Journal of Hydrology 292 (2004) 210–228 www.elsevier.com/locate/jhydrol Landscape controls on the hydrology of stream riparian zones Philippe G.F. Vidon*, Alan R. Hill Department of Geography, York University, 4700 Keele Street, Toronto, Ont. Canada M3J 1P3 Received 10 December 2002; revised 15 December 2003; accepted 8 January 2004 Abstract Increased knowledge of hydrology is essential to an understanding of the water quality function of stream riparian zones. We examined the effect of upland surficial aquifer size, topography and riparian sediment lithology on the subsurface hydrology of eight riparian sites on glacial till and outwash landscapes in southern Ontario, Canada. Riparian sites had a permanent subsurface hydrologic connection to the adjacent upland in landscapes where upland permeable sediment depths were 2 – 15 m. In contrast, a hydrologic connection was absent in summer and autumn at riparian sites with ,2 m of permeable sediments overlying an aquitard. Riparian zones linked to thicker and more extensive upland aquifers had large relatively constant groundwater inputs that maintained a stable riparian water table. Sites that were seasonally disconnected from uplands had large annual water table drawdowns that dried out the riparian area unless the water table was sustained by the stream. Peats with low hydraulic conductivity and a thinning of highly permeable sediment layers occurred in some riparian zones producing upward groundwater flow that created seeps and surface rivulets. Subsurface flow data indicated a consistent downslope flow path in areas of the riparian zone where the slope gradient was . 5%. Where the riparian zone was level to gently sloping (,5%), subsurface flow directions were influenced by stream water level. Riparian subsurface flows at these sites often shifted from a hillslope to stream pattern to a down valley and even a stream to hillslope direction during summers. A conceptual framework of riparian hydrologic types based on hydrogeologic setting is developed to generalize the hydrology of riparian zones at the landscape scale. q 2004 Elsevier B.V. All rights reserved. Keywords: Riparian zone; Groundwater hydrology; Soil lithology; Topography; Water table; Hydrogeology 1. Introduction During the past 20 years, the fate of agricultural contaminants in stream riparian zones has been extensively studied (Peterjohn and Correll, 1984; Lowrance, 1992; Haycock and Pinay, 1993; Jordan et al., 1993; Gilliam, 1994; Hill, 1996). These studies * Corresponding author. Tel.: þ1-416-7362100x88620; fax: þ 1416-736-5988. E-mail address: [email protected] (P.G.F. Vidon). 0022-1694/$ - see front matter q 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2004.01.005 have shown that in order to understand and predict the fate of contaminants in riparian zones, we first need to understand riparian zone hydrology. The hydrology of stream riparian zones is strongly influenced by the landscape hydrogeologic setting (Winter 1992; Brinson, 1993; Lowrance et al., 1997; Baker et al., 2001). This encompasses the location of the riparian zone in the catchment in relation to surface and groundwater flows as well as the geological characteristics such as topography, stratigraphy and hydraulic properties of sediments that control hydrology. P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 The size of upland aquifers controls the magnitude and seasonality of subsurface flow and elements inputs to riparian areas (Hill, 2000). The threshold between riparian zones that are seasonally and permanently connected to uplands may be the result of small differences in average depth of upland storage in some landscapes (Devito et al., 1996). The size and seasonality of upland hydrologic connection also influences riparian zone water table fluctuations and the extent of surface saturation with consequent effects on soil microbial redox reactions and water chemistry (Roulet, 1990; Hill and Devito, 1996; Devito and Hill, 1997). The depth and permeability of saturated sediments overlying a confining layer in riparian zones can influence riparian hydrologic flowpaths, water residence times and renewal rates (Schnabel et al., 1994; Correll et al., 1997). A confining layer at a shallow depth increases the interaction of subsurface flow with vegetation root systems and surface soils and enhances the potential of the riparian zone for nitrate removal (Hill, 1996). In riparian zones where a confining layer is absent, groundwater flow at depth may bypass the riparian zone and discharge directly to the stream channel (Bohlke and Denver, 1995). Gravel layers in the soil profile beneath less permeable sediments can be zones of preferential subsurface flow across the riparian zone (Burt et al., 1999). Conversely, sediments with low hydraulic conductivities (e.g. peats, clays in oxbow channels) can divert water upward or downward altering subsurface flowpaths in the riparian zone (Hill, 1990; Brusch and Nilsson, 1993; Burt, 1997). Research also indicates that topography affects the hydrological functioning of riparian zones (Devito et al., 2000a). The slope gradient, especially at the riparian zone-upland margin, influences the hydraulic gradient and the volume and velocity of water entering the riparian zone. A concave riparian profile typically favours interaction between subsurface water and superficial soil horizons where denitrification and root uptake are likely to occur, whereas convex topography favours deeper water tables and subsurface flowpaths. Burt et al. (2002) have shown that when the riparian zone is flat, the stream water level acts as a fixed point around which the water table fluctuates, whereas when the slope directly borders 211 the river, the effect of stream water level on water table dynamics in the riparian zone is negligible. Although the importance of landscape hydrogeologic setting is recognized, researchers have not studied whether characteristics such as upland aquifer thickness, topography and riparian lithology can be used together to understand and predict the hydrological functioning of riparian zones. In order to develop a better understanding of how these landscape characteristics control riparian zone hydrology, we need to conduct a comparative analysis of riparian zones in landscapes that differ in hydrogeologic characteristics. With the exception of a recent study of water table fluctuations in European riparian zones by Burt et al. (2002), researchers have examined single riparian sites. In this study, we examine the effect of upland aquifer size, topography and riparian sediment lithology on the subsurface hydrology of eight riparian sites located on glacial till and outwash landscapes in southern Ontario, Canada. The riparian zones were selected to provide a range of sediment permeability, slope gradients and upland aquifer depths. The comparison of the hydrology of these riparian zones focuses on two main questions: (1) what is the influence of increasing upland aquifer size on the magnitude and duration of upland –riparian hydrologic linkages? (2) how do hillslope discharge, topography and riparian sediment lithology interact to influence water table fluctuations, riparian groundwater flow patterns and fluxes entering the riparian zones? Our understanding of riparian hydrology is used in the concluding section of the paper to develop a conceptual framework of different classes of riparian groundwater dynamics based on hydrogeologic setting to generalize about riparian zone hydrology at the landscape scale. 2. Study sites The study sites are located in agricultural catchments in southern Ontario near Toronto (Fig. 1). The annual precipitation in the area is 800– 900 mm/yr with 120 –240 mm falling as snow between December and April (Singer et al., 1997). The mean annual temperature is 7.2 8C with a mean January temperature of 2 6.7 8C and a mean July 212 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 Fig. 1. Location of study sites in relation to geomorphic regions. (1) Glacial till, (2) sand plain, (3) clay plain, (4) spillways and gravel terraces, (5) Kame moraine. The dashed line indicates the Niagara escarpment. Letters identify riparian sites. E, Eramosa; B, Boyne; R, Road 10; S, Speed; M, Maskinonge; G, Ganatsekiagon; H, Highway 27; V, Vivian. temperature of 20.5 8C. There are frequent mid-winter thaws but spring snowmelt in the March –April period is the main runoff period. Lowest stream discharge occurs in July –August when evapotranspiration is highest. Considerable weather variations occurred during the 2000– 2002 study period. Rainfall in May and June 2000 was more than two times higher than the 30 year normal for these months. Snowfall in the winter of 2000 – 2001 was 34% above the long-term average and the snow cover duration of 104 days was the longest ever recorded. Rainfall from June to August 2001 was only 45% of normal for this period and represents the driest summer recorded in the past 50 years. Winter precipitation in 2001 – 2002 was 22% below normal and the mean temperature of 4.5 8C above normal was the warmest recorded in the past 140 years in southern Ontario. Summer 2002 ranked as the eighth driest summer of last 50 years with 15.4% less precipitation than normal. Two of the riparian sites are located to the west of the Niagara escarpment on the Eramosa and Speed rivers, 4th and 2nd order streams respectively, that flow in former glacial meltwater channels, bordered by extensive gravel and cobble terraces (Fig. 1). The Road 10 and Boyne river riparian sites are located on the Alliston sand plain that forms an unconfined 9 – 12 m thick aquifer underlain by a thick sequence of silts and clays (Devito et al., 2000b). The Boyne River is a fourth order stream draining into the Nottawasaga River and the Road 10 site is adjacent to a first order tributary of the Nottawasaga River. The other riparian sites are located on glacial till (Maskinonge, Ganatsekiagon and Highway 27) or outwash silt (Vivian). The Highway 27 site is located along an intermittent first order tributary of the Nottawasaga River. The Maskinonge and Vivian Creek sites border second order streams in catchments that drain into Lake Simcoe. The Ganatsekiagon site is on an intermittent first order tributary of Duffin’s Creek, that flows into Lake Ontario. The Boyne and Eramosa riparian zones are forest sites dominated mainly by white cedar (Thuja occidentalis L.). The other riparian sites are covered with an herbaceous plant community with scattered shrubs and deciduous trees. All riparian sites are located downslope from fertilized cropland. 3. Methods An extensive network of wells and piezometer nests was installed extending from the field-riparian zone margin to the stream at each riparian site. For most sites, two separate transects of wells and piezometers were installed 20– 80 m apart in order to assess site variability and subsurface flow paths at the riparian zone scale (Fig. 2). Each piezometer nest consisted of piezometers constructed from 1.27 cm ID PVC pipe with 20 cm long slotted ends installed at depth between 0.5 and 5.5 m depending on sites. Groundwater wells (ID 5.1 cm ABS pipe 1.5– 2 m long) perforated throughout their length were installed at most piezometer nests and at various other locations within the riparian sites. When necessary, a bentonite clay seal was used to prevent contamination by surface water. Riparian zone lithology was first determined from visual inspection of cores collected by hand augering during piezometer and well installation. Soil samples P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 213 Fig. 2. Topography of the riparian zones showing the location of piezometer nests. The dashed line indicates the field edge. (a) Eramosa; (b) Boyne; (c) Road 10; (d) Speed; (e) Maskinonge; (f) Ganatsekiagon; (g) Highway 27; (h) Vivian. Contour elevations are in meters above an arbitrary datum. were later collected and the different fractions of sand and the clay-silt fraction were separated by sieving and then weighted to determine the soil texture. Saturated hydraulic conductivities were measured in piezometers and wells using the Hvorslev water recovery method (Freeze and Cherry, 1979). For each site, the confining layer defined as a horizon with a low hydraulic conductivity (# 1026 cm s21) restricting subsurface flow in the riparian zone and adjacent upland, was determined using a combination of soil survey and hydraulic conductivity distribution data. When the confining layer was deeper than the deepest point surveyed in the field (3 – 5 m depending on sites), geology and water resource reports were used to estimate the confining layer depth (Deane, 1950; Sibul and Choo-Ying, 1971). Topographic maps were used to estimate the length of the upslope contributing area to each 214 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 riparian zone. Both the depth of permeable sediments and the slope length in the upland were used as indicators of upland aquifer size. Groundwater level and hydraulic heads were measured at least once a month beginning in March 2000 (Maskinonge, Ganatsekiagon, and Vivian sites), in May 2000 (Highway 27 site), in June 2000 (Boyne and Eramosa sites) and July 2000 (Road 10 and Speed sites) until September 2002 for all sites. The topography of the riparian zone and the adjacent upland was determined using a total station. Equipotential lines for the vertical cross-sections were determined by triangulation by hand. The anisotropy ðKh =Kv Þ of sand and gravel is often approximately 10 (Barwell and Lee, 1981) and would be higher for finer texture sediments and range from 0 to 1000 for peats (Chason and Siegel, 1986). The vertical exaggerations on the riparian cross-sections (Figs. 3 and 6) vary between 3 and 14 depending on sites and are therefore, approximately equivalent to the transformed isotopic section for an overall anisotropy of 10– 100 (Freeze and Cherry, 1979). Topographical and water table contour maps for plan view illustrations of riparian sites were determined using Surfer 7 mapping software (Golden Software, 1999). Water fluxes entering each riparian zone at the perimeter were calculated using Darcy’s Law: Q ¼ Ks ðdh=dlÞA ð1Þ where Q is the water flux (l day21); Ks is the saturated hydraulic conductivity (m day21) and dh=dl is the hydraulic gradient comprising the change in hydraulic head ðhÞ between two adjacent equipotential lines with distance ðlÞ along the direction of flow. Variations of Ks in relation to lithology in piezometer nests adjacent to the riparian perimeter were used to calculate fluxes for each layer of the soil profile. The hydraulic heads used to calculate the hydraulic gradient were based on the initial several nests along transects near the riparian perimeter. Temporary shallow piezometers were also installed in the field at the Maskinonge, Vivian, Ganatsekiagon and Highway 27 sites to better estimate the hydraulic gradient. For the riparian zones where equipotentials were approximately vertical, the groundwater flow was calculated for each soil layer extending from the water table for that date to the top of the confining layer. The area A of each of these layers parallel to equipotential lines was estimated as the product of the vertical depth times a width of 1 m. Fluxes for each of the layers were added to provide an estimate of the groundwater flux entering each riparian zone. Subsurface water flows with an upward component at the riparian perimeter of the Eramosa and Maskinonge sites. For these sites, groundwater flow was calculated as the sum of fluxes for several flow cells of area A on a section of the riparian zone parallel to equipotential lines using piezometers from adjacent nests extending downslope in the zone of upward flow. 4. Results 4.1. Site hydrogeomorphic characteristics The key hydrogeomorphic features of the eight riparian sites are summarized in Table 1. The range of hydraulic conductivity expressed as an order of magnitude using Ks values from several piezometers is indicated for the main lithologic units in each of the riparian sites. The Eramosa site is a 220 m wide riparian zone with a concave topography connected to an approximately 750 m long upslope aquifer. The upland – riparian zone boundary has a steep slope gradient of 24% that declines rapidly to , 2% towards the river (Fig. 2a). The upland is underlain by coarse gravel and cobble sediments which are 9– 10 m thick at the field edge. At the slope base, 2.5 m of loamy-sand and coarse gravel (Ks ¼ 1024 cm s21) thins rapidly downslope where a 1 m thick humic peat deposit with Ks values of 1025 cm s21 extends towards the river (Fig. 3a). Beneath the gravel and peat deposits, a clay till forms a confining layer (Ks ¼ 1026 cm s21) throughout the riparian zone. The Boyne river site is located in a valley incised approximately 10 –12 m below the adjacent upland sand plain surface and connected to the Lake Algonquin sand aquifer. The length of the aquifer recharging the Boyne site is approximately 1400 m. The topography at this site is concave with a very steep slope at the field –riparian zone boundary of 38.6% and a flat, 180 m wide riparian zone extending to the river channel which is incised 3 m below the floodplain (Fig. 2b). The lithology at this site consists of humic peat deposits (Ks ¼ 1026 =1025 cm s21) which thicken from 0.5 m at the valley side to 3 m P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 215 Fig. 3. Vertical cross section along the main transect showing riparian lithology. Dots represent piezometer slot zones. (a) Eramosa site; (b) Boyne; (c) Road 10; (d) Speed river site; (e) Maskinonge; (f) Ganatsekiagon; (g) Highway 27; (h) Vivian (SL, sandy loam; LS, loamy sand; FS, fine sand; S, sand; CS, coarse sand; G, gravel). within the riparian area (Fig. 3b). The peats overlie fine sands and a layer of coarse sand and fine gravel with Ks values of 1023 cm s21. The regional clay underlies the sands and gravels at approximately 6 m near the valley side and 2 m below the river channel. The Road 10 site is a 30 m wide riparian zone with a mainly flat topography located on the Alliston sand plain. There is an initial drop of approximately 1 m in elevation at the field edge and then a relatively flat riparian area leading to the stream. Although this site is underlain by the 9 – 12 m thick unconfined Algonquin sand aquifer, the stream bottom is only 3 m below the sand plain surface. Subsurface flow path data (Section 4.5) suggest that groundwater deeper than 5 m below the riparian area (6 m below the sand plain surface) does not significantly interact with the stream and riparian zone. The soils at the Road 10 site are mainly sands (Ks ¼ 1023 cm s21), 216 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 Table 1 Site physical characteristics Topography type Eramosa River (fourth order) SGTa Boyne River (fourth order) SP Road 10 (first order) SP Riparian width (m) Upland permeable sediment depth (m) 220 9–10 204 30 Speed river (second order) SGT Maskinonge (second order) T Ganatsekiagon (first order) T 66 Riparian vegetation Slope gradient (%) 750 SW 15 1400 SW 6 200 H 200 H þ HW 23.6/ , 2,b overall: 5.7 38.6/ , 2, overall: 5.2 18.0/ , 1, overall: 5.2 5.0/ , 1, overall: 2.9 13.1/ , 1, overall: 5.1 Overall: 13.2 2.8–3 Upland slope length (m) 45 2 230 H þ HW 25 1.4 300 H Highway 27 (first order) T 33 1.2 400 H þ HW Vivian (second order) KM 37 0.9 250 H 20.1/1–2, overall: 11.3 Overall: 1 HW, hardwood; SW, softwood; H, herbaceous. The initials indicate the landform information. SGT, spillways and gravel terraces; SP, sand plain; T, till; KM, Kame moraine. b The first value indicates the gradient of the steepest section of the riparian zone at the upland/riparian boundary and the second indicates the slope gradient of the remaining part of the riparian zone. a however, a loamy-sand layer with lower Ks values of 1025 cm s21 extends across the riparian zone at depths 1 –3 m (Fig. 3c). The terrain at Speed River is gently sloping and the riparian zone and adjacent upland are underlain by a 3 m thick coarse gravel and cobble deposit with hydraulic conductivities of 1024 cm s21 (Fig. 3d). Beneath this layer, substrates with low Ks values of 1026 cm s21 form a confining layer for groundwater flow. The lengths of upslope areas contributing subsurface flow to Road 10 and Speed sites were approximately 200 m. The Maskinonge and Highway 27 riparian zones have a concave topography with moderate to steep slopes at the upland perimeter and level terrain near the streams. Soils at the riparian perimeter and in the adjacent upland at the Maskinonge site are loamysands that extend to a depth of 2 m and have Ks values of 1024 cm s21. At greater depths, a dense stony silty sand till (Ks ¼ 1027 cm s21) restricts subsurface flow. Within the riparian zone, peat increases in thickness from 0.4 m, near the perimeter, to a maximum of 2.7 m, 15 m from the river (Fig. 3e). Hydraulic conductivities decline from 10 25 cm s 21 in the upper 50 to 10 26 cm s 21 at greater depths in the peat. The Highway 27 site is developed on a dense sandy loam till (Ks ¼ 1027 cm s21) that forms a confining layer at a depth of 1.2 m at the field –riparian zone margin. The soil above the confining layer is a sandy loam on the slope changing to a loamy-sand (Ks ¼ 1024 cm s21) near the stream. The lengths of the contributing areas at the Maskinonge and Highway 27 sites were 230 and 400 m, respectively. The Ganatsekiagon site receives runoff from a 300 m long upland. The riparian zone is 27 m wide with a slightly convex topography and an average slope of 13.2% that extends to the stream (Fig. 3f). The soil profile is composed of a coarse sandy ablation till with thin layers of gravel (Ks ¼ 1023 cm s21) between 50 and 100 cm. A dense sandy loam basal till with Ks values of 1026 cm s21 restricts subsurface flow to depths , 1.4 m in both the riparian zone and the adjacent upland. The topography at the Vivian site is flat (slope 1%) and outwash silt with Ks values of 1026 – 1027 cm s21 forms an aquitard that varies in depth from 0.9 m at the riparian perimeter to 1.4– 1.5 m near the stream (Fig. 3h). The sediment profile above this confining layer in the riparian area is a sandy-loam (Ks ¼ 1025 cm s21) between 0 and 50 cm and a loamy-sand mixed with gravel between P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 50 and 90 cm (Ks ¼ 1024 cm s21). The length of the upland area recharging the Vivian site was estimated to be approximately 250 m. 4.2. Upland –riparian zone linkage Fig. 4 shows the duration of the upland – riparian zone subsurface hydrologic connection for the eight riparian sites. The period where the riparian zone is hydrologically connected to its upland aquifer is defined as the time interval during which the water table at the field – riparian zone boundary is above the confining layer. Five of the riparian sites had a permanent hydrologic connection to the adjacent upland throughout the study. Depths of permeable upland sediments which maintain this linkage ranged from 2 to 15 m. In contrast, permeable sediment depths overlying an aquitard were , 2 m in upland areas at the Highway 27, Ganatsekiagon and Vivian riparian sites where an intermittent hydrological linkage was found (Fig. 4). Upland slope lengths at these three riparian sites were similar to the slope lengths at Speed and Maskinonge sites that maintained a permanent hydrologic link. The duration of the upland –riparian zone connection and the dates of disconnection from the upland aquifer varied among sites and from year to year. 217 At Ganatsekiagon, the upland – riparian zone connection ceased between late July 2000 and late January 2001. In 2001, flow ceased in late July until late October. In 2002, the riparian zone remained connected to the upland aquifer until mid-June. The Highway 27 riparian zone was disconnected from the upland aquifer from mid-July 2000 to February 2001. In 2001, flow ceased between late June and late October. In 2002, the riparian zone became disconnected from its upland aquifer in early July. At Vivian creek, the upland –riparian zone linkage ceased in early August 2000, re-connected briefly in September, and ceased again until February 2001. In 2001, there was no hydrologic linkage between late June and late December and in 2002, the riparian zone remained connected to its upland aquifer until mid-July. 4.3. Water inputs Large, relatively stable water inputs were observed at the Eramosa and Boyne sites which were connected to large upland aquifers (Table 2). In contrast, at the Speed and Maskinonge sites, where upland permeable sediment depths were 2– 3 m, fluxes were lower and more seasonally variable. Nevertheless, small inputs during periods of maximum water table drawdown Fig. 4. Duration of the upland– riparian zone hydrological connection versus upland permeable sediment depth for the riparian zones. 218 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 Table 2 Subsurface water inputs (L day21 m21 width) from adjacent upland to the riparian zones during high, medium and low water table periods Eramosa Boyne Road 10 Speed Maskinonge Ganatsekiagon Highway 27 Vivian High water table Medium water table Low water table 394 320 44 66 72 244 30 0.7 355 300 0 22 26 1.8 9 0.4 300 280 0 14 4 0 0 0 maintained a continuous upland riparian zone linkage. Significant hillslope discharge at the Road 10 site only occurred during high water table conditions in the spring. The change in elevation between the field and the riparian zone was only 1 m at this site and water inputs from the field ceased as the hydraulic gradient decreased to almost zero. Fluxes were very variable at Ganatsekiagon, Highway 27 and Vivian sites, where hillslope discharge ceased during summer and early autumn. Groundwater inputs to the riparian zones during periods of intermediate water table elevation had a significant positive correlation with upland permeable sediment depth ðr 2 ¼ 0:77; p , 0:05Þ and upland slope length ðr 2 ¼ 0:72; p , 0:05Þ: However, these relationships were not significant for high water table periods because of the large subsurface input at the Ganatsekiagon site where the upland aquifer was small. When this site was excluded from the analysis, we found a significant positive correlation between water input during high water table conditions and both upland permeable sediment depth ðr 2 ¼ 0:75; p , 0:05Þ and upland slope length ðr 2 ¼ 0:65; p , 0:05Þ: 4.4. Water table fluctuations Seasonal variations in riparian water table fluctuations at the eight sites are shown in Fig. 5. These water table elevations were averaged for wells located between the perimeter and the mid-point of each riparian area where the water table was influenced Fig. 5. Mean water table depth below ground surface (BGS) for (a) sites with a continuous upland–riparian hydrologic connection and (b) sites with a discontinuous upland–riparian hydrologic connection. mainly by upland subsurface inputs rather than inflow from the stream. The water table remained close to the ground surface at the Eramosa and Boyne sites. The minimum mean water table height at the Eramosa site was only 13 cm. At the Boyne site, the minimum elevations of 47 and 50 cm occurred for a few weeks in the dry summers of 2001 and 2002, respectively. Annual water table fluctuations at the Road 10, Speed and Maskinonge sites were larger than at the Boyne and Eramosa sites. Although the water table was often near the surface at these sites in spring, it dropped to 112, 134 and 126 cm in the summers of 2000– 2002 at the Road 10 site. Minimum mean water table heights at the Speed and Maskinonge sites in the dry summer of 2001 were 137 and 62 cm, respectively (Fig. 5). The largest water table drawdowns were recorded for the Ganatsekiagon, Highway 27 and Vivian sites, which were hydrologically disconnected from the adjacent upland in summer and autumn. The water table declined below the confining layer at these sites in 2001 and 2002 to minimum elevations of . 150 cm (Ganatsekiagon), . 200 cm (Highway 27) and . 250 cm (Vivian). P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 4.5. Groundwater flow paths in the vertical plane Groundwater flow patterns in a vertical cross-section along the main piezometer transect for each site are shown for two dates representative of periods of high and low water table, respectively (Fig. 6). Except for 219 the Eramosa site where there was only one piezometer transect, the second transect, which had fewer piezometer nests, showed groundwater flow patterns which were similar to the main transect at each site. At the Eramosa site, equipotential lines indicate up-welling at the slope bottom between nests 105 Fig. 6. Hydraulic head contours (cm) for a vertical cross section along the main transect of each riparian zone for one date in spring 2001 (high water table) and one date in summer 2001 (low water table). (a) Eramosa; (b) Boyne; (c) Road 10; (d) Speed; (e) Maskinonge; (f) Ganatsekiagon; (g) Highway 27; (h) Vivian. Dashed line represents location of the water table. The small flow arrows only represent the general direction of flow because the vertical exaggeration affects the angle with which the flow system crosses the equipotential lines. 220 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 Fig. 6 (continued ) and 103. Groundwater discharged to the surface in seeps creating surface rivulets, which flowed across the riparian zone from nests 105 to 102A in summer and beyond nest 100 in the spring (Fig. 6a). Equipotential lines also indicate that near surface water recharged to deeper sediments, downslope from nest 102A, especially during the driest months of the year. The flow net indicates predominantly horizontal flow from the riparian perimeter to the river at the Boyne site (Fig. 6b). Springs were present at the surface of the peat between nests 3 and 4 although the hydraulic heads did not indicate strong vertical gradients within the peat. Small surface rivulets produced by the springs extended to nests 6– 7 before recharging to the peat from March to May. There was enough hillslope discharge to maintain a hydraulic gradient from field to stream in the spring at the Road 10 site (Fig. 6c). However, during summer, P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 hillslope discharge decreased and the hydraulic gradient from field to stream declined to almost zero by late summer. Equipotential lines indicate up-welling of groundwater from the Alliston sand aquifer beneath the riparian zone in spring. This upward flow persisted near the stream during the summer. At depths of . 5 m, groundwater flow was predominantly horizontal beneath the riparian area (Fig. 6c). At the Speed site, hillslope subsurface inputs maintained a field to riparian zone hydraulic gradient in the upper part of the riparian zone (nests 5 –3) throughout the year (Fig. 6d). Nevertheless, as the riparian zone dried out during the summer, there was a decrease of the hydraulic gradient between nests 5 and 3 and a reversal of the water table gradient was observed in the lower part of the riparian zone (nests 1– 2A). The pattern of groundwater flow observed at the Maskinonge site was similar to the Eramosa site (Fig. 6e). Hydraulic head contours showed upward gradients of water toward the surface adjacent to the upland – riparian perimeter between nests 12 and 11B. In spring, groundwater seeps produced surface rivulets that recharged into the riparian zone soil between nests 11A and 11. Although this riparian zone is permanently connected to the upland aquifer, hillslope discharge significantly decreased during the summer. In the lower part of the riparian zone (nests 11– 10), the hydraulic gradient was almost nil during that period, indicating that groundwater discharge to the stream was very limited during summer months (Fig. 6e). During periods of upland – riparian hydrologic linkages, subsurface flow paths were parallel to the slope surface and the shallow underlying confining layer at the Highway 27, Ganatsekiagon and Vivian sites (Fig. 6f – h). Hillslope discharge was very intermittent at the Ganatsekiagon site and the water table declined rapidly in early summer each year (Fig. 6f). Flow reversal situations were never observed at this site during the study period. When the Highway 27 riparian zone was disconnected from the upland in the summer and autumn, the upper part of the riparian zone dried out and the hydraulic gradient near the stream was almost zero (Fig. 6g). Hillslope discharge was also very intermittent at the Vivian site. During summer, the water table first 221 declined parallel to the ground surface and then, as hillslope discharge ceased, the stream became the main source of water to the riparian zone and a steep hydraulic gradient was established towards the slope (Fig. 6h). 4.6. Subsurface flow patterns in the horizontal plane The configuration of well and piezometer nests provides a three-dimensional perspective of groundwater flow for all sites except Eramosa, where transects did not extend across the entire riparian area and where the three-dimensional flow pattern was not established. Water table elevations measured at monthly intervals revealed three different patterns of groundwater flow. The first pattern was found at Vivian Creek and Road 10 sites where large seasonal changes in subsurface flow direction occurred across the entire riparian zone. At Vivian Creek, groundwater flow was parallel to the main topographical gradient (Fig. 2h) during high water table conditions in spring resulting in a slightly oblique flow direction across the riparian area (Fig. 7b). During summer and early autumn, when there was no runoff from the hillslope, the riparian water table was sustained by stream inflow. During this period, we observed a large shift in flow direction across the riparian zone with water flowing in a stream-to-riparian zone direction with a strong downstream component (Fig. 7d). By December, when water table levels were similar to those observed in the spring, the flow pattern returned to a hillslope to stream orientation. Similar changes in flow direction between spring, summer and autumn were observed at the Vivian site each year. However, changes in flow direction were less severe in 2000, a year with high rainfall in May and June. The groundwater flow pattern in October 2000 (Fig. 7a), during the period of maximum annual water table drawdown, was similar to June 2001 and did not show the large reversal of hydraulic gradient observed in September 2001 (Fig. 7d). At Road 10 site (data not shown), the direction of groundwater flow was from field-to-stream in the spring. However, during the summer, the field to riparian zone water table gradient decreased to near zero producing a change in flow pattern to a down valley orientation parallel to the stream. In contrast to 222 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 Fig. 7. Water table contour maps for Vivian Creek riparian site. (a) October 2000, (b) April, (c) June, (d) September, (e) November and (f) December 2001. Water table contours are 20 cm intervals. Arrow sizes are not proportional to magnitude of flow. The dashed line indicates the field edge. Vivian Creek, a flow reversal from the stream to the slope was not observed at this site. A second type of flow pattern was observed at the Speed River site where hillslope discharge was large enough to maintain a field-to-riparian zone gradient throughout the year in the upper half of the riparian zone (Fig. 8). However, in the lower part of the riparian zone significant changes in flow direction were observed. As the water table declined, the downstream component of subsurface flow became predominant. In September 2001, a flow reversal was observed from the stream towards the centre of the riparian zone (Fig. 8d). As the water table levels increased in November and December, the groundwater flow direction returned to the spring pattern (Fig. 8e and f). As previously noted for the Vivian site, the strong reversal of hydraulic gradient near the stream did not occur during the summer and autumn of 2000 at the Speed site (Fig. 8a). A third riparian groundwater flow pattern occurred at the Boyne River, Maskinonge, Highway 27 and Ganatsekiagon sites. Groundwater flowed in the direction of the steepest slope and was perpendicular Fig. 8. Water table contour maps for Speed River riparian site. (a) November 2000, (b) April, (c) June, (d) September, (e) November and (f) December 2001. Water table contours are 20 cm intervals. Arrow sizes are not proportional to magnitude of flow. The dashed line indicates the field edge. P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 Fig. 9. Water table contour maps for Maskinonge Creek riparian site. (a) April, (b) June and (c) September. Water table contours are 25 cm intervals. Arrow sizes are not proportional to magnitude of flow. The dashed line indicates the field edge. to the stream at these sites. This pattern was constant throughout the year at the Boyne and Maskinonge sites and occurred during the period of upland hydrologic connection at Highway 27 and Ganatsekiagon. The Maskinonge site is representative of this pattern (Fig. 9). As hillslope discharge decreased during the summer, the water table declined parallel to the ground surface but the water table hydraulic gradient remained relatively constant from the hillslope toward the stream. 5. Discussion These data indicate that small differences in the depth of permeable upland sediments capable of serving as aquifers control the duration of the hydrological connection between the riparian zone and the adjacent upland. All riparian sites where the depth of permeable sediments overlying a confining layer was approximately 2 m or greater were permanently connected to their upland aquifer, whereas the upland – riparian zone linkage was intermittent at all sites with , 2 m of permeable upland sediments (Fig. 4). In contrast, the length of upland slope contributing subsurface flow to riparian zones with an intermittent link was similar to several sites that had a permanent hydrologic connection. The duration of the upland – riparian zone linkage for the sites connected to a thin aquifer is very strongly affected by periods of low precipitation input. In November and December 2000, none of the sites in this category were reconnected with their upland aquifer, whereas in 223 2001, hillslope discharge began by late October – November. After a dry early autumn, precipitation in late November and December 2000 occurred mainly as snow, so there was almost no recharge during this period. In contrast, although the summer of 2001 was very dry, precipitation in September – November was 29% higher than the 30 year normal. Devito et al. (1996) also showed that the upland – wetland linkage in southern Ontario Canadian Shield catchments with less than 2– 3 m of till overlying bedrock ceased during the summer in response to variations in precipitation. Our results and those of Devito et al. (1996) suggest that approximately 2 m of permeable sediments above a confining layer is a minimum depth required for a permanent upland – riparian zone linkage in the southern Ontario climatic regime. The size of the upland aquifer influences the magnitude and seasonality of fluxes entering the riparian zone. However, the absolute values of these groundwater inputs should be treated with caution because of inherent measurement errors. The largest and most constant water inputs were observed at the Eramosa and Boyne sites where the upland slope lengths were 750 and 1400 m, respectively, and permeable sediment depths were 10 – 15 m. The riparian zones connected to smaller upland aquifers had lower and more seasonally variable subsurface fluxes. Nevertheless, our correlation analysis for high water table periods only indicated a significant positive relationship between riparian water inputs and both upland permeable sediment depth and slope length after exclusion of the Ganatsekiagon site. The occurrence of large water inputs at this site suggest that topography and soil permeability are also important variables controlling groundwater inputs to some riparian areas. At the Ganatsekiagon site, despite a very thin upland aquifer (, 1.4 m), the combination of steep topography and highly permeable sediments produced fluxes that were similar in magnitude to much thicker and more extensive aquifers. However, because of the small upland aquifer, these fluxes were limited to periods of high water table during a few weeks in spring and autumn whereas at a site like Boyne River, large fluxes occurred throughout the year. The results of this study also show that the relationship between upland aquifer depth and the maintenance of a hydrologic connection has 224 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 an important influence on riparian zone water table fluctuations. Sites connected to a thick extensive upland aquifer (Eramosa, Boyne) had small annual water table fluctuations whereas sites that become disconnected from uplands (Highway 27, Ganatsekiagon, Vivian) had annual water table drawdowns of . 1 – 2 m which can result in the complete drying out of the riparian area unless the water table is sustained by the stream. Similar variability in the amplitude and seasonal duration of water table elevation in relation to upland hydrologic linkages has been reported for small forested headwater wetlands (Devito et al., 1996; Hill and Devito, 1996). Sediment stratigraphy also influences water table dynamics and subsurface flow paths in some riparian sites. At the Eramosa site, the rapid decrease in thickness of the gravel and cobble layer from 12 m at the field edge to , 2 m overlying a low conductivity clay till at the slope base produced groundwater discharge to the riparian surface. This decrease in aquifer thickness and continuous hillslope discharge maintained a relatively stable water table position within 15 cm of the surface even during summer droughts. Permanent hillslope discharge at the Boyne site associated with a decrease in thickness of the sand aquifer beneath the riparian area and the presence of low conductivity peats, was also linked to limited water table fluctuations and the occurrence of surface saturation within 70 –80 m of the riparian perimeter for most of the year. Hinton et al. (1993) have reported similar effects of decreasing sediment thickness along flow paths on groundwater discharge and surface saturation in a Canadian Shield glacial till catchment. At the Maskinonge site, upward flow due to the damming effect produced by low conductivity peats maintained surface saturation for considerable periods of the year. As a result, water table variations were lower here than at sites such as Speed River which have a larger upslope depth of permeable sediments, but where riparian subsurface flow is not restricted by low conductivity sediments. Brusch and Nilsson (1993) also observed that groundwater flow from a sand aquifer in Denmark was forced to the surface at the perimeter of a riparian zone by a low permeability organic deposit. Surface groundwater seeps produced surface rivulets at the Eramosa, Boyne and Maskinonge sites. This flowpath reduces residence times and contact with riparian sediments which can limit nitrate removal (Warwick and Hill, 1988; Gold et al., 2001). Landscape topography has a considerable influence on riparian subsurface flow pattern. Our data suggest that subsurface water tends to flow downslope in areas of the riparian zone where the slope gradient is steeper than 5%. The Road 10 site, which has a slope gradient of 18% at the upland perimeter, is an exception to this pattern. However, the drop in elevation at the perimeter is only 1 m and this is not enough to influence the water table gradient. The sites with a steep concave topography (Boyne, Eramosa) that we studied are connected to thick extensive upland aquifers. The combination of large, continuous hillslope discharge with a steep topography (. 15%) at the field edge, maintained a field to stream water table gradient throughout the year even on the flat areas of the riparian zone. The Speed and Vivian riparian zones occur in landscapes with level to gentle slopes (, 5%) where a low hydraulic gradient between uplands, riparian areas and the stream increase the probability that the water table in the riparian zone may fall below the stream water level resulting in flow reversals (Figs. 7 and 8). Burt et al. (2002) also indicate that when the topography is flat, stream water level can strongly influence water table dynamics in the riparian zone. 6. Conclusions and conceptual model The eight riparian sites examined in this study represent a large portion of the range of hydrologic behaviour typical of glacial till and outwash landscapes in southern Ontario where local groundwater flow systems connect uplands to adjacent riparian areas. A conceptual model of stream riparian hydrology in these landscapes is presented in Fig. 10. This model emphasizes the importance of upland permeable sediment depth as an indicator of aquifer size and topography of the riparian zone and adjacent upland for the identification of different hydrologic categories of riparian zones. The first hydrologic type occurs on areas of level to sloping topography underlain by clay soils behaving as aquitards that extend almost to the surface. P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 225 Fig. 10. Conceptual framework identifying riparian hydrologic types. Although this landscape setting was not included in the present study, we suggest that subsurface flux will be very small because of low soil permeability and that runoff will occur mainly as overland flow during snowmelt and rain events allowing contaminated water to bypass the riparian soil. A second category of riparian zone hydrologic functioning occurs in landscapes with a thin upland aquifer (1 – 2 m) and level to gently sloping topography (, 5%). The upland –riparian zone linkage is generally very intermittent and hillslope discharge is seasonally variable and precipitation dependent. Water table drawdowns between 1 and 1.5 m are common in normal years but can exceed 2 m in especially dry years. Consequently, these sites have a limited role in regulating subsurface contaminant flows between uplands and streams. Large changes in subsurface flow direction are likely to be observed in this riparian hydrologic type, with water flowing in a field-to-stream direction in the spring and then in the direction of the valley gradient or even in a stream-tofield direction during the summer. A third type is represented by riparian zones linked to thin upland aquifers on steeper topography (. 5%). These sites have water table dynamics and seasonality of hillslope discharge that is similar to the second riparian type. However, during periods of high water table, sites with steeper slopes in combination with highly conductive surficial sediments may produce large subsurface fluxes of water and contaminants that can exceed the buffering capacity of the riparian zone. Because of the steeper topography, subsurface water flows in the direction of the steepest slope gradient whenever these sites are hydrologically active. It is important to note that a slope extending to the stream decreases the buffer role of the riparian zone. 226 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 This could apply to several of the riparian types presented here when moderate to steep slopes are connected directly to the stream. Another hydrologic type is represented by riparian zones linked to upland aquifers with intermediate thickness (2 – 6 m) and level to gently sloping topography. These sites have considerable seasonal variations in groundwater input but maintain a continuous upland – riparian zone hydrological connection. Water table drawdowns between 0.3 and 1.2 m in normal years and up to 1.5 m in especially dry years are likely to be observed. These sites have seasonal changes in subsurface flow direction with the magnitude of the changes being directly related to the balance between water level at the field edge and stream water level dynamics. Subsurface flows may shift from a field –riparian to a down valley gradient in most summers, however, flow reversal situations generally occur only in exceptionally dry years. In contrast, subsurface flow paths in riparian zones linked to intermediate aquifers on steeper topography (. 5%) are generally in the direction of the steepest slope except very close to the stream. Riparian zones linked to thick aquifers (. 6 m) may be located on level to gently sloping topography, where the stream channel is incised only a few meters below the landscape surface. As a result of the low hydraulic gradient, groundwater interaction with the riparian zone is restricted to the upper few meters of the aquifer. Water table fluctuations and changes in subsurface flow direction in this category of riparian zone are similar to sites linked to intermediate aquifers (2 – 6 m) on level to gently sloping topography. Our conceptual model identifies a final riparian hydrologic type which is linked to a thick upland aquifer and has a steep concave topography (locally . 15%) as a result of valley incision. These riparian zones have large seasonally constant groundwater inputs, which maintain stable water tables. The initial steep slope associated with a large and permanent hillslope discharge generally maintains a stable field to stream subsurface flow path across the entire riparian zone. In some of these riparian zones, the water table intersects the ground surface at the slope bottom producing seeps and surface rivulets. In sites with conductive sediments at depth, groundwater may bypass the riparian area and discharge upward to the stream channel. These flow paths may limit the effectiveness of this riparian type as a water quality buffer. It may be possible to use widely available maps of topography, soils and surficial geology to identify these different riparian hydrologic types at the landscape scale. Variations in riparian stratigraphy and sediment permeability can also locally modify water residence times and subsurface flow paths. These site-specific characteristics may be difficult to evaluate without individual site studies. This conceptual model of riparian zone hydrological functioning complements and extends previous research on landscape setting and riparian hydrology (Devito et al., 1996; Hill, 2000; Baker et al., 2001; Burt et al., 2002). Our study is the first to use data from a wide range of riparian sites to examine in detail the interacting effect of upland aquifer size, topography and lithology on riparian hydrology. Our classification of riparian hydrologic types can probably be used with minimum modifications in most glacial till and outwash landscapes. However, this conceptual framework may require modification in other landscape and climatic settings. For example, the critical depth of permeable sediments required to maintain a continuous upland –riparian hydrologic link may vary depending on seasonal patterns of precipitation and evapotranspiration in different climatic regimes. It is important that future research emphasises the development of process-based conceptual frameworks that link differences in riparian hydrology to key landscape hydrogeologic characteristics. These conceptual models provide an essential template for understanding and predicting similarities and differences in the water quality function of riparian zones at the catchment and landscape scales. Acknowledgements We thank Alan Michalsky for preparing the piezometers and wells, Graham Carlyle, Robert McDonald and Tim Duval for assistance in the field, and Shan Sanmugadas and Jackson Langat for laboratory assistance. Thanks are also due to landowners for access to the riparian sites. We thank Kevin Devito and Tim Burt for comments on P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 the manuscript. The research was supported by grants from the Natural Sciences and Engineering Research Council of Canada to A.R. Hill. References Baker, M.E., Wiley, M.J., Seelbach, P.W., 2001. GIS-based hydrologic modeling of riparian areas: implications for stream water quality. J. Am. Water Resour. Assoc. 37, 1615–1628. Barwell, V.K., Lee, D.R., 1981. Determination of horizontal-tovertical hydraulic conductivity ratios from seepage measurements on lake beds. Water Resour. Res. 17, 565–570. Bohlke, J.K., Denver, J.M., 1995. Combined use of groundwater dating, chemical, and isotopic analyses to resolve the history and fate of nitrate contamination in two agricultural watersheds, Atlantic coastal plain, Maryland. Water Resour. Res. 31, 2319–2339. Brinson, M.M., 1993. Changes in the functioning of wetlands along environmental gradients. Wetlands 13, 65– 74. Brusch, W., Nilsson, B., 1993. Nitrate transformation and water movement in a wetland area. Hydrobiologia 251, 103–111. Burt, T.P., 1997. The hydrological role of buffer zones within the drainage basin systemi. In: Haycock, N.E., Burt, T.P., Goulding, K.W.T., Pinay, G. (Eds.), Buffer Zones: Their Processes and Potential in Water Protection, Quest Environ., pp. 21– 32. Burt, T.P., Matchett, L.S., Goulding, K.W.T., Webster, C.P., Haycock, N.E., 1999. Denitrification in riparian buffer zones: the role of floodplain hydrology. Hydrol. Proc. 13, 1451–1463. Burt, T.P., Pinay, G., Matheson, F.E., Haycock, N.E., Butturini, A., Clement, J.C., Danielescu, S., Dowrick, D.J., Hefting, M.M., Hillbricht-Ilkowska, A., Maı̂tre, V., 2002. Water table fluctuations in the riparian zone: comparative results from a pan-European experiment. J. Hydrol. 265, 129–148. Chason, D.B., Siegel, D.I., 1986. Hydraulic conductivity and related physical properties of peat, Lost River Peatland, northern Minnesota. Soil Sci. 142, 91–99. Correll, D.L., Jordan, T.E., Weller, D.E., 1997. Failure of agricultural riparian buffers to protect surface waters from groundwater nitrate contamination. In: Gibert, J., Mathieu, J., Fournier, F. (Eds.), Groundwater/Surface Water Ecotones: Biological and Hydrological Interactions and Management Options, Cambridge University Press, Cambridge, pp. 162 –165. Deane, R.E., 1950. Pleistocene Geology of the Lake Simcoe district, Ontario. Geological Survey of Canada, Memoir, 256. Devito, K.J., Hill, A.R., 1997. Sulphate dynamics in relation to groundwater – surface water interactions in headwaters wetlands of the Southern Canadian shield. Hydrol. Proc. 11, 485–500. Devito, K.J., Hill, A.R., Roulet, N., 1996. Groundwater–surface water interactions in headwater forested wetlands of the Canadian Schield. J. Hydrol. 181, 127–147. 227 Devito, K.J., Creed, I.F., Rothwell, R.L., Prepas, E.E., 2000a. Landscape controls on phosphorus loading to boreal lakes: implications for the potential impacts of forest harvesting. Can. J. Fish. Aquat. Sci. 57, 1977–1984. Devito, K.J., Fitzgerald, D., Hill, A.R., Aravena, R., 2000b. Nitrate dynamics in relation to lithology and hydrologic flow path in a river riparian zone. J. Environ. Qual. 29, 1075–1084. Freeze, R.A., Cherry, J.A., 1979. Groundwater. Prentice-Hall, Englewood Cliffs, NJ. Gilliam, J.W., 1994. Riparian wetlands and water quality. J. Environ. Qual. 23, 896 –900. Gold, A.J., Groffman, P.M., Addy, K., Kellogg, D.Q., Stolt, M., Rosenblatt, A.E., 2001. Landscape attributes as controls on groundwater nitrate removal capacity of riparian zones. J. Am. Water Resour. Assoc. 37, 1457–1464. Golden Software, 1999. Surfer 7. Golden Software, Inc, Golden, Colorado 80401-1866, USA. Haycock, N.E., Pinay, G., 1993. Groundwater nitrate dynamics in grass and poplar vegetated riparian buffer strips during the winter. J. Environ. Qual. 22, 273–278. Hill, A.R., 1990. Ground water flow paths in relation to nitrogen chemistry in the near-stream zone. Hydrobiologia 206, 39–52. Hill, A.R., 1996. Nitrate removal in stream riparian zones. J. Environ. Qual. 25, 743 –755. Hill, A.R., 2000. Stream chemistry and riparian zones. In: Jones, J., Mulholland, P. (Eds.), Streams and Ground Waters, Academic Press, London, pp. 83–110. Hill, A.R., Devito, K.J., 1996. Hydrological–chemical interactions in headwater forest wetlands. In: Trettin, C.C., Jurgensen, F., Grigal, D.F., Gale, M.R., Jeglum, J. (Eds.), Northern Forested Wetlands: Ecology and Management, CRC/Lewis, Boca Raton, FL, pp. 217–233. Hinton, M.J., Schiff, S.L., English, M.C., 1993. Physical properties governing groundwater flow in a glacial till catchment. J. Hydrol. 142, 153–175. Jordan, T.E., Correll, D.L., Weller, D.E., 1993. Nutrient interception by a riparian forest receiving inputs from adjacent cropland. J. Environ. Qual. 22, 467 –473. Lowrance, R., 1992. Groundwater nitrate and denitrification in a coastal riparian forest. J. Environ. Qual. 21, 401–405. Lowrance, R., Newbold, J.D., Schnabel, R.R., Groffman, P.M., Denver, J.M., Correll, D.L., Gilliam, J.W., Robinson, J.L., Brinsfield, R.B., Staver, K.S., Lucas, W., Todd, A.H., 1997. Water quality functions of riparian forest buffers in Chesapeake Bay watersheds. Environ. Manag. 21, 687–712. Peterjohn, W.T., Correll, D.L., 1984. Nutrient dynamics in an agricultural watershed: observations on the role of a riparian forest. Ecology 65, 1466–1475. Roulet, N.T., 1990. Hydrology of a headwater basin wetland: groundwater discharge and wetland maintenance. Hydrol. Proc. 4, 387–400. Schnabel, R.R., Gburek, W.J., Stout, W.L., 1994. Evaluating riparian zone control on nitrogen entry into northeast streams, Ecosystems in the Humid US, Functions, Values and Management, National Association of Conservation Districts, Washington DC, pp. 432–445. 228 P.G.F. Vidon, A.R. Hill / Journal of Hydrology 292 (2004) 210–228 Sibul, V., Choo-Ying, A.V., 1971. Water Resources of the Upper Nottawasaga River Drainage Basin. Water Resources Report No. 3, Ontario Water Resources Commission, Toronto, Ontario, 128 p. Singer, S.N., Cheng, C.K., Scathe, M.E., 1997. The hydrogeology of southern Ontario, Hydrogeology of Ontario Series, Report 1. Ministry of Environment and Energy, Toronto. Warwick, J., Hill, A.R., 1988. Nitrate depletion in the riparian zone of a small woodland stream. Hydrobiologia 157, 231–240. Winter, T.C., 1992. A physiographic and climatic framework for hydrologic studies of wetlands. In: Robarts, R.D., Bothwell, M.L. (Eds.), Aquatic ecosystems in semi-arid regions: implications for resource management, NHRI Symp. Ser. 7, pp. 127–148.