Seven hydrogeological terrains characteristic of southern Ontario

In addition to surface water, groundwater is an important component of the hydrological cycle of southern Ontario. It is significant to water supply as it provides high-quality water across most of the region, particularly in rural areas. In addition, groundwater discharge (GWD) to surface water maintains streamflow between precipitation events, particularly during summer months. It also moderates in-stream temperatures, stream water quality, and habitat for some aquatic species.

Understanding of the hydrologic cycle can advance by tracing events from climate inputs to the land surface, to overland flow, to infiltration, and recharge to groundwater storage and flow to streams. Progress in understanding improves by developing a hydrogeological framework of typical terrains to identify geological and topographical controls on water movement through the southern Ontario landscape (e.g., Piggott and Sharpe 2007; Sharpe et al. 2014). Given a measured precipitation flux, bedrock, sediment cover and thickness, as well as permeability and porosity, combined with topography, are important factors that control the movement of water as it arrives on the land surface and then enters subsurface pathways. Knowledge of terrain (geology, topography) and climate flux, allows assessment of surface water–groundwater interaction (e.g., Hinton 2014; Conant et al. 2019). Our understanding of the hydrological cycle in southern Ontario progresses with water monitoring data integrated with terrain information. This allows for a more complete description of the hydrogeological response of water-bearing formations. Groundwater levels within water bearing formations respond to the elevation within the landscape and to rock or sediment properties. Surface topography and buried structures (e.g., channels) provide/alter gradients that drive water flow, whereas internal rock or sediment properties (texture, compactness) control the storage of water and the rate of flow. Combined, these characteristics are critical to land use, land-use planning and land-use impacts in southern Ontario (e.g., O’Connor 2002) and the wise use of water resources as climates change (Piggott et al. 2005; Grannemann and Van Stempvoort 2016). Improved knowledge of regional hydrogeological settings prepares Ontario for more effective land and water resource management. It may also provide a conceptual perspective for understanding abstract numeric modelling results for nonspecialists. Groundwater quality changes as it passes through different rock or sediment types (Hamilton 2015; Sharpe et al. 2020); however, that is not a focus in this paper despite its important land-use and health implications.

The purpose of this paper is to identify, characterize, and assess a variety of hydrogeological settings, chosen to support a fundamental characterization of regional surface water and groundwater conditions across southern Ontario. The analysis uses a modest set of available water monitoring and terrain data. A conceptual understanding supported with graphical monitoring data attempts to be instructive in the hydrogeological function of each of seven settings. The study focuses on integrating a limited set of terrain (surface geology and elevation) and hydrometeorological data (climate, streamflow, groundwater level), that are readily accessible to a range of practitioners, to describe and assess regional landscape hydrology and related subsurface flow. Three case studies presented previously (Sharpe et al. 2014) illustrate the value of connecting monitoring hydrographs to typical hydrogeological settings. A review of some limitations of data selection in this study helps refine the findings.

The study area is of hydrogeological interest as part of the Great Lakes basin (Grannemann et al. 2000; Conant et al. 2016). A large Paleozoic sedimentary basin that is part of the Appalachian and Michigan basins, between the Ottawa valley and Georgian Bay, underlie Southern Ontario (Carter et al. 2020). These sedimentary rocks unconformably overlie the Canadian Shield. The area has low-relief bedrock topography interrupted by buried valleys and prominent escarpments, such as the Niagara Escarpment, a 100 m high bluff that extends from Niagara Falls to Bruce Peninsula and beyond (Sharpe et al. 2014). It is an erosional upland with fractures and cavities in soluble carbonate cap rocks. Glaciers and meltwater have shaped the landscape leaving behind moraines, till, clay, and sand plains (Fig. 1). Large sandy moraines (e.g., Oak Ridges Moraine (ORM)) have 50–75 m of relief and form important drainage divides to major lakes. Smaller moraines, drumlins and eskers provide 10–25 m of local relief, and meltwater channels manifesting as surface valleys yield relief of 25–50 m. Low-relief clay and sand plains, incised by modern rivers, have 10–25 m of relief. Gently dipping Paleozoic carbonate and shale beds influence regional groundwater flow and storage beneath thin to thick (10–200 m), carbonate-rich, glacial sediment. Sediment is thinnest near escarpments and along river valleys and thickest along buried bedrock valleys and in major moraines (e.g., Logan et al. 2020). The distribution, thickness, and physical properties of surface geological units, bedrock, sand, clay, till, and gravel, as well as slope and vegetation, affect infiltration to groundwater. These varied and complex glacial landscapes result in equally complex local surface- and groundwater-flow patterns.

Regional groundwater systems across southern Ontario (Sharpe et al. 2014; Carter et al. 2020) have been identified based on geology, the nature of groundwater flow (i.e., fractured, karst or porous media flow), and on groundwater storage. Major hydrogeological domains include: (i) thick sediment cover (∼50–200 m), (ii) exposed/thin sediment-covered bedrock, (iii) contact zones between sediment and bedrock, and (iv) deep bedrock (>140 m depth of fresh water). Within these major domains, water moves through fractures (or solution-enlarged joints) in solid rocks and through pore spaces in unconsolidated sediment.

Glacially derived sediment covers bedrock (Barnett et al. 1991), providing the regional landscape with gentle topography and fertile, well-drained soils. Despite proximity to Great Lakes surface water, groundwater is a vital resource for agricultural and provides 90% of inland (away from the Great Lakes) potable-water use, including significant discharge to stream flow. The region has a humid continental climate (Brown et al. 1968): warm to hot summers, cold winters, and consistent precipitation from one year to the next. The Great Lakes modify climate in the region, particularly temperatures inland from the lakes and produce lake-effect precipitation in areas east of Lake Huron and Georgian Bay. Mean annual temperature also varies with latitude and elevation from roughly 4 to 10 °C. Mean annual precipitation varies across the region from 790 mm near Toronto to 1200 mm east of Lake Huron (overall average 960 mm). The warm growing season and abundant rainfall support extensive agriculture, woodlands, and wetlands.

Groundwater systems operate across drainage basins at various nested scales (Meyboom 1962; Tóth 1963). These can be described loosely as hydrogeologic landscapes in which water flow to streams is determined by topography, geology, and climate (Winter 2001; Dahl et al. 2007). Similar landscape classifications in Europe (defined by hills, hillslope, and lowlands), correspond to dominant hydrological regimes: overland flow, storage-controlled subsurface flow, and deep percolation (McMillan et al. 2000). Further, the regional concept of hydrodomains supports the assessment of regional hydrogeological pathways that describe conceptual understandings of shallow geological systems (Price et al. 2007).

Hydrologic landscapes become hydrogeologic terrains when the surficial geology of the landscape is explicitly integrated into the assessment. Traditionally, focus has been on hydrological study of catchments where hydrological boundaries can be easily defined (e.g., Devito et al. 2005; Buttle 2006). Small, homogeneous catchments were often selected to “control” for climate and geology, with an emphasis on predicting hydrologic response mainly on the basis of topography. This approach led to an overreliance on digital elevation models (DEM) (Gharari et al. 2011), in which surface topography combined with channel networks persists as the principal hydrological boundary. This reflects the fact that a watershed boundary is defined traditionally using surface topography. Recent efforts in watershed assessment, however, illustrate the need to integrate surface water with groundwater processes (Winter 2001), and to highlight the role of surface geology (and stratigraphy) on soil moisture distribution and runoff response (Buttle et al. 2004). In short, streamflow (i.e., the flow of water through wetlands, lakes, and rivers) consists of a rapidly varying runoff component due to overland flow directly to surface water, and, a slowly varying base flow component due to water storage and release processes, largely contributed by GWD (Neff et al. 2005).

The broadest hydrogeological landscapes (sediment and bedrock) reflect the geographic distribution of prominent aquifers in bedrock and thick sediment (Hinton et al. 2007; Fig. 2). These were mapped using bedrock elevation and sediment thickness as derived from water well records, which show most water supply and monitoring wells as sourced in bedrock or in thick sediment.

A specific categorization of southern Ontario groundwater regimes is presented by means of a set of typical geological terrains. Water monitoring data linked to terrain (geology, slope) allow the timing and nature of groundwater flow (i.e., fractured, karst, or porous media flow) and groundwater storage to be inferred. Of the major hydrogeological domains, thick sediment (∼50–200 m), and exposed/thin sediment-covered-bedrock terrains are highlighted. Contact zone sediment/bedrock and deep bedrock flow are mentioned in passing. Characteristics under consideration include high conductivity flow/relatively low storage capacity groundwater regimes, such as fractured or karst bedrock; lower conductivity/higher groundwater storage capacity groundwater regimes, such as silt/clay sediment.

The focus is on characteristics of smaller scale (<1300 km²) hydrogeological settings so that factors such as variations in climate and human contributions to base flow are less likely (Piggott et al. 2006). At the lower end of this watershed range, groundwater flow across boundaries is more probable and therefore values of base flow index may reflect the transfer of groundwater to and from adjacent watersheds with differing geology. Seven basic terrains provide a characterization of the land surface across southern Ontario (Fig. 1).

This paper develops and applies conceptual models as typical hydrogeological terrains, which as a group, are representative of most of southern Ontario. Surficial sediment texture and stratigraphy, bedrock types, physiography and topographic gradients yield distinctive local characteristics that are representative of larger areas. In each setting, the movement of water exhibits differences in fluxes and temporal patterns as it flows from the surface to the subsurface and back. Of the seven hydrogeological terrains identified (locations shown in Figs. 1 and 2), five are in sediment and two in bedrock (Table 1). These characterize most of the terrain spanning southern Ontario; however, wetland areas (∼3%) have not been directly included in this overview, yet they contribute, albeit variably, to water storage and surface water-generated base flow on the landscape (Piggott and Sharpe 2007).

Primary analysis is completed on the elementary terrain and on monitoring case-study-specific datasets. The derived terrain hydrogeological models are then compared with the literature to test, expand, and refine understanding and provide a path for characterization of southern Ontario terrains beyond these seven representative cases.

A full assessment of the hydrology of a landscape describes the complete water cycle and can be quantified as a water budget (Healy et al. 2007), parameterized as shown in Fig. 3. Precipitation (P), which reaches the land and does not evaporate or is not absorbed/transpired through plants (evapotranspiration, ET), either runs off (Ru) the surface and contributes directly to streamflow SF, or it infiltrates¹ the ground and flows as groundwater to contribute to recharge (Re) as it reaches the water table (WT). This is the level below which the ground is fully saturated, and at which level a water well may be productive. Groundwater can flow to deeper regions² (storage, GWS) or it can resurface or discharge to a stream (GWD) to contribute to streamflow (SF). This GWD to streamflow is a component of baseflow (BF); the other component comes from stored surface water (e.g., lakes, wetlands, and reservoirs; Neff et al. 2005). Baseflow and surface runoff to streams combine to yield total streamflow (eq. 4).

A typical watershed budget is a summary of average water fluxes. For example, for the Grand River watershed (Sharpe et al. 2014), precipitation minus ET losses provides a watershed water surplus (WS) of ∼440 mm/year. This quantity, as expressed in mm/year, represents an annual volume of water averaged across the area of the watershed under consideration. This surplus contributes to watershed flow in two ways: by surface runoff to streams (Ru, 260 mm/year) and by recharge of groundwater (Re, 180 mm/year). A substantial portion of groundwater flow resurfaces within the watershed as GWD to surface water and contributes to BF of streams or ponds. Some groundwater flows to groundwater storage (GWS), a portion of which may be lost to discharge outside of the watershed (Jyrkama and Sykes 2007). Groundwater pumping from wells may also remove water from the local groundwater flow system. It is possible to relate stream BF, identified as the slowly varying component of streamflow compared to overland flow, to geological factors (Neff et al. 2005). The geological factors of interest allow for estimates of the amount of groundwater discharge relative to surface water (Piggott and Sharpe 2007). Thus, different surficial geology units each yield a characteristic baseflow index BFi (dimensionless value between zero and one; Piggott and Sharpe 2007), which is baseflow expressed as a fraction of total streamflow (Neff et al. 2005). Geological controls on BFi can also be examined using statistical analysis: see an illustration from the Oak Ridges Moraine area (Buttle et al. 2015). Groundwater baseflow varies considerably from place to place and it contributes almost one-half of total streamflow averaged across southern Ontario. In other words, almost one-half of streamflow is water that has flowed underground (Piggott and Sharpe 2007). The approach in this paper is simply to present baseflow as derived graphically from total streamflow stripped of the short-term overland flow events.

The water budget described above depends on a dataset more extensive than what was targeted for use in the present overview. The object here is to identify the principal characteristics involved in each of our cases based on the subset of parameters (macro indicators) identified in Fig. 3 in red, and traced visually from hydrograph to hydrograph. The presented cases are simple regional representatives of hydrogeological settings, which aim to demonstrate the variable pattern of hydrologic events within surface and subsurface geological units.

Hydrogeological processes across southern Ontario are affected by the physical properties of variable type, thickness, and stratigraphic arrangements of sediment covering bedrock (e.g., Sharpe et al. 2014). The variety of surficial sediment thickness and texture across southern Ontario (Fig. 2) controls rates of infiltration of rainfall and snow melt, primarily as porous media flow (although fracture flow may occur in dense till or clay). Sediment thickness is well mapped regionally due to the widespread availability of water-well records (Gao et al. 2006). Glacial sediment, as much as 200 m thick in areas such as bedrock depressions/valleys and in sandy moraine complexes, has variable lithology, geometry and internal structure, a reflection of complex sedimentation histories (e.g., Sharpe et al. 2002a). Wells in sediment are mainly screened, so the presence of a screen is an indicator of the most transmissive zones in these thick sediment areas (Hinton et al. 2007).

Sediment varies from very permeable (gravel, sand) to impermeable (clay) with all in-between variations possible. Typical rates of flow in gravel and sand are 3–7 orders of magnitude higher than in unfractured silt and clay (Domenico and Schwartz 1990). Depending on grain size, sorting, and packing, sediment can have porosity that is typically 40%–45% (Wolf 1982), resulting in large groundwater storage.

Wells in sediment aquifers often provide abundant potable water because higher permeability leads to ease of groundwater infiltration, storage and flux, and generally low dissolved solids and mineral content (shale-rich tills are an exception) due to short flow paths and residence times. Lower flow rates and longer residence times allow percolating water to take up the chemical character of the finer grained bounding sediment as it seeps toward the sediment-bedrock interface and/or water table. Shallow sediment aquifers recharge from direct surface infiltration and from surface water. In turn, they commonly recharge sediment-bedrock contact aquifers and bedrock aquifers.

In Ontario, the primary bedrock aquifers are either carbonate (i.e., karst terrain) or fractured crystalline rock in areas of the Canadian Shield (see Fig. 2). They are conceptually similar in that flow occurs via visible rock openings, but on different scales of permeability and storage capacity. For example, fractured (jointed) crystalline rocks, such as the granite aquifer in the Tweed case, are markedly different from fractured carbonate rocks that have been enhanced by solution enlargement with increased flow capacity. Shield rocks are typically of low porosity, permeability, and variable fracture patterns, commonly with limited connectivity and usually with low water yields (Gleeson et al. 2009). The presence of joints, faults, and fracture zones allows groundwater flow to take place locally in Shield rocks, at different scales with fresh water down to a depth of ∼100 m due to recent recharge (Gascoyne and Kamineni 2012). Recharge to fracture zones increases where thin (<1 m) sandy sediment covers Shield rocks and provides increased groundwater storage and recharge to bedrock (Gleeson et al. 2009).

Karst-prone Paleozoic carbonate bedrock (Brunton and Dodge 2008) requires special note. Paleozoic carbonate bedrock areas include Manitoulin Island, Bruce Peninsula, Niagara Escarpment, eastern Lake Simcoe, Lake Ontario shorelines, and eastern Ontario (see Fig. 1; Brunton and Brintnell 2020). Paleozoic basin strata form regional-scale, fresh-water aquifers, and near-surface carbonate bedrock units (Brunton and Dodge 2008; Carter et al. 2015; Priebe et al. 2019), are zones for significant potential recharge, particularly in highlands near Niagara Escarpment.

Exposed, or thinly covered carbonate bedrock, eroded by glacial meltwaters or leached by acidic surface and groundwater over thousands of years, forms karst (Brunton and Dodge 2008). The leaching creates enhanced permeability, especially along major joints or fractures in the upper tens of metres resulting in pipe-like groundwater flow and greater connectivity between surface water and groundwater. Distinctive features of karst terrain depend on chemical erosion of various bedrock types (e.g., limestone, dolostone, marble, gypsum, and salt). Such terrains display pocket-like surface depressions, well-developed underground drainage systems, but few surface streams. Interaction of chemical, physical and biological processes may enhance dissolution. Related features such as sinkholes, sinking streams, caves and large springs may result from solution interaction of circulating groundwater and related streamflow (Brunton and Dodge 2008). Surface karst areas represent an important hydrogeological terrain and may connect to deeper paleo-karst horizons (Brunton and Brintnell 2020). One aspect is that shallow and, in some cases, even deeper groundwater aquifers in karst terrains are more susceptible to biological and chemical contamination (Worthington et al. 2012,; Hamilton et al. 2017). Karst conditions are common across southern Ontario and the Napanee limestone plain case study highlights their behaviour.

Hydrogeological terrain types were selected as representative with the help of a simplified version of the surficial geology map of Ontario (Fig. 1; Ontario Geological Survey 2010). Surface geological units, including a simple stratigraphic component, can be related to distinct hydrogeological characteristics because values of baseflow index primarily reflect the extent and properties of the geological units within the watersheds (Piggott and Sharpe 2007; Piggott and Sharpe 2022). The results of these interpretations are spatial estimates of a fraction of streamflow that is due to groundwater discharge and, equivalently, the partitioning of precipitation into groundwater recharge. Hydrogeological characteristics were assessed within long-term (∼30-year) hydrological monitoring records (hydrographs) a proximal climate station (precipitation, temperature), a stream gauge in the principal surface drainage (total streamflow), and a groundwater monitoring well (groundwater level) that could be considered representative of water fluxes in the watershed. A representative year of monitoring was chosen based on observation support, 1974, so that hydrologic events could be compared event-to-response rather than any event correlation being obscured by any longer term averaging. The study focused on understanding the characteristic patterns of hydrological events between watersheds with different geology, rather than carrying out hydrologic or water budget accounting.

Stream-gauge records were a limiting selection factor because of the sparse distribution of such records (Water Survey of Canada) across southern Ontario (Neff et al. 2005; Piggott et al. 2006) and the stipulation of using records within the historic years of observations in proximity to a weather station and a suitable monitoring well. The monitoring wells and associated groundwater-level data are from a monitoring network that was operated from 1974 to 1980 by the Ministry of the Environment. Many of the wells from this historical network have been reactivated since 2001 for inclusion into a recently established Provincial Groundwater Monitoring Network (PGMN³). Small watersheds (<1300 km²) were chosen to improve reliable hydrologic representation and comparison.

The approach of using a selection of readily available monitoring and terrain data to graphically track landscape hydrologic patterns has potential limitations.

Simple graphical hydrological comparisons (climate, streamflow, and water levels) can effectively identify important hydrogeological patterns, for example understanding groundwater levels. These graphical comparisons only characterize a small portion of southern Ontario for a single year of historic well monitoring. Such interpretations require judgement and may overlook other influences on observed groundwater levels discussed below.

A mathematical solution or model of groundwater flow can estimate groundwater levels as a function of temperature and precipitation (e.g., MacRitchie and Piggott 2012). This model accurately estimated levels from historic and current monitoring networks, and it represents analysis over a longer period and interprets the data in a more objective manner than the reported graphical approach. A successful algorithm was tested for the Norfolk sand-plain setting, yet similar test algorithms are needed for other terrain settings.

The availability of long-term hydrological monitoring records, including historic MOE monitoring wells, means that many years of recent hydrologic monitoring did not inform the analysis. Long-term climate data may affect the amplitude of the hydrologic response during changing climate, yet the characteristic patterns of response related to each terrain type are expected to maintain their hydrologic links. Even with changing climates, if excess precipitation is reasonably uniform within each watershed, then the values of the geological component of baseflow index for a watershed can be estimated using values for the units within the watershed (Piggott and Sharpe 2022). The selection of monitoring sites within small watersheds minimizes variations in climate/weather (storm events) across the watershed (Piggott et al. 2006). Trends observed in longer records can help distinguish man made and other hydrological patterns.

Temperature and precipitation data for each station were adjusted for snow accumulation and melting.

Nevertheless, some differences in the magnitude of hydrologic responses might be tied to variable snow cover (e.g., less snow in Essex — more snow in York or Saugeen) may introduce variation in response that would ideally require collecting snow depth monitoring data to address.

The use of one monitoring well for each watershed requires review. Recording the representativeness of watershed flux to a well depends on several minimum items: (i) recharge across the watershed (function of surface geology and slope), (ii) stratigraphic complexity, and (iii) aquifer extent. To address these issues requires significant data gathering, analysis and perhaps numerical modelling. The chosen single watershed monitoring wells were drilled deep enough to access an aquifer that extends to watershed boundaries and to intercept basin-wide recharge. Watershed-scale recharge response is inferred rather than variable local-scale flow paths (e.g., Tóth 1963) by the use of water well cross-sections for each watershed. The result focuses on the main water flux patterns in landscape hydrology using this sparing selection of data.

There is an inherent assumption in the study, that precipitation-melt events are the main driver of groundwater level change, yet other inputs may include surface water seepage and artificial recharge to groundwater. Alternate groundwater level changes are considered in case studies related to pumping (dewatering) during the growing season in the clay- and sand-plain settings and to possible pressure loading effects (e.g., Van Der Kamp and Maathius 1991), reviewed in the clay-plain setting and in Discussion.

The graphical baseflow method used in this study aids simple visual comparison with related watershed hydrographs. Nevertheless, different methods of baseflow separation can produce varying results (Stadnyck et al. 2015). Pertinent to this study, the average values of baseflow index for gauged watersheds across southern Ontario calculated using different methods varied by ∼25% (Piggott et al. 2005; Rutledge 1998). Baseflow indices, however, are closely correlated and provide consistent indicators of at least the relative magnitude of baseflow (Piggott and Sharpe 2022). Yet, a limitation of baseflow index is that it does not measure the persistence of groundwater discharge, which can vary for similar watersheds.

The cause of a relatively low baseflow yield for a watershed carries uncertainties related to (i) basin-wide low permeability, (ii) good surface drainage, or (iii) the stream not being the only place of groundwater discharge in the basin. In addition, groundwater flow is more likely across the boundaries of the smaller watersheds chosen and therefore values of baseflow index for these watersheds may reflect the transfer of groundwater to and from adjacent watersheds.

Elevation is not explicitly assessed in this study; however, it is an important factor in watershed characterization that affects hydrologic function (Devito et al. 2005). Differences in baseflow estimates were identified using a low-resolution elevation model (Piggott and Sharpe 2022): (i) interpretations overestimate baseflow index for gauged watersheds where elevation deviation is less than 50 m and (ii) underestimate baseflow index where elevation deviation is greater than 50 m. Watersheds with this magnitude of topography are associated with prominent features such as the Niagara Escarpment and Oak Ridges Moraine where high hydraulic gradients, deeply incised valleys, and complex three-dimensional (3D) groundwater flow regimes may result in elevated discharge that cannot be estimated solely on the basis of geological factors.

The essence of the seven case studies as presented in Fig. 4 also serves as the legend for the terrain study set (see Figs. 5–11). Each study contains a watershed illustrated by a geology map set on an elevation model (Fig. 4a), and a conceptual hydrogeological sketch (Fig. 4b) showing water movement through the landscape and subsurface. Arrows in the block model show water flow on the landscape and in the ground with differing lengths implying differing fluxes. Vertical arrows indicate a downward-flow gradient; horizontal arrows indicate local to regional flow systems to streams or to lakes or basins, or flow within bedrock.

Three graphs (Fig. 4c) illustrate the key components of the water cycle over a period of 1 year: (i) precipitation and temperature, (ii) streamflow, and (iii) groundwater levels. When rain falls or snow melts (i), it is registered as an event (e.g., main events, E₁–E_x); some water runs off the surface directly to streams (ii); and some infiltrates the ground and recharges the water table (iii) where it contributes to groundwater storage and is measured as water level in a well. Water flow measured in a stream consists of quite variable input from surface runoff (seen as spikes in light blue Fig. 4c, ii), and less variable flow from groundwater discharge into the stream (some possible flow from surface storage); this contributes to baseflow and occurs steadily over extended periods (shown in dark blue).

The three graphs provide climate data, streamflow, and groundwater levels for the purpose of comparison between sites, all taken from the same climate year, 1974. This year had the maximum number of observations, thereby improving the opportunity for site-to-site comparison of hydrogeological characteristics. Note that in all hydrographs, streamflow is expressed in millimetres (mm) as a daily volume of flow per unit area of watershed. This normalization allows the data to be directly compared with millimetres of precipitation, and for better comparison between watersheds. Groundwater levels are in metres relative to an annual mean (Piggott et al. 2006). In each case, the well was drilled into a likely watershed-wide aquifer horizon, five into bedrock (cased or open hole) and two into sediment with screens. If the watershed-wide aquifer horizon was identified correctly, it is supportive of the use of one monitoring well to assess the watershed groundwater system. Marked fluctuations on this watershed-scale, well-level graph are considered to result from water inputs from the land surface, for hydrograph comparison. To aid this comparison, a mean GW level was assigned to indicate levels above or below the zero (0) datum throughout the season (Fig. 4c, I, red dashed line). Plotted in this format, increasing water levels indicate an increase in the volume of groundwater in storage while decreasing values indicate the release of groundwater from storage.

The well is located a short distance from the weather station in (i), and the stream gauge in (ii), so that the watershed-scale proximity of those three monitors is assumed to render a credible representation of water-flux models that supports event-response hydrological comparison and understanding. Most of the case studies (except the clay-plain setting, with strong confined aquifer conditions), for example, show a strong correlation between precipitation/melt events and a response in stream flow and groundwater level readings (event, E₁–E_x tie lines) to assess the presented analysis. Relevant data related to all monitoring sites are included in Table 1.

Total streamflow, groundwater recharge/discharge, and groundwater/water level fluctuations are not always directly related to total annual precipitation, owing to seasonal climatic effects. Seasonal effects determine the form of precipitation (rain/snow) and the movement of water between atmospheric, surface, and subsurface reservoirs. Thus, the hydrogeological response to seasonal variations helps to characterize a site. For example, frozen water (ice and snow) is stored on the landscape and affects the permeability of frozen land sediments. Release of this frozen stored water varies with weather patterns; hence, years with the same total annual precipitation can have variable hydrograph patterns controlled by weather variation year to year and thawing of the ground.

There is also the possibility of surface water influence. A large precipitation event of sufficient regional scope could influence streamflow first, and then streamflow affects groundwater levels. This is possible in the Shield setting where wetlands are prominent.

Other factors, such as sediment thickness, stratigraphic sequence and variation of sediment type with depth play a role in determining groundwater flow into and out of the subsurface, but these factors are not explicitly assessed in this work other than in the thick-till case (Rouge) and Paris Moraine studies. Some monitoring wells in sediment settings penetrate to bedrock to access a watershed-wide aquifer and to improve terrain-to-terrain comparisons, with the primary focus on sediment in the landscape. Contact zone (bedrock/sediment contact) aquifers and flow to deeper storage (e.g., Carter et al. 2015) are not part of the formal analysis.

Seven hydrogeological settings are identified and distributed with respect to simplified surface geology (Figs. 1 and 2). Sediment texture and stratigraphy, bedrock types, physiography and topographic gradients yield distinctive hydrogeological terrains that are representative of larger areas of southern Ontario, five in sediment and two in bedrock (Fig. 1). The movement of water in each setting exhibits typical differences in hydrologic patterns and fluxes as it flows from the surface to the subsurface and back. Wetland areas (3%) have not been included as a monitored case in this overview, yet they variably contribute to water storage and release on the landscape (Bradford 2016).

A clay-plain setting is located in Essex County in the Ruscom River watershed, north of Lake Erie (Fig. 5). The watershed drains northward to Lake St. Clair and forms a low-relief clay plain with minor sand patches in an area of nearly 100% agriculture. A 58.5 m deep well (164A) samples a regional limestone aquifer beneath the laterally extensive, ∼20 m thick, clay layer⁴ (Figs. 5a and 5b). The impervious nature of the clay layer means that there is likely very little hydraulic (flow) connection between the deep aquifer and surface hydrology. Precipitation and snowmelt (i) run off the clay surface directly into the streambed, resulting in very sharp, simultaneous stream-flow peaks, i.e., a typical quick response in the early part of the year (Fig. 5c, ii, E₁–E₅). Crop cover may intercept or reduce peak stream flows during the growing season. These observations, combined with low stream baseflow (BFi = 0.16), indicate limited infiltration to groundwater recharge and discharge. Groundwater levels do not respond to individual precipitation and snow melt events and there is minimal seasonal variation of levels (<0.6 m) although some recession of water levels does occur during July and August, followed by a gradual increase during November and December and May (subtle changes in comparison with other case sites). The summer water level recession appears to be a result of seasonal groundwater withdrawals likely for irrigation in this agricultural landscape, with subsequent redistribution of groundwater within the aquifer. Short-term (∼1–2 day) variation in groundwater level (appearing at this scale as a fine saw-tooth character) is likely related to changes in atmospheric pressure and indicative of confined conditions in the aquifer as imposed by the thick, continuous clay cover. Other short-term variations with slightly larger amplitude appear to line up with climate, surface water and stream flow events (e.g., E₂–E₅), yet it is not clear how hydraulic connection could be established. Very limited shallow groundwater flow and little groundwater-level response due to confined conditions at depth typify this setting. It is likely that other, low-permeability, clay-plain settings in southern Ontario (Fig. 1) function in a similar manner, behaving as two almost separate hydrologic systems, namely the deep (bedrock) aquifer and the shallow fractured clay. Note: the bedrock aquifer is regional in extent (Fig. 5b) and likely tied to Lake St. Clair, a possibility to test by examining lake-level trends for the same monitoring period.

Detailed hydrogeological studies in the area of southwestern Ontario provide support for the above analysis (Ruland et al. 1991; Husain et al. 1998), and confirm the conceptual model of water movement for a clay-plain terrain setting. Fractures in the clay-plain surface indicate an active infiltration and shallow groundwater flow in the top 5–10 m below ground, despite very low measured hydraulic conductivities (10⁻⁷ to 10⁻⁸ cm/s; Ruland et al. 1991). Fractures most likely form in warm, dry summer weather when the precipitation surplus is low. Hydraulic head and isotope profiles show little flow deeper than ∼ 10 m except by molecular diffusion (Ruland et al. 1991) with little or no advection (Husain et al. 1998). Dating of deep groundwater (clay pore water at 30 m) shows ages of ∼9–10 ka compared to modern water in the fractured flow system (Desaulniers et al. 1981). These results explain the very low response of the regional aquifer to near surface hydrological events (Fig. 5) and the long duration records of separate shallow and deep flow systems.

The above results indicate that in this confined clay basin setting it is highly unlikely that surface water is getting to the bedrock aquifer; neither in short 1–2 day, nor in seasonal timeframes in this confined clay basin setting. Most short-term water level change indicates barometric pressure fluctuation, while some small seasonal effects may indicate pressure-induced hydraulic loading, (e.g., van der Kamp and Maathius 1991). Changes of total moisture mass above an aquifer occur in all observation well records, particularly for confined aquifers (van der Kamp and Schmidt 2017) yet additional data (barometric readings) and analysis are required to assess these effects.

This sand-plain setting is located within Venison Creek watershed in Norfolk County, northwest of Long Point, Lake Erie (Fig. 6). The sand plain, with its level to very gently rolling terrain and 10–20 m deep incised stream valleys, is a classic tobacco-growing area in which the soils are well drained, in sharp contrast to the Essex clay plain. A 23.8 m deep well (413A) penetrates a sand aquifer in a laterally extensive, low-relief, sand plain (Figs. 6a and 6b), which extends across the watershed. Streamflow in Venison Creek responds readily_; but not nearly as rapidly as on a clay plain, and with lower peak flows to precipitation and snowmelt events (the latter inferred from the temperature graph) (Fig. 6c). This, together with substantial baseflow (BFi = 0.78), indicates significant surface infiltration and groundwater recharge and discharge (i.e., hydraulic connection) and results in the rather gradual drop off in streamflow after each precipitation event (Fig. 6c, E₁–E₄). Four substantial increases in groundwater level (notice wide, modest curves coinciding with streamflow and precipitation events) in the first few months of the year are further indication of surface infiltration, storage capacity and permeability, all of which contribute to maintenance of steady baseflow throughout the year. Of note is that the rise in GW groundwater level is slightly offset from a precipitation event. This delay indicates that there is timing separation between the surface events (P, runoff and streamflow) and the well, due to the time necessary for water to move through the groundwater flow system (sand aquifer) to the stream. With the onset of summer and elevated ET, precipitation-triggered streamflow enhancements cease, and water levels drop. Water level decline likely includes pumping from irrigation wells. However, a noticeable level of baseflow persists, consistent with an efficiency of groundwater recharge and storage, despite an insufficient amount of summer precipitation to contribute directly to streamflow. This behaviour is likely representative of southern Ontario sand-plain settings, a relationship that was recognized in an earlier study using water well records (Singer et al. 2003).

The Norfolk sand plain hydrologic characteristics and monitoring analysis are supported by recent watershed studies (Novakovic and Farvolden 1974). Infiltration and recharge in the sand-plain setting contributes to substantial groundwater storage and discharge as recorded in a high base flow index, 0.78 (compared to low BFi, 0.16 in the clay plain). The sand-plain aquifer extent and storage also contributes to a sizeable water supply, with predictable gradients toward main streams and Lake Erie. Singer et al. (2003) reported that the water table is <10 m deep in most of ∼1000 aquifer wells, and yields ∼275–2300 L/min occurred in 53 wells across the regional-scale Norfolk sand aquifer. There is also sufficient base flow consistency to support a trout fishery (Stadnyck et al. 2015). Thus, in contrast to the clay-plain setting, infiltration in sand plains creates and maintains baseflow through the year, whereas run-off from clay plains bypasses infiltration, recharge and groundwater storage with no response in water level or in stream (baseflow) discharge.

A stratified porous-medium aquifer with abundant sand and gravel straddles the Eramosa River and Blue Springs Creek watersheds (Fig. 7). It is part of the rolling, hummocky Paris Moraine, north of Guelph, Wellington County (Figs. 7a and 7b), a watershed with elevation relief of >150 m from north to south. It is an agricultural and mixed forest area with 12% wetland. The landscape is capped with thin till and dissected by glacial meltwater channels that cut into the sand/gravel aquifer. The lower reaches of Eramosa and Blue Springs creek flow on carbonate bedrock, with karst dissolution features. The aquifer sampled by a 32.0 m deep well (432A), has a screen in sand and gravel (∼28–32 m), <5 m above dolostone bedrock. The moraine uplands present a more variable terrain as compared to the first two cases; more complex surface geology and stratigraphy, greater and enclosed topographic relief, and resultant hydraulic gradient and pathways. In spring, streamflow shown for the Eramosa River exhibits rapid onset in response to precipitation events and snowmelt, a result of large peak flows (three main events, E₂–E₄), from drainage across till-covered slopes and from wetland-gravel channels (Fig. 7b). A significant late November precipitation event (E₅) shows a muted stream flow response compared to similar spring events, likely due to depleted soil moisture contents. Baseflow (BFi = 0.59) is significant and persistent, attributable to a combination of ample permeability and groundwater storage in coarse sediment. Groundwater levels (Fig. 7c) also respond to precipitation and snowmelt with relatively large, step-like increases; those increases are followed by short, then by a prolonged gradual recession in the second half of the year. Groundwater level response is slightly delayed before it reaches a peak, compared to precipitation and streamflow (E₁–E₃); this likely is due to the time for groundwater to flow through the complex moraine geology and to discharge to streams. The late November precipitation event (E₅) shows a similar muted groundwater level response that likely relates to replenishing depleted groundwater storage after summer loss to discharge.

The 2-week recession of groundwater level in late May coincides with a precipitation event and a stream-flow event. This could be a result of a small, isolated precipitation event during elevated ET, perhaps creating a soil-moisture deficit (plausibly connected to the May drop in water level). Alternatively, the peak could be considered as a delayed increase, possibly the effect of temporary storage in valley wetlands. Otherwise, the groundwater-level profile indicates rapid recharge and storage in sandy upland aquifer sediment. This is followed by recession discharge from uplands, then longer, slower recession discharge from valley aquifer storage and wetlands. An important hydrological factor is that stream-gauge response for Blue Springs Creek sub-watershed (Fig. 7b) has more baseflow contribution from Paris moraine sediments than the gauge response for the Eramosa River. Blue Springs Creek subwatershed has ∼50% moraine cover (topography and higher permeability sediment, including 27.9% gravel content in well records), while Eramosa watershed has only ∼25% moraine sediment. Streamflow per unit area for Blue Springs Creek is larger (∼30%) than for Eramosa River, likely reflecting added discharge from sandy gravel moraine sediment. It is also possible that Blue Springs Creek watershed receives additional groundwater flow from gravel beds and near surface karst flow that extend in the subsurface beyond its surface drainage divide boundaries.

The complex setting of the Paris Moraine also includes groundwater flow in bedrock. Groundwater flows through shallow portions of the underlying carbonate bedrock and it may discharge to surface water, both Eramosa River and Blue Springs Creek. This flow is the result of the north to south slope and gradient on the bedrock surface topography (Burt and Dodge 2016), with flow and discharge assisted by interconnected karst pathways (Priebe et al. 2019). Well 432A is screened in sand and gravel ∼5 m above bedrock (Logan et al. 2020) with possible connection to the shallow bedrock groundwater flow system. The bedrock surface variation includes an 80 m deep, 1 km wide bedrock valley with ∼40 m of sand and gravel fill, trending NNE-SSW (Burt and Dodge 2016). The bedrock valley lies east of the study watershed but may induce groundwater flow from the area.

Groundwater monitoring shows a significant recession (∼2.5 m) in water levels from July to November (Fig. 7c, iii). This depletion in groundwater storage is due to summer-fall discharge from an inferred connected sediment, shallow karst bedrock reservoir and flow system. It may be speculated that discharge from this system may have left the watershed via gravel channels that cross the landscape (Figs. 7a and 7b). Storage loss may also be due to downward gradients in karst-connected flow in bedrock (GRCA 2013) close to the Niagara Escarpment, a notion in part signalled by low base flow to streams during this period. In addition, it appears that Blue Springs Creek is gaining groundwater across the watershed boundary based on its larger long-term value of baseflow recession (more persistent flow) compared to Eramosa River, a result that is supported by modelling (MacRitchie and Piggott 2012). A decline in base flow is expected during the warm summer-fall period when WS is low. In contrast, storage capacity in the sand-plain study is sufficient to maintain persistent baseflow throughout the summer-fall interval (Fig. 6c, iii).

The rolling upland of the north Saugeen River area, southwest of the town of Chatsworth in Grey County, is a thin till setting (Figs. 8a and 8b). The shallow till landscape is dissected by meltwater channels containing sand and gravel. The total sediment package is thinner (<20 m) than the Paris Moraine terrain (∼37–80 m), and has much less water-storage capacity in the porous sediment. A 115.2 m deep well (194A) samples a deep carbonate (limestone) aquifer that provides substantial storage capacity beneath the thin sediment cover. Streamflow in the North Saugeen River near Paisley responds strongly to precipitation and snowmelt throughout the year, in part as run-off from upland slopes (Fig. 8b). Overland flow is pronounced on the rolling slopes in winter and early spring when ground may be frozen and (or) saturated from rapid melting. There seem to be more summer and fall precipitation events here than at other case study areas (likely lake effect); however, the events register as definite but diminished streamflow with little effect on groundwater level (except early spring events, E₃); later events (E₅, E₆) possibly record an effect of ET typical of an agricultural area. Baseflow also occurs throughout the year from ample groundwater storage, particularly in spring and again in smaller amounts in early winter (Fig. 8c). This indicates high rates of groundwater recharge, storage and discharge, depending on whether from channels of permeable sediment or from bedrock reservoirs. Evident is <1 m annual water-level variation (Fig. 8c) in comparison to ∼3 m variation noted in the Paris Moraine (Section 3.3). The spring baseflow likely derives from sediment; early winter baseflow likely derives from bedrock storage, but the change in water level is only slight because of the large size of the bedrock reservoir as the watershed-scale aquifer. Strong correlation between precipitation and streamflow, together with water-level increases in January to April imply good hydraulic connection (three main spring events (E₁–E₃; Fig. 8c). Good connection is to be expected as the stream intersects gravel and flows on bedrock through the lower part of the watershed; thereby, such GWD affects the stream gauge promptly. The January to April step increases in water level are followed by very slow recession during summer and early fall (Fig. 8c), an effect related to considerable groundwater storage. Short-term variation in levels (sawtooth character) occurs throughout the year, probably related to changes in atmospheric pressure, a sign of confined aquifer conditions (the shale, silt-clay layers acting like a semi-continuous aquitard and providing a partial seal within bedrock). Geological complexity includes relatively thin gravel and till sediment, with the river running in places on bedrock (giving way to persistent, ready discharge) and zones of possible karst conditions (see karst example in Napanee; Section 3.7) in a deeply connected flow system. The overall result is rapid groundwater recharge and slow recession of groundwater levels due to considerable storage, mostly in bedrock.

The north Saugeen mid-watershed well is cased to 41.5 m and is open below to 98 m in dolostone bedrock. The multiple levels (4) at which water entered the well below casing (>41.5 m depth), explains the minimum influence of near surface precipitation events on the watershed-wide groundwater system. Large, mainly spring events (E₁–E₃) show clear, but muted, slightly delayed water level responses to surface hydrologic events. The general water level trend shows a delayed water level high compared to the stream flow high, indicating hydrologic connection (recharge) across ∼100 m depth of groundwater flow, in addition to small-scale barometric signals. Such depth of marked hydraulic response supports the inference of karst-enhanced bedrock flow. The deep bedrock well in the clay plain watershed shows a similar subdued groundwater level response as the deep well in this Saugeen watershed.

A gently undulating agricultural and urban area in York Region, near Markham, has a thin cover of sand and clay-silt resting on the well-mapped thick regional Newmarket Till (Fig. 9a). To the north, stratigraphically overlying sandy Oak Ridge Moraine (ORM), drapped by thin Halton Till, is an important topographic upland (>275 m asl) forming the northern margin of the Rouge River study watershed. A 30 m deep well (398A, collared at well at ∼190 m asl) provides access to groundwater from a significant sand, silt and clay aquifer below regional till (Sibul et al. 1977). The well screen is set into Thorncliffe Formation (Sharpe et al. 2002a) at ∼21–23.5 m depth (just below water level) beneath the regional Newmarket Till aquitard, caped here and there by thin Halton Till. Streamflow in the Rouge River near Markham gauging station, responds sharply to snowmelt or precipitation, mostly in the spring (four main events) with response decreasing in the summer with only minimal response in the late fall (Fig. 9c). Overland flow is pronounced in winter and early spring when ground cover may be frozen and (or) saturated (particularly sand), to create rapid flow events. The effect is enhanced by large areas of pavement in the adjacent urban areas (∼ 30%; Buttle et al. 2015^,5). Baseflow is modest year-round and low but persistent during summer months as to be expected because of ET loss. Baseflow is predominantly supported by groundwater discharge to the river from the ORM aquifer higher in the watershed, and from the Thorncliffe aquifer (closer to Markham; Hinton 1995, 2000). In addition, baseflow may be augmented by release from a shallow, sandy surface aquifer. Groundwater level (Fig. 9c, iii) shows a very distinct response of moderate amplitude to precipitation and snowmelt events in spring, a local effect likely commensurate with the 5 m thick surface sand layer around the well site (Fig. 9b). A very modest recession of groundwater level occurs during mid-summer and early fall when inputs are low, likely sustained by the substantial reservoir capacity of the Thorncliffe aquifer (Sharpe et al. 2011), where Newmarket Till has been eroded away by the Rouge River, to expose Thorncliffe sediment in the riverbed. The sharp recession from individual events (of a spikey nature) illustrates the rapid response to surface hydrological events (runoff from hard Newmarket Till, clay and urban surfaces). Baseflow contribution from the Newmarket Till is low. The contribution is up to five times greater from ORM aquifer headwater streams overlying Newmarket Till (Fig. 9b), where flow from headwater streams in the north increases downstream near Markham. Near Marlham, the Thorncliffe aquifer is intersected by the Rouge streambed (Hinton 1995; see front panel of conceptual model, Fig. 9b). In sum, potential groundwater contributions to Rouge River flow come from ORM, from near-surface glaciolacustrine sand (minor) and from deeper Thorncliffe Formation aquifers.

The considerable geological and hydrogeological research in this apparently “simple” till plain area reveals a more intricate hydrogeological function with multiple aquifers in a complex stratigraphic setting; this extends the above analysis based on a limited set of monitoring and terrain data. Improved understanding of stratigraphic complexities of a thick till plain flow system, bordering the ORM upland, accrued using 3D mapping, modeling, hydrogeological analysis, geochemistry and baseflow surveys (e.g., Sibul et al. 1977; Howard and Beck 1986; Gerber and Howard 1996; Sharpe and Barnett 1997; Gerber and Howard 2000; Meriano and Eyles 2003; Hinton 1995; M.J. Hinton, personal communication, 2022; Logan et al. 2006). This collective work supports the hydrograph analysis and shows flow through a multi-aquifer complex of four aquifers and aquitards that control surface and groundwater flow. The regional Newmarket Till has such low hydraulic conductivity that as little as ∼35–50 mm of recharge penetrates the till plain per year (Gerber and Howard 2000). The Halton Till aquitard also reduces recharge to less than 150 mm (Gerber and Howard 1996), while the sandy ORM uplands generates ∼300–400 mm of recharge per year that contributes to flow in the Rouge watershed (Hinton 2000; Meriano and Eyles 2003). In addition, baseflow surveys along the Rouge watershed showed a partitioning of streamflow due to groundwater discharge from ORM, from the Thorncliffe aquifer system (Hinton 1995; Buttle et al. 2015), and perhaps from the 5 m thick surface sand area, which mutes expected runoff from Newmarket Till.

The use of a single well to characterize the hydrology of a watershed with complex stratigraphy (multiple aquifer and aquitard properties and geometry), may be inadequate for the Rouge watershed case study. The role of dense Newmarket Till in creating rapid flow to streams is illustrated and the resultant flashy runoff and modest well response is as expected in a low permeability terrain setting (Meriano and Eyles 2003).

It is not expected that the water level in the Rouge monitoring well 398A should show a spikey pattern in groundwater >20 m below a low permeability Newmarket Till aquitard (Gerber and Howard 2000). This spikey response requires a different explanation. It appears that Rouge River flow is directly connected to the Thorncliffe aquifer through gaps or breaches in Newmarket Till (Sharpe et al. 2002b). The Rouge River flows across Thorncliffe discharge (Fig. 9b, MC), is recorded in a significant increase in baseflow (∼3 times higher; M.J. Hinton 2000 unpublished).

The commonplace low-relief Canadian Shield terrain is represented here by the forested Moira River watershed, Hastings County (Fig. 10). Undulating Grenville bedrock structure comprising gneiss, marble and metavolcanic rocks controls the local topography. Thin, sparse sandy till, sand and silt cover rock here and there between the many lakes and wetlands. Palaeozoic rocks are exposed immediately to the south (Fig. 10a). A 12.2 m deep well (230A) samples a jointed-granite aquifer in wetland, Shield terrain near Tweed (Figs. 10a and 10b). Streamflow in the nearby Moira River responds to precipitation and snowmelt correlated to peak flows (5 main spring events, E₁–E₅), somewhat subdued in amplitude in comparison to the peaks in water level, (Fig. 10c), albeit with slight delay. This pattern, along with persistent medium-size baseflow (BFi = 0.54; Neff et al. 2005), suggests early-season groundwater recharge, storage, and discharge along with surface storage in wetlands and in forested areas (Devito et al. 1996), that moderate and contribute delayed runoff to streamflow. Most precipitation events show a direct response in well water levels throughout the year (E₁–E₇; Fig. 10c, ii). Significant baseflow occurs during winter, especially extending into spring and then again late fall, perhaps when storage is full, and ground is not frozen (before April pulse). Baseflow during summer and early fall, however, is minimal in part due to small storage capability in fractures of Precambrian bedrock, which is the predominant storage mode (Wolff 1982). Groundwater flow within bedrock likely occurs via interconnected fractures that have limited capacity to retain groundwater for discharge during lower recharge periods, i.e., summer and early fall (Fig. 10c). The observation well, located very close to the Moira River, appears to indicate that groundwater levels are directly influenced by surface water levels and vice versa, as observed water level follows Moira streamflow very closely. Borehole water level starts to climb in September along with baseflow, likely because ET has fallen off and the wetlands are leaking into the bedrock (wetland bottoms being somewhat retentive by way of fracture infilling with fine-grained sediment and organic material), and leaking into streamflow. The direct hydraulic associations between well and river via fracture flow, and the fact that flow into the granite aquifer may be restricted by other overlying, lower permeability sediments, means that a relatively small exchange of surface water may influence groundwater levels for some distance from the river.

An important aspect of Shield hydrology is the role of surface water and the prominence of wetlands in the area. Observed baseflow is not all related to groundwater discharge. When recharge paths are inactive, it is more likely that wetland discharge creates baseflow, particularly in winter (Buttle et al. 2004). Wetland base flow is likely consistent due to wetlands comprising 14% of the Moira watershed⁶, and it may help explain why base flow tails off in summer when greater ET from wetlands uses up excess water. It is also possible that wetlands leak directly into the Moira River once ET slows down. Thus, connection to and contribution of water from the low-storage, fractured rock groundwater system is minimal (Gleeson et al. 2009). The low rates of groundwater discharge are noted in previous studies of small watersheds in the Canadian Shield; they also indicate that groundwater discharge is limited where sediment cover is minimal (Buttle et al. 2004). The direct response of groundwater to precipitation events, as measured in rapid rises in hydraulic heads (Miles and Novakowski 2016) confirm the low storage yield in fractured Shield rocks. In addition, these results show that as little as 0.4 m of sediment cover can eliminate the rapid well response on exposed Shield rock surfaces. For comparison, Shield watersheds, such as Moira River, have a ratio of groundwater discharge to precipitation that is only ∼1/3 that of groundwater discharge to precipitation in watersheds with porous sediment (Gleeson et al. 2009).

The Wilton Creek watershed setting is a Paleozoic carbonate terrain with thin-sediment cover. It occurs in a meltwater-eroded bedrock valley, ∼25 m below the main Napanee limestone plain. Scattered, drumlinized sandy/silt till uplands host agricultural fields. Wilton Creek valley has discontinuous sandy fan deposits (Shaw 1988). This landscape is difficult to assess because of the possible existence of deep solution cavities within the limestone (karst features — Gorrell, personal communication, 2022) which have little surface expression and cannot be considered in the realm of fractured or porous-media flow aquifers as our other cases lend themselves. A 31.7 m deep well (478A) samples a shallow unconfined fresh-water aquifer in limestone (Figs. 11a and 11b) northeast of Napanee. Streamflow, in Wilton Creek, responds immediately to precipitation and snowmelt as water runs off the limestone plain (Fig. 11c) with minimum modification to high surface flows by patchy, thin till. At the same time, groundwater level shows rapid increases and abrupt decreases in response to precipitation and snowmelt as precipitation percolates readily into openings in the limestone. A protracted period of low streamflow occurs during summer growing season and early fall when no precipitation events are reflected in streamflow. Groundwater levels also decline dramatically (∼5.7 m) beginning in mid-May and continuing until mid-October; they are lowest from mid-August to late September. The relatively constant levels that occur during this late summer period may indicate that groundwater storage provides minimal discharge to surface water consistent with the modest baseflow index that occurs during the period (seasonal BFi = 0.38). These characteristics may be the result of a groundwater-flow regime that is connected to upland surface water as rapid drops in water level lead to very low local groundwater gradients, and therefore low baseflow. However, one can speculate that intricate, pipe-like pathways in the limestone terrain provide connectivity to large-cavity groundwater storage at greater depth (below valley bottom?). In this scenario, the valley bottom acts as a drain to inter-valley uplands, creating loss to deeper aquifers below valley bottom. If this is the case, then the incised limestone plain can be considered a two-level/two-stage-storage hydrogeological landscape. The valley-bottom drain effectively draws groundwater down rapidly from upland terrain even with replenishment by considerable spring precipitation; and in summer months, the shortage of upland water is aggravated by ET loss. Baseflow originates only from shallow (upland) limestone fractures and from sparse sediment and soil cover during winter and spring times, as the main limestone aquifer level is below the Wilton Creek stream bed and is ∼25 m below the upland. The resultant rapid variations in streamflow, particularly very low summer baseflow, may be enough to affect in-stream water quality and stress the aquatic ecosystem.

The Napanee limestone plain presents some unusual hydrographs. First, a 40 mm precipitation event in April created a nearly 20 mm response in stream flow, the largest response of all case studies. Second, this unusual response, and related spring responses, experienced rapid declines in well water levels after each spring rainfall-snowmelt event. This pattern suggests flash flow across the limestone uplands, followed by the lack of stored water in sediment or in shallow limestone cavities to sustain storage gain. Only the clay-plain setting comes close to this flash flow behaviour. Third, and more significant, is the steep decline in groundwater levels (<6 m) following the initial spring events. This is the largest drop in groundwater storage in all of the presented case studies. Possible downward gradients in Wilton Creek help to explain any loss of stream flow due to karst flow paths. Steep gradients at the ∼25 m deep valley margins of Wilton Creek readily drain upland solution-enhanced fracture and cavity groundwater storage.

For comparison, the next largest groundwater storage loss is a decline of ∼3 m in the dolostone aquifer beneath the Paris Moraine. Dolostone in the Eramosa River area has karst-enhanced flow (Priebe et al. 2019), which prompts the speculation that such large seasonal groundwater storage swings are typical of karst terrain.

The key hydrogeological attributes of the various terrains of southern Ontario, summarized in Table 1, include information on weather stations, stream gauges, well monitors, and watershed attributes. A broad distinction considers whether the overlying sediment or the bedrock predominates in defining the hydrogeological behaviour of the terrain. This landscape distinction applies even if the chosen watersheds are located in areas where sediment or rock roles are not apparent from regional mapping (Figs. 1 and 2). The cases are arranged in order of increasing complexity to promote systematic development of an appreciation of the important terrain attributes. The five sediment settings all involve flow through intergranular porous media in the unconsolidated stratigraphic units (e.g., Heath 1988), yet recharge response is mainly measured in regional flow in bedrock aquifers. The two bedrock terrains depend on entirely different types of groundwater flow: Shield rocks involve fracture flow and carbonate rocks often entail pipe flow in cavities as well as solution-enhanced fracture flow.

Few studies (short of numerical modelling) combine essential hydrological data (precipitation/snow melt, streamflow and water level monitoring) as short-hand indicators to assess the flow of water from the land surface into the subsurface to groundwater storage or flow and discharge to streams, ponds and wetlands. Visual hydrograph comparisons in archetypal case settings allows one to infer the amplitude, rate, timing, scale and reservoir fate of water movement in most southern Ontario landscapes, and to permit some generalizations.

Five chosen sediment settings represent a range of sediment characteristics. Two of these sediment settings show end-member contrasts in hydrological behaviour. Clay plains display high surface runoff and little infiltration, and thus, there is little if any recharge, storage, baseflow or fluctuation in groundwater levels. Low-permeability clay plains also create confined, pressure-sensitive conditions in the underlying aquifer. Sand plains, on the other hand, readily allow rain and snowmelt to infiltrate the surface, transmitting flow to water table and groundwater storage, particularly in late winter and spring. This stored water releases slowly under unconfined conditions throughout the summer as discharge to streams when rainfall inflow is significantly less, and ET loss is high. The contrasting hydrogeology of sand and clay plains affects water availability throughout a typical year over areas of comparable terrain covering about a third of southern Ontario. Clay plains likely support few in-stream species because of sporadic, flashy flow, whereas sand plains likely support more abundant in-stream species as they absorb more water and sustain more constant, cool baseflow (e.g., Stanfield et al. 2009; Stadnyck et al. 2015).

Upland settings differ from either clay or sand plains, in part due to their higher gradients. Sand/gravel channels also dissect both till-covered upland settings, i.e., North Saugeen River and Paris Moraine uplands. Eroded slopes and channels promote rapid recharge and steady baseflow from infiltrated run-off within till-upland landscapes and from direct recharge to sand and gravel in channels. Thick coarse-grained sediment in Paris Moraine allows for a much larger fluctuation of groundwater levels than the thin sediment found in the Saugeen uplands (∼3 m compared to 0.8 m annual water level variation, or groundwater storage). The Napanee limestone plain also shows a similar flashy runoff pattern as the Essex clay plain, and its upland position provides enough gradient to create large annual groundwater storage fluxes perhaps typical of karst modification. We speculate that in the Saugeen uplands, spring-fall baseflow likely derives more from sediment storage, whereas summer baseflow is lower and may derive more from bedrock storage as the north Saugeen River runs over karst-prone bedrock.

No direct comparative hydrogeological terrain describes the importance of sandy stratified moraines (e.g., ORM, Oro, Waterloo moraines) based on the selection criteria of small watersheds with suitable hydrographs of climate, stream gauging and groundwater monitoring well, although the Paris Moraine setting offers an approximation. The role of stratified moraines becomes apparent by comparing two streams that rise in the hummocky topographic ridge of the Paris Moraine (Fig. 7a). Streamflow per unit area for Blue Springs Creek is larger (∼30%) than for Eramosa River, reflecting added discharge from twice the area of sandy, upland moraine sediment, normalized to watershed area. In the Paris Moraine area, it is possible that Blue Springs Creek also receives groundwater flow from outside its watershed boundaries (possible karst bedrock flow). These same attributes of landscape response occur in the Oak Ridges Moraine where a greater sandy moraine upland with recharge up to 400 mm per year results in added baseflow (Buttle et al. 2015), and, groundwater flows also extend beyond watershed boundaries (e.g., Hinton 1995; Gerber and Howard 2000).

The thick sediments of the Newmarket Till plain in the Rouge River watershed illustrate a multi-aquifer-aquitard landscape, supplemented by detailed hydrogeological context (e.g., Gerber and Howard 2000). The Newmarket Till plain exerts a significant control on surface hydrology by diverting most surface water directly to streams.  Recharge and baseflow on Newmarket Till increase where surface sand covers the till (Fig. 9a). Water also infiltrates to deeper groundwater flow where meltwater channels breach the Newmarket Till aquitard (e.g., Sharpe et al. 2002a, 2002b; Fig. 9b). Incision of the Rouge River beneath the base of Newmarket Till allows an underlying regional aquifer system (e.g., Thorncliffe; Sibul et al. 1977; Gerber et al. 2018) as well as shallow aquifers (e.g., ORM, surface sand) to contribute variable discharge to the Rouge along its course. ORM, with elevated recharge and topography, discharges to streams through multi-layered sequences and incised channels in the area, both within and from outside the Rouge River watershed boundaries (e.g., Hinton 1995).

The two terrains classified as bedrock dominant also illustrate distinctive hydrogeological behaviour. Both allow precipitation and snowmelt to run off readily to streams and to infiltrate rock openings. The Shield setting with thin, discontinuous sediment and organic/wetland deposits on low-relief, fractured rock has significant baseflow during wet periods, winter, spring and late fall, when storage is full, yet small storage capacity in fractures results in minimal flow during recession. Significant baseflow is likely delayed discharge from surface storage in prominent wetlands in the area (e.g., Moin and Shaw 1986). Fracture-flow discharge is limited during summer (e.g., Gleeson et al. 2009) when levels are low; on the other hand, it appears that well levels, when close to rivers, can show direct influence from fluctuating river stages.

The thin-sediment-covered Paleozoic carbonate terrain is widespread in the region. Fractures and joints enhanced by dissolution in carbonate rocks readily provide infiltration (Worthington 2002). Karst dissolution, common in limestone terrain (Brunton et al. 2007), is a more pronounced version of joint enhancement: it exerts a large influence on the amount, timing, and distribution of groundwater recharge as well as the depths and distances of active groundwater flow and discharge as observed in the limestone terrain near Napanee. Groundwater storage is significant in karst terrain yet can be reduced in summer when upland water levels fall below discharge points within incised bedrock valleys (see Winter 1981; Fig. 2), and streams such as Wilton Creek may effectively go dry. The large seasonal groundwater storage drops, ∼3–6 m, in the Paris Moraine dolostone and the limestone aquifer at Napanee, point to the notion that such large groundwater storage swings may occur in other karst terrains.

Better determining the factors that affect groundwater levels is one theme in which the sparse data approach used in this study could benefit from long monitoring records and other attributes (e.g., barometry) to improve the identification and characterization of representative hydrogeological terrains⁷. For example, groundwater level change may include man-made inputs, pumping, artificial recharge, and natural inputs, precipitation, surface water seepage (recharge) to the water table, as well as under-recognized moisture loading effects. Changes of total moisture mass above an aquifer (snow accumulation, soil moisture, and storage at the water table) represent changes of mechanical load acting on the aquifer, effects that occur in all observation well records, in particular, confined aquifers (van der Kamp and Schmidt 2017). Removal of barometric and other effects from observation well records may allow one to identify short- and long-term moisture loading for comparison with hydrological observations, such as effects in response due to precipitation, runoff, and ET.

An important issue with groundwater level changes is whether possible moisture loading is interpreted as a groundwater recharge event.  For example, in the Essex deep well (Fig. 5c, iii), an approximate 100 mm rise in water level in the first half of the year may be moisture loading (snow) rather than surface recharge, in light of the absence of recharge as reviewed above (e.g., Ruland et al. 1991). The 5 years of hydrograph records following 1974, show a small but steady decline in the average water level (∼200 mm), apparently due to irrigation pumping from the aquifer. In the second half of 1974, an approximate 200 to 300 mm decline of the groundwater level likely confirms irrigation pumping of the aquifer. While difficult to confirm in this report, it is possible that moisture unloading due to ET loss explains part of the 200–300 mm decline. Moisture loading was documented rather than recharge in annual fluctuations for deep observation wells in Saskatchewan, (van der Kamp and Maathuis 1991). Thus, if water level changes are interpreted as groundwater recharge rather than moisture loading (if confirmed), an over-estimate of recharge to aquifers can result.

Building upon the significant terrain concept “hydrologic landscapes” (e.g., Winter 1981; Devito et al. 2005; Buttle et al. 2015), the paper endeavours to identify the geological and physiographical controls that determine how water typically moves through southern Ontario landscapes, encountering very different hydrogeological responses depending upon terrain. Specifically, variations in surface sediment cover, type and thickness, bedrock, porosity and permeability, physiography and topographic gradients combine to yield distinctive settings. Readily available climate-, stream- and well-monitoring data integrated with local terrain information, allow one to assess the response of typical settings. The seven hydrogeological settings (clay, sand, gravel, thin till, thick till, granite, and limestone) characterize ∼90% of all terrain with variations based on the simplified geological map for southern Ontario. As a result, this terrain framework highlights the value of ongoing watershed-scale hydrological monitoring as the main hydrogeological characteristics can only be derived from hydrograph trends for each terrain. These insights may be useful in anticipation of hydrological stresses under changing climatic scenarios, or in any case helpful to guide water-management scenarios at the regional and watershed scales (e.g., Allen et al. 2013).

Considering issues identified in this is paper (e.g., single watershed well, longer climate records, more terrain types, etc.) the present study constitutes progress as a proof of concept that can be improved with added data support, analysis, and literature context.

This study combines readily available water monitoring (weather, stream, and groundwater) and terrain (relief and geology) data to describe seven hydrogeological terrain responses that are representative of water movement on and below the landscape across most of southern Ontario. The proof-of-concept approach, data and watershed selection process is efficient and leads to significant insights into the patterns of hydrologic response across the landscape, less so on the hydrologic accounting of the terrain. Results of this hydrologic terrain characterization are consistent with those of more detailed hydrogeological analysis (e.g., Buttle et al. 2015), yet an improved set of predictive hydrologic terrain models can be advanced to increase the scope and refine this characterization with added data support and analysis. The derived hydrologic terrain models can inform water and land resource management related to water availability, protection, and aquatic habitat as future climates change.

Andrew Piggott contributed to this work when he was a Research Scientist with Environment Canada (now Environment and Climate Change Canada). He declined to be an author for private reasons. Andrew contributed the essential hydrological datasets and analysis for the paper. He also provided three terrain case studies, included here, to a chapter on the Hydrogeology of southern Ontario (Sharpe et al. 2014). A version of this paper was presented at an annual session on the Regional-scale groundwater geoscience in southern Ontario: Hydrogeological terrains and typical settings of southern Ontario, February 2019 (Russell et al. 2019). Alf Dyck provided guidance on the structure, clarity, and detail of the paper. Reviews by Marc Hinton, Steve Holsh, Riley Mulligan, Robert Sharpe, and Hazen Russell are much appreciated, as are comments from Garth van der Kamp and David Ferris. Charles Logan created all geology maps (Figs. 5a–11a) using Mapinfo and geology from OGS (2010); John Glew (deceased) drew the block models (Figs. 5b–11b). Donna Ferguson completed final figure drafting. The work was completed under the Geological Survey of Canada Groundwater Geoscience Programme, Archetypal Aquifer Project. NRCAN contribution 20210713.

Data are as presented.