquifer. The aquifer is characterized with significant fracture permeability predominated by horizontal bedding plane fractures. It also has an intergranular porosity. These features are evidenced by the apparent “semi-confined” and dual
porous or delayed drainage effects observed in some pumping tests in the Wilmot River Watershed. Work performed in the Winter River Watershed located ~30 km east of the present study site concluded that horizontal layering of the aquifer along with the predominance of horizontal bedding plane fractures results in a stratified aquifer with a vertical component of hydraulic conductivity ranging from one to three orders of magnitude less than horizontal values (Francis, 1989). Multi-level slug tests in the watershed have shown that hydraulic conductivity of the aquifer decreases with depth (Paradis et al., 2006), which agrees with previous findings in the Winter River Watershed. From the viewpoint of water supply, the permeability of the bedrock reduces to near negligible levels at depths of 200 m. Well yields ranged from 300 to 2000 m3/d.
Mean annual precipitation in the study areas is ~1060 mm. Most of the precipitation occurs as rain state (80%) and the rest as snow. The aquifer receives precipitation recharge at an annual mean of ~400 mmthrough the till or outcropping red beds, and discharges as base flow, evapotranspiration, pumping withdrawal and seepage at the coastline. Discharge mainly occurs along stream channels and fresh water wetlands. The water table configuration mimics topography. The aquifer demonstrates rapid hydraulic response to recharge stress from infiltration. A major recharge event due to snow melting occurs in April followed by a recession throughout the summer and early fall. A second recharge event often occurs in October or November corresponding with fall rains and lack of evapotranspiration. Current groundwater development utilizes ~10% of total annual recharge in the study areas (112 km2).
Stream-aquifer interaction is one of the key processes governing groundwater flow regime in the watershed. The Wilmot River and its tributaries discharge groundwater as seepage along most segments in the watershed. Modeling work (Jiang et al., 2004) showed base flow accounts for ~66% of annual stream flow and >80% in the late summer and fall months of many years in the watershed. The main stem of the stream is ~13.4 km in length and the stream width ranges from ~0.1 m at the head and 30.0 m at estuary segment. Sediments, comprised of a mixture of sand, silt, fine and clay, cover the bedrock/till streambed. Seepage meter measurements in the Winter River Watershed (Francis, 1989) showed vertical hydraulic gradient exists within the streambed sediments. The streambed materials, typical 1.0 to 1.5 m in thickness in the Winter River Watershed, act as weakly permeable unit that retards the hydraulic link between surface water and groundwater.
Groundwater Flow Modeling
A 3D groundwater flow model was developed using Visual MODFLOW in 2003 to assess the impacts of potential groundwater withdrawals on surface water/groundwater regimes in the Wilmot River Watershed. Details on the modeling exercise can be found in Jiang et al. (2004). Some modifications are made to the original model to accommodate the requirements for nitrate transport modeling. Vertically, the simulated thickness is expanded to 200 m and is refined from 3 into 15 layers with thicknesses varying from 10 to 30 m. Horizontal grid spacing is set at ~95 m and stream boundaries are replaced with river boundaries. The model is further calibrated based on multi-level head measurements and stream discharge data.
3D simulations suggest the horizontal hydraulic conductivity (~10-4 m/s) is 3-4 orders larger than the vertical hydraulic conductivity (10-7-10-9 m/s), which conceptually agrees with field observations and is fairly consistent with the findings in the Winter River Watershed by Francis (1989). The characteristic of significant transverse anisotropy forms a laterally-dominated flow in the aquifer. It is found that vertically multiple flow systems exist, roughly corresponding to the stream network; the shallow systems, occurring from the water table to a depth of ~50 m, discharge into the nearby tributaries and the boundaries of the deep flow systems (below ~50 m) do not necessarily follow the surface water divides as the shallow flow systems do. This model serves as the flow model for nitrate transport modeling.
Nitrate Transport modeling
Groundwater receives nitrate loading from fertilizer, mineralization-nitrification of soil organic nitrogen, sewage disposal, and atmospheric deposition. The leaching concentrations are estimated from field-scale nitrate budget analyses based on approaches and parameters in Delgdo et al. (2001), Kraft et al. (2003) and Macleod et al. (2002).
It is assumed that nitrate transport in the groundwater is controlled by the advection-dispersion processes. Based on evidences from N, H and O isotopes and hydrogeochemistry analyses (Savard et al., 2004), nitrate is further assumed non-reactive, retardation and absorption are negligible. It is well recognized that dispersivity is scale-dependent (Gelhar et al., 1992; Schulze-Makuch, 2005). In the Wilmot case, longitudinal dispersivity is set at 10 m and the ratios of horizontal/longitudinal and vertical/longitudinal dispersivities are set at 0.1 and 0.01 respectively. The effective porosity used is 0.05-0.07.
For nitrate transport modeling, advective fluxes are defined at flow sources/sinks, constant head and river boundaries (Zheng et al., 1998). Nitrate transport is modeled using MT3DMS.
Transient mass transport simulation starts from the beginning of 1965 when nitrogen fertilizer began to be intensively applied in the watershed and terminates by the end of 2100. The initial nitrate concentration in the groundwater is set at 1.0 NO3-N mg/l, which is close to assumed background levels, and leaching concentrations from different land uses from present to the year of 2100 is assumed as status quo. The maximum time step is set at 40 days. During the simulation, Generalized Conjugate Gradient Solver is selected and the implicit finite difference method with upstream weighting is applied to the advection term.
Nitrate transport modeling shows elevated nitrate concentrations in surface water/groundwater are temporally correlated with fertilizer application rates in the watershed. The simulated integrated discharge concentrations agree very well with the long-term observations in the middle reach of the Wilmot River. The simulated concentrations in the sampled wells do not match the measurements as well because of the uncertainties with the source release processes, local hydraulic properties and well construction information. The model shows surface water, dominantly fed by shallow groundwater, responds relatively rapidly to nitrate leaching from the crop lands. The nitrate plume in the shallow flow system reaches quasi-steady state within ~5 years and the nitrate plume front is moving into the deeper portions of the aquifer within the simulated time framework.
Since the year of 2000 to present, the difference of mass input and output from the aquifer in the modeled areas (~112 km2) has been estimated at ~100 kg N3O-N/day. The surplus mass is transferred into the deeper portions of the aquifer and contributes to the degradation of the quality of the less contaminated groundwater.
Conclusions
Groundwater models are developed to assess nitrate contamination in groundwater in the Wilmot River Watershed in the central west portion of Prince Edward Island. Vertically the flow model maps out multiple flow systems within the active flow zone and these systems roughly correspond to the stream network. Significant transverse anisotropy of the aquifer forms a laterally-dominated flow, which governs the distribution and migration of nitrate in the aquifer.
It is found that elevated nitrate concentrations in surface water/groundwater are temporally correlated with fertilizer application rates in the watershed; surface water dominantly fed by shallow groundwater responds relatively rapidly to nitrate leaching from the crop lands, and the plume in the shallow flow system reaches quasi-steady state within about 5 years and the plume front is moving into the deeper portions of the aquifer within the simulated time framework. This implies reducing the source input by land use changes can improve the nitrate levels in the streams within a few years and gradually reverse the trend of increasing nitrate levels in the aquifer.
Acknowledgements
We would like to acknowledge with thanks to Brad Potter for his GIS assistance and Mary Lynn McCourt and Barry Thompson for providing land use data.
References
[1] Delgdo, J., R. Riggenbach, R. Sparks, M. Dillon, L. Kawanabe and R. Ristau.2001. Evaluation of Nitrate-nitrogen transport in a potato-barley rotation. Soil Science Society of America Journal, 65:878-883.
[2] Francis, R. 1989. Hydrogeology of the Winter River Basin, Prince Edward Island. Department of the Environment, PEI.
[3] Gelhar, L., C.Welty , and K. Rehfeldt. 1992. A critical review of data on field-scale dispersion in aquifers. Water Resources Res., 28(7): 1955-1974.
[4] Jiang, Y., G. Somers and J. Mutch. 2004. Application of numerical modeling to groundwater assessment and management in Prince Edward Island. In Proc. from the 57th Canadian Geotechnical Conference and the 5th Joint CGS/IAH-CNC Conference, Quebec, Canada.
[5] Kraft, G., and W. Stites. 2003. Nitrate impacts on groundwater from irrigated-vegetable systems in a humid north-central US sand plain. Agriculture, Ecosystems and Environment, 100: 63-74.
[6] Macleod, J., J.B. Sanderson, A.Campell and G. Somers. 2002. Potato production in relation to concentrations of nitrate in groundwater: current trends in PEI and Potential management changes to minimize risk. In Proc. from National conference on agricultural nutrients and their impact on rural water quality, Waterloo, Canada.
[7] Paradis, D., J.M. Ballard, M.M. Savard, R. Lefebvre, Y. Jiang, G. Somers, S. Liao and C. Rivard. 2006. Impact of agricultural activities on nitrates in ground and surface water in the Wilmot Watershed, PEI,Canada. In Proc. from the 59th Canadian Geotechnical Conference and the 7th joint CGS/IAH-CNC Conference, Vancouver, Canada.
[8] Savard, M.M., S. Simpson, A. Smirnoff, D. Paradis, G. Somers, E. van Bochove, G. Thériault. 2004. A study of the Nitrogen cycle in the Wilmot River Watershed, Prince Edward Island: Initial Results. In Proc. from the 57th Canadian Geotechnical Conference and the 5th Joint CGS/IAH-CNC Conference, Quebec, Canada.
[9] Somers, G. 1998. Distribution and trends for occurrence of nitrate in PEI groundwater. In Proc. from nitrate-agricultural sources and fate in the Environment-Perspectives and Direction, Grand Falls, Canada.
[10] Schulze-Makuch, D. 2005. Longitudinal dispersitivity data and implications for scaling behavior. Groundwater, 43(3): 443-456.
[11] Van de Poll. 1981. Report on the Geology of Prince Edward Island. Department of Tourism, Industry and Energy, PEI.
[12] Young,J., G. Somers and B. Raymond. 2002. Distribution and trends for nitrate in PEI groundwater and surface waters, in Proc. from national conference on agricultural nutrients and their impact on rural water quality, Waterloo, Canada.
[13] Zheng, C. and P. Wang. 1998. MT3DMS documentation and User’s Guide. Waterloo Hydrogeologic, Inc.
