Geophysical , geological , and hydrogeological characterization of a tributary buried bedrock valley in southern Ontario 1

Buried bedrock valleys infilled with Quaternary-aged sediment have the potential to become productive aquifers owing to prevalent sand and gravel deposits often associated with these topographic lows. In areas where groundwater is drawn from the underlying bedrock aquifer, buried bedrock channels may significantly affect the spatial distribution of recharge and localized contaminant pathways. Therefore, understanding the form, distribution, and the nature of Quaternary infill sediments within these buried bedrock river valleys, and their relationship to hydraulically transmissive bedrock features is an important aspect of groundwater resource management. Here, we evaluate the effectiveness of electrical resistivity and seismic refraction collected over a partially urbanized 150 ha area with variable vegetation, roads, and structures, to map the spatial distribution of sediments and delineation of a channel segment associated with a regional bedrock valley. Electrical resistivity and seismic refraction was performed along 13 (covering 11.6 km) and seven transects (covering 0.9 km), respectively, to map and characterize the bedrock surface morphology beneath a variable thickness of unconsolidated deposits. Three continuously cored holes and downhole geophysical logs, supplemented with four nearby water well records captured the in-channel as well as adjacent Quaternary stratigraphy ( 15–40 m). Cores recorded multiple glacial till deposits and ice-marginal processes associated with ice advances and retreats. Hydraulic transmissivity of the bedrock around the valley feature was evaluated using a FLUTe hydraulic transmissivity profiling technique. This study demonstrates the potential of combining several surface geophysical methods with sedimentological analysis of continuous cores and hydraulic data for characterizing tributary bedrock channel morphology and Quaternary infill at a scale relevant to localized studies of municipal production well recharge zones and contaminant transport and fate. Résumé : Les vallées ensevelies taillées dans le roc remplies de sédiments d’âge quaternaire ont le potentiel de devenir des aquifères productifs en raison des importants dépôts de sable et de gravier souvent associés à ces dépressions topographiques. Dans les régions où l’eau souterraine est tirée de l’aquifère rocheux sous-jacent, les chenaux ensevelis taillés dans le roc pourraient avoir une importante incidence sur la répartition spatiale de la recharge et des voies de propagation des contaminants. La compréhension de la forme, de la répartition et de la nature des sédiments quaternaires qui remplissent ces vallées fluviales ensevelies et leur lien avec des caractéristiques du substrat rocheux transmissif est un aspect important de la gestion des ressources d’eau souterraine. Nous évaluons l’efficacité des méthodes de résistivité électrique et de sismique réfraction utilisées dans une zone partiellement urbanisée de 150 ha présentant une répartition variable de la végétation, des routes et des ouvrages, pour cartographier la répartition spatiale des sédiments et les limites d’un tronçon du chenal associé à une vallée taillée dans le roc d’ampleur régionale. Des levés de résistivité électrique et de sismique réfraction ont été réalisés le long de 13 traverses (couvrant 11,6 km) et sept traverses (couvrant 0,9 km), respectivement, pour cartographier et caractériser la morphologie de surface du substrat rocheux sous des dépôts meubles d’épaisseur variable. Trois forages carottés en continu et des diagraphies de géophysique de fond, en plus des rapports de quatre puits d’eau situés à proximité, rendent compte de la stratigraphie quaternaire du chenal et des zones attenantes ( 15–40 m). Les carottes contiennent différents dépôts de till glaciaire et témoignent de processus proglaciaires associés avec des avancées et retraits de la glace. La transmissivité du roc autour de la vallée a été évaluée par la méthode FLUTe de détermination des profils de transmissivité. L’étude démontre le potentiel qu’offre la combinaison de plusieurs méthodes géophysique à l’analyse sédimentologique de carottes continues et à des données hydrauliques pour caractériser la morphologie de chenaux tributaires taillés dans le roc et leurs matériaux de remplissage quaternaires à une échelle pertinente pour des études localisées de zones de recharge de puits de production municipaux et du transport et du devenir des contaminants. [Traduit par la Rédaction]


Introduction
Buried bedrock valleys (BBVs) represent an important component of the hydrogeologic conceptual model of southern Ontario and other previously glaciated terrains such as the Canadian Prairies, northern United States, and Europe (e.g., Kehew and Boettger 1986;Sandersen and Jørgensen 2003;Russell et al. 2004;Cummings et al. 2012;Sharpe et al. 2013).In southern Ontario, these complex and variably distributed features comprise an extensive network of bedrock valleys that formed through, or were enhanced by, intense glacial and glaciofluvial erosion during the Pleistocene (Karrow et al. 1979;Eyles et al. 1997;Kor and Cowell 1998;Gao 2011), and were subsequently infilled with Quaternary-age glacial and interglacial sediment (Karrow 1968;Barnett 1992).Although province-wide mapping (Gao et al. 2006) and regional mapping of some specific areas (e.g., Burt 2011;Marich et al. 2011;Bajc et al. 2012) have been carried out, the location and geometry of these BBVs and their tributaries is not well constrained in many parts of Ontario.
Source water protection has become an integral component of a community's groundwater management plan since the introduction of the Clean Water Act (2006), S.O. 2006, c. 22 (Ministry of the Environment and Climate Change 2006).This has led to significant investments in three-dimensional (3D) sediment and bedrock surface mapping across Ontario both by the geological surveys (e.g., Gao et al. 2006;Sharpe and Russell 2006;Bajc 2008;Davies et al. 2008;Bajc and Rainsford 2010;Bajc et al. 2012Bajc et al. , 2014;;Burt 2012) and municipalities and conservation authorities that were tasked with creating source water protection plans (e.g., Gartner Lee Ltd. 2004;Golder Associates Ltd. 2006;AquaResource Inc. 2010).These activities have supported the conceptualization of aquifer and aquitard units and the delineation of regional-scale groundwater flow systems within and around BBVs.
In some cases, here in Ontario and elsewhere, thick successions of sand and gravel deposits have infilled these valleys, resulting in highly productive aquifers (Kehew and Boettger 1986;Russell et al. 2004;Wiederhold 2009).Alternatively, buried valleys can be infilled or covered by low-permeable Quaternary-age diamicton or a combination of sediment types with complex geometries and very different hydraulic properties (Jørgensen et al. 2003a;Meyer and Eyles 2007;Cummings et al. 2012), further complicating hydrogeologic conceptualizations (Cummings et al. 2012;van der Kamp and Maathuis 2012).Therefore, these environments require more detailed characterization of both form and infill to ensure accurate quantitative analysis of local groundwater flow system geometries.
The underlying sedimentary bedrock across southern Ontario contains ubiquitous interconnected fractures with variable lengths and apertures (Johnson et al. 1992;Kennel 2008;Munn 2012).In addition to these fractures, the Silurian dolostone aquifera prolific aquifer for private and municipal water across southern Ontariocontains dissolution-enhanced features (e.g., fractures or vugs) and conduits that increase its groundwater resource potential, but also its susceptibility to surface contamination (e.g., Perrin et al. 2011).Many communities across southern Ontario rely almost exclusively on these fractured and dissolution-enhanced bedrock aquifers for their drinking water supply.The complex interdependent relationship between Quaternary deposits and these fractured sedimentary rocks has become an important component in the development of effective groundwater management strategies for these communities (e.g., Gartner Lee Ltd. 2004;AquaResource Inc. 2010).Given the potential association between BBVs and increased recharge and transmissivity of bedrock aquifers (Cole et al. 2009), successful groundwater management strategies will depend on sufficiently high-resolution (spatial and temporal) quantitative information, enabling accurate and well-constrained geologic and hydrogeologic conceptualizations of buried valley features (e.g., Shaver and Pusc 1992;Smerdon et al. 2005).
Geophysical methods have the capacity to unravel complex geologic relationships and can play a critical role in advancing our understanding of shallow groundwater flow systems in these settings (van Dam 2012).Studies utilizing multiple geophysical methods combined with geologic and hydrogeologic data are particularly effective for 3D mapping of hydrostratigraphic units (e.g., Gabriel et al. 2003;Jørgensen et al. 2003aJørgensen et al. , 2003b;;Steuer et al. 2009; Stumpf and Ismail 2013;Cassidy et al. 2014;Sapia et al. 2014).In Ontario, shallow seismic reflection, gravity, and electromag-netic surveys have been used at mostly regional scales to delineate the nature of the infilled sediment and bedrock valley geometry (e.g., Greenhouse and Monier-Williams 1986;Greenhouse and Karrow 1994;Zweirs et al. 2008;Burt 2011;Bajc et al. 2012;Pugin et al. 2013).Elsewhere, airborne transient electromagnetic methods are mostly used (e.g., Sandersen and Jørgensen 2003;Jørgensen et al. 2003aJørgensen et al. , 2003b;;Sørensen and Auken 2004;Sapia et al. 2015).
This paper focuses on a local-scale high-resolution electrical resistivity study conducted within the City of Guelph, Canada, with a primary goal of demonstrating the capability and limitations of the technique for mapping a BBV at a relatively local scale.Surface electrical resistivity and seismic refraction results are combined with sedimentological analysis of cores, downhole geophysical logs, and borehole hydraulic information to develop a data-driven model of bedrock channel morphology and Quaternary infill, and associated hydraulic characteristics at a scale relevant to localized studies of municipal production well recharge zones and flow and transport characteristics.

Regional geologic and hydrogeologic setting
The Guelph region is underlain by the Middle Silurian unsubdivided Amabel (i.e., locally identified as the Irondequoit, Gasport, and lower Goat Island formations), Eramosa, and Guelph Formation dolostone (Brunton 2009;Armstrong and Carter 2010;Fig. 1).These massive and fossiliferous units represent an extensive shallow water carbonate platform that formed over the Algonquin Arch, a tectonic high situated between the subsiding Michigan and Appalachian basins (Brett et al. 1990;Johnson et al. 1992).Sedimen-  Armstrong and Carter (2010) and with proposed revisions by Brunton (2009).*Cabot Head shale of the Cataract Group.tary facies across these dolostone units are considered varied and complex owing to a diversity of paleoenvironments that ranged from high-energy reefal to quiet water back-reef lagoons (Brunton 2009).Widespread till plains and a series of moraines have been deposited over the bedrock during the Pleistocene through multiple ice advances and subsequent retreats (Barnett 1992).This glacial activity resulted in highly variable drift thickness, with bedrock valleys infilled with stacked successions of coarse to finegrained sediment of varying characteristics and geometries (e.g., Meyer and Eyles 2007;Burt 2011), preserving tills that represent several ice advances during the last Wisconsinan (Fig. 2).
The bedrock topography across southern Ontario has been extensively sculpted by preglacial, glacial, and glaciofluvial erosion (Fig. 3) as evidenced by uneven and complex bedrock valleys and troughs (Eyles et al. 1997;Gao 2011).In the Guelph region, the bedrock surface exhibits several valleys trending in a southwestnortheast direction (Fig. 4) (Karrow et al. 1979;Cole et al. 2009), with drift thicknesses ranging between 0 and 62 m.Based on regional mapping of the surficial geology, the study site primarily consists of glaciofluvial outwash sand and gravel, poorly sorted sandy silt to silty sand, and drumlinized till plain deposits (Karrow 1968(Karrow , 1987)).
Using geological and hydraulic data sets from consultant reports completed for the City of Guelph, Gautrey (2004) noted a spatial relationship between the location of high-capacity municipal production wells and BBVs within the City of Guelph; based on these data and two complete bedrock cores, Cole et al. (2009) hypothesized that bedrock near buried valleys in the Guelph region may possess greater abundance of dissolution-enhanced fractures and conduits that would enhance hydraulic connectivity with deeply buried aquifers and between fractured bedrock and Quaternary sediment.However, the type and distribution of Quaternary sediment covering these BBVs as well as the valley's geometry and orientation with respect to groundwater flow direction will likely affect its significance to the overall flow system.Perrin et al. (2011) examined the role of dissolution-enhanced flow features on the transport and fate of contaminants within the same dolostone aquifer covered by up to 50 m of Quaternary deposits; their study found that deposition of Quaternary sediments altered the boundary conditions governing the conduit flow system A more recent study within the Guelph area showed the impact of varying glacial sediment distribution on the migration and fate of agricultural contaminants to the underlying fractured dolostone aquifer (Best et al. 2015).While the exact relationship between BBVs and increased well productivity has yet to be fully understood, it stands to reason that the potential yield and quality of water withdrawn from a bedrock aquifer near a Quaternary infilled channel will be influenced by the composition and architecture of the unconsolidated sediments and the morphology of the bedrock surface; hence, there is value to improving our understanding of these conditions.

Study site description
The study site is located within the City of Guelph and includes a large portion of the University of Guelph's Arboretum and an adjacent golf course to the north-northwest (Fig. 4); various municipal roads and private drives cross the site.The area largely consists of open grasslands, areas of dense brush regrowth, manicured fairways, and is bounded by the Eramosa River to the northeast.Surface topography is relatively flat toward the southeast and slowly decreases in elevation toward the Eramosa River where it also begins to exhibit increasing variability.
Data from a number of existing water supply and irrigation wells are available in the study area (Table 1).Well records indicated that the bedrock elevation is highly variable across the study site.Available well records and regional bedrock elevation data compiled by the Ontario Geologic Survey (Fig. 3) indicated the presence of a local-scale BBV, which likely connected to the regional Guelph-Rockwood BBV (Fig. 4; Karrow et al. 1979;Greenhouse and Karrow 1994;Eyles et al. 1997;Gao et al. 2006).Based on these point data, the feature appears to be approximately 30 m deep, hundreds of metres wide, striking east-west with an approximate valley floor elevation of 290 m above sea level (masl).

Geophysical investigations of buried valleys
Buried bedrock valleys and their Quaternary fills can be studied using a wide range of surface geophysical techniques (Greenhouse and Karrow 1994).However, the majority of studies to date have focused on the use of gravity (e.g., Greenhouse and Monier-Williams 1986;Gabriel et al. 2003;Gabriel 2006;Møller et al. 2007;Zweirs et al. 2008;Bajc and Rainsford 2010;Burt and Rainsford 2010), which exploits changes in subsurface bulk density and ground and airborne transient electromagnetic methods (e.g., Jørgensen et al. 2003aJørgensen et al. , 2003bJørgensen et al. , 2013;;Sørensen and Auken 2004;Steuer et al. 2009;Høyer et al. 2015;Oldenborger et al. 2016).While gravity methods can be highly effective in mapping changes in bedrock elevation, they provide limited insight about Quaternary infill and architecture, and are frequently accompanied by significant uncertainty in the true bedrock elevation in the absence of borehole logs.Although transient electromagnetic methods are relatively effective in the characterization of valley infills and bedrock morphology, these methods have large sampling volumes, making them more suited to deeper and larger-scale investigations; these methods are also very susceptible to cultural electromagnetic interference, which may limit applications in urbanized areas.Advancements in seismic reflection technology (i.e., data processing equipment and Fig. 2. Regional stratigraphy in the Guelph area based on mapping by Karrow (1987) and stratigraphy of the nearby Rockwood buried bedrock valley based on borehole logging and regional mapping by Burt and Dodge (2016); refer to Fig. 3 for location of Rockwood buried bedrock valley.
Pagination not final (cite DOI) / Pagination provisoire (citer le DOI) software) have led to more frequent applications in shallow (30-100 m) environments (Steeples 2000).Numerous studies demonstrate the capacity of shallow seismic reflection to characterize the rock surface and Quaternary fill architecture (e.g., Boyce et al. 1995;Büker et al. 1998;Gabriel et al. 2003;Jørgensen et al. 2003b;Ahmad et al. 2009;Ogunsuyi and Schmitt 2010;Stumpf and Ismail 2013;Burt 2015).Although shallow seismic reflection has been recognized as a powerful tool for investigating shallow bedrock and Quaternary deposit geometry (e.g., Hunter et al. 1984;Pugin et al. 2014) to support or enhanced hydrogeologic characterization (e.g., Sharpe et al. 2003;Ogunsuyi and Schmitt 2010), data acquisition remains costly, can have poor resolution, and in many cases, data processing and interpretation can be challenging (Steeples et al. 1997;Steeples and Miller 1998;Ismail et al. 2012), with varying quality in the results.These data are susceptible to cultural noise, making data acquisition in urban areas challenging.
Surface electrical resistivity methods are an alternative to gravity, transient electromagnetic, and seismic reflection surveys for imaging shallow (<100 m) geologic formations, especially in areas impacted by infrastructure, cultural noise, and electromagnetic interference.Electrical methods also have considerably lower operational costs, in some cases are easier to implement, and provide measurements that, in practice, are more straightforward to interpret.Like all geophysical methods, electrical methods have limitations; unlike seismic reflection, electrical resistivity measurements do not provide a direct image of the subsurface, tend to combine layers, and have relatively poor resolution of interfaces with increasing depth of investigation or near the model boundaries.
Most modern apparent resistivity measurements are conducted with multicore cables and interpretations that rely on a tomographic inversion process where measured data are reconstructed from forward models of a specific physical property distribution (Snieder and Trampert 1999;Loke et al. 2013).Although data inversion is a standard practice in the interpretation of most geophysical data, the solutions of inverse problems in applied geophysics are not inherently unique.In other words, the model that best matches the measured data is not necessarily an exact representation of the subsurface parameter distribution.Nevertheless, electrical imaging techniques using either induced or direct current sources have provided promising results for imaging valley morphology (e.g., Jørgensen et al. 2003a;Ahmad et al. 2009;Steuer et al. 2009;Tye et al. 2011), and in some cases, the infilled Quaternary sediment architecture (e.g., Baines et al. 2002;Leopold et al. 2013), particularly when combining multiple geophysical techniques with geological information (e.g., Greenhouse and Karrow 1994;Gabriel et al. 2003;Jørgensen et al. 2003b;Ahmad et al. 2009).

Borehole lithology, geophysical logs, and hydraulic profiling
Continuous cores and downhole geophysics were collected at GDC-1, GDC-2B, GDC-6 (adjacent bedrock cored hole to GDC-10A), and GDC-10A (Fig. 5), recording geologic characteristics of bedrock and Quaternary sediments across the bedrock valley.Sedimentary characteristics of the unconsolidated material overlying bedrock were logged in the field (centimetre scale) using standard sedimen-Fig.3. Regional bedrock elevation showing major buried bedrock valley features across southern Ontario (Gao et al. 2006).The study area within the City of Guelph examines a branch of the Guelph buried valley, which is an extension of the larger Rockwood valley to the northeast toward Toronto.Coordinate system: North American Datum 1983 (NAD83) Universal Transverse Mercator (UTM) zone 17N.[Colour online.]tological techniques to record changes in texture, sorting, sedimentary structures, and clast characteristics with depth based on PQ mud rotary (GDC-1, GDC-2B) and rotosonic (GDC-10A) continuous core records.Quaternary cores were 7.6 cm (GDC-10A) to 12.7 cm (GDC-1 and GDC-2B) in diameter and ranged in length from 10.4 m (GDC-10A) and 16.5 m (GDC-2B) to 43.4 m (GDC-1).Overall, core recovery was excellent in both GDC-10A (97%) and GDC-1 (93%), and adequate in GDC-2B (80%).Disturbance in the core is minimal, and sedimentary bedding and structures well preserved once mud is scraped off the cores recovered by PQ mud rotary drilling, whereas the primary sedimentary structures tend to be homogenized in the cores recovered by rotosonic drilling.Textural classi-fication of diamict (a poorly sorted deposit consisting of mud, sand, and gravel that is commonly found in glaciated settings) is based on Hambrey and Glasser's (2003) classification where diamict is either clast poor or clast rich (1%-5% or 5%-50% gravel content, respectively) and either sandy (>66% sand), intermediate (33%-66% sand), or muddy (<33% sand).
Table 1.Lithologic descriptions from on-site water supply wells.) or broken rock fragments (GDC-1).Note that the Amabel Formation is shown here with local informal subdivisions of Irondequoit, Gasport, and Goat Island formations as proposed by Brunton (2009).b, unsealed bedrock interval; cl, clay; cps, counts per second; g, gravel; mbgs, metres below ground surface; s, sand; si, silt; Vf, very fine.
logged in a similar fashion (10 cm scale), noting lithological properties such as rock type, colour, hardness, crystallinity, fossil abundance, fossil type, dissolution-vug intensity, characteristics of fractures, and presence of styolites (Fomenko 2015).Borehole geophysical logs included natural gamma, induction conductivity (expressed here as the reciprocal apparent resistivity), and primary (P)-wave velocities within the bedrock.Steel casings were installed to bedrock with the exception of GDC-6, which was designed with a polyvinyl chloride casing.
Open boreholes at GDC-1, GDC-2B, and GDC-6 were hydraulically tested using a flexible downhole liner method (Keller et al. 2013), which provides a measure of formation transmissivity (e.g., Quinn et al. 2015).With this method, flexible impermeable liners are everted into an open borehole while measuring the rate of descent at a constant hydraulic head and associated forces (Keller et al. 2013).As the everting liner passes and seals transmissive features or sections of the borehole, changes in the liner velocity indicate the position of a permeable feature and provide estimates of total open borehole and depth-discrete transmissivities using the Thiem equation for steady radial flow (Keller et al. 2013).

Electrical resistivity tomography
The electrical resistivity tomography (ERT) method measures the apparent resistivity of the subsurface by applying a direct current electrical current between two electrodes inserted into the ground and measuring the potential difference across two other electrodes at various spacings and separation distances from the current source.These data were used to construct twodimensional (2D) apparent resistivity pseudo-sections of the earth that were inverted to obtain a reasonable image of the subsurface materials (i.e., electrical property distribution).A more theoretical discussion of geophysical inversion and the surface resistivity method can be found in Oldenburg and Li (2005) and Loke et al. (2013), respectively.
A fully automated Syscal Jr. Switch resistivity system (Iris Instruments, Orleans, France) equipped with 5 and 10 m electrode spacings connected to stainless-steel electrodes inserted into the ground were used to collect measurements.The system has the capacity to handle 48 electrodes for a total array length of 235 and 470 m for 5 and 10 m electrode spacing configurations, respectively.A maximum investigation depth of 70 m was achieved using the 10 m electrode spacing with a Wenner-Schlumberger electrode configuration; the Wenner-Schlumberger configuration was used for all resistivity surveys because of its optimal lateral and vertical sensitivity and higher signal-to-noise ratio relative to other arrays (Dahlin and Zhou 2004).Full 2D profiles were constructed by progressively rolling along the profile in 115 m (for 5 m electrode spacing) and 110 m (for 10 m electrode spacing) increments, reaching maximum 2D survey lengths of 570-1070 m across the site (Fig. 4).A total of 13 transects were collected over the study area spanning ϳ11.6 km (Fig. 4).
Measured apparent resistivity data were manually filtered to remove bad data points and outliers prior to being inverted using RES2DINV version 3.58 (Loke 2012).A robust inversion scheme was used with moderate dampening parameters to reduce impacts of noise on the inverted model.Owing to the distal proximity of the boreholes with respect to the resistivity profile, the inversions were not constrained using any depth-to-bedrock measurements from boreholes.Still, very reasonable agreement was achieved between inverted models and nearby borehole information.Each resistivity data set was inverted using the same set of parameters.
The ERT arrays were collected along two orientations: lines 1-5 were collected in a northeastern direction, and lines 7-11 were collected in a northerly direction (Fig. 4).Given the minimum segment lengths (>570 m) needed to achieve the necessary depth of investigation and lateral coverage, as well as the added complications of paved roadways with traffic, and restricted access to the adjacent golf course, the initial resistivity measurements were collected in a southwest-to-northeast direction roughly parallel to College Avenue.Based on these preliminary results, additional resistivity surveys were conducted with a north-south orientation, with a broader lateral coverage across the study area that included the adjacent golf course.

Seismic refraction
A number of seismic refraction surveys were carried out in areas of low cultural noise to confirm the apparent depth-to-bedrock interpreted from the resistivity models (Fig. 4).Refraction surveys were not conducted over the suspected valley given the large shot offset distances and strong energy source needed to reach the bedrock surface.Seismic surveys were conducted using 24 and 48 channel Geode seismographs (Geometrics, USA) connected to a laptop computer.A 10 kg sledge hammer and steel plate were used to generate a P-wave source.A variable geophone array was used for all surveys, which included 2 m geophone spacings on the ends and center of the array (48 channel arrays only) and wider 5 m spacings elsewhere.A variable spacing array was used to better detect the direct ground wave and the cross-over position of the first refraction; standard forward and reverse shot procedures were used on-end and off-end (Milsom 2003), as well as at various points along the array (e.g., middle of array and opposite ends of 2 m spaced geophone segments).All data were recorded using a sample interval of 0.0625 ms with a 0.25 s window; 100 Hz P-wave geophones were used for the surveys.A total of seven transects were collected over the study area, with a total line length of ϳ0.9 km (Fig. 4).
Seismic data were processed using Reflex-Win version 7.0 (Sandmeier 2012) seismic software, which included basic signal processing: low-pass temporal filtering, automatic gain control, bandpass frequency filter, and time cut.First breaks corresponding to the arrival of the direct ground and refraction events were manually identified at each geophone location.Refraction arrival times were analyzed using Hagedoorn's (1959) plus-minus "delaytime" method to provide a depth to bedrock estimate across the survey area.These data were also interpreted using a more advanced refraction tomography technique (i.e., simultaneous iterative reconstruction technique) available within Reflex-Win software.The seismic tomographic method is similar to the resistivity tomography in that it optimizes a 2D velocity model to the measured arrival times.As a result, this method can be more effective at incorporating spatial heterogeneities and gradational boundaries relative to conventional interpretation methods (Sheehan et al. 2005).Seismic inversion parameters were optimized for model stability and convergence.

Geology, geophysical logging, and hydraulic transmissivity profiles
Detailed Quaternary logging on either side of the bedrock valley (GDC-2B, GDC-10A), and from within the valley (GDC-1), reveal relatively variable stratigraphy (Fig. 5).GDC-2B and GDC-1 are both located in an area that is mapped as drumlinized till plain and are dominated by thick packages of diamict, whereas GDC-10A is located in an area mapped as glaciofluvial outwash plain and is dominated by variable stratified sediments.Bedrock was encountered at 318.4, 290.2, and 316.9 masl in GDC-2B, GDC-1, and GDC-6, respectively.Logging of the bedrock cores in GDC-1, GDC-2B, and GDC-6 identified the Eramosa (upper and lower), Goat Island, Gasport, Irondequoit, Rockway, Merritton, and Cabot Head formations.The Gasport formation is a regional semi-confined aquifer, which is locally overlain by the regionally discontinuous Eramosa Formation, whereas the overlying Guelph Formation is an unconfined bedrock aquifer, and the basal Cabot Head shale is a regional aquitard (Brunton 2009).
Quaternary sediments in GDC-2B (Fig. 5), on the southeast edge of the valley, are dominated by multiple layers of diamict (2-16.4m below ground surface (bgs)).Diamict beds (tens of centimetres to ϳ3 m in thickness) are distinguished by colour, degree of consolidation, matrix texture (sandy-intermediate and muddy), clast size, roundness (subangular to subrounded), and amount (clast rich to clast poor) as well as minor sand interbeds.This thick diamict package can be divided into three main units: a basal overconsolidated clast-rich intermediate diamict at the sediment bedrock interface, overlain by an overconsolidated clast-poor muddy diamict (ϳ11.5-15m bgs) with bullet-shaped clasts, and a clast-rich intermediate diamict (ϳ2-11 m bgs).The diamict package is capped by a thin bed of gravel, a thick bed of gravelly (25%) very fine sand, and the current soil profile.The basal diamict and the overlying mud-rich diamict likely record deposition in a subglacial environment based on their overconsolidated nature, presence of bullet-shaped clasts, and stratigraphic position (Karrow 1968(Karrow , 1987;;Evans et al. 2006).In contrast, the layering, sand interbeds and lack of consolidation in the coarser grained clast-rich upper diamict unit suggest that this unit records deposition by melt out or mass flow (Boulton 1972;Evans et al. 2006).Based on its stratigraphic and topographic position relative to modern rivers, the uppermost pebbly sand and gravel likely record glaciofluvial processes during deglaciation (Karrow 1968(Karrow , 1987)).
The valley infill (GDC-1, Fig. 5) consists of just over 6 m of cobble gravel composed of broken angular black shale fragments at the base with trace sand, overlain by a thick package (ϳ28 m) of very stiff clast-poor muddy diamict, with a silty clay and some mediumfine sand matrix.Crude layering within this thick package of diamict is observed based on changes in colour (dark grey to olive grey to trace red), plasticity, or size and composition of gravelsized clasts (ranging from granule to cobble and dolostone to shale, respectively).This thick package of diamict is in turn overlain by silty clay (2.3 m) and a clast-poor, stiff diamict with a sandy clay matrix (2.2 m).The core is capped by a 4.2 m bed of gravel.This sequence is interpreted to record accumulation of bedrock rubble over competent bedrock, deposition of till in a subglacial environment, an interval of clay settling in a ponded environment, another till deposited in a subglacial environment, and capped by gravel that most likely records glaciofluvial conditions in an ice-marginal environment.
GDC-10A, located on the northwest edge of the valley (Fig. 5), is dominated by stratified sediments of variable texture.A thin (60 cm) basal clast-rich sandy diamict occurs at the bedrock sediment interface.It is overlain by 3 m of crudely stratified pebbly medium sand with some cobbles; the stratification is defined by changes in clast size, content, and cohesiveness.The next 1.5 m includes beds of clast-rich intermediate diamict, pebble gravel, and pebbly fine sand.The top 5 m is dominated by sandy silt and clay with minor interbeds of pebbly sand and pebble gravel.Based on its stratigraphic position, the glacial history of this area (Karrow 1968(Karrow , 1987)), and the variable types of sediments ranging from gravel to clay, this succession is interpreted to record a dynamic environment with variable depositional and erosional conditions ranging from higher energy glaciofluvial flows to quieter depositional conditions where silts and clays were allowed to settle in a ponded setting.The depositional origin of the diamict is unclear; the lack of overconsolidation suggests these units record debris flows in an ice-proximal environment.
In terms of regional till stratigraphy, the following tentative correlations are proposed.The basal diamict in GDC-2B (ϳ15-16.5 m bgs) is very distinct from the other tills in the area and is similar to that observed by Burt (2011;borehole BH40-OF_2010) and interpreted to be a pre-Catfish Creek Till.The overlying dense mud-rich grey to olive diamict (ϳ11-15 m bgs) and the corresponding thick package of dense grey diamict in GDC-1 (ϳ9-37 m) is likely Catfish Creek Till (Late Wisconsinan, Nissouri phase), based on texture, colour, and its overconsolidation.The depositional origin of the unconsolidated diamict in the upper part of GDC-1 and GDC-2B is difficult to establish.Based on its stratigraphic position and clay-rich nature, the uppermost diamict in GDC-1 is tentatively correlated with the regionally mapped Maryhill Till (Karrow 1987).Based on stratigraphic position, the absence of overconsolidation, and the coarse-grained nature of diamict in GDC-2B, it likely records sediment reworking and debris flows during the retreat of Port Stanley ice (Late Wisconsinan; Port Bruce phase; Karrow 1968Karrow , 1987)).
The natural gamma logs exhibit a moderately variable response across the upper 10-15 m of Quaternary sediment at GDC-1, GDC-2B, and GDC-6 (27.4 m away from GDC-10A).The gamma response at GDC-2B indicates sharply bounded Quaternary sediment subdivisions, including two clay-rich layers between 0-2 m bgs (332.9-334.9masl) and 11-14 m bgs (320.9-323.9masl).In contrast, the response within the Quaternary sediment portion of GDC-6 indicates gradational contacts with more subtle but frequent subdivisions; however, the grout used behind the polyvinyl chloride casing at this location may be contributing to the more erratic response across the unconsolidated sediments.The response observed at GDC-1 within the upper 10 m shows sharper boundaries before transitioning into a relatively high and uniform response between 15 and 32.5 m bgs (301.2-318.7 masl).The more variable gamma signal at the top of the succession corresponds well with the gravel at surface and the clay-rich interval at ϳ7 m, whereas the more consistent signal between 15 and 32.5 m depth corresponds to the thick package of grey diamict that makes up the majority of the valley infill.A sharp change in gamma count was encountered at a depth of 38 m (295.7 masl), which is consistent with the unit of broken black shale bedrock fragments encountered below the diamict and above the more competent dolostone at the base of GDC-1 (Fig. 5).Apparent resistivity and sonic logs were not collected across the Quaternary owing to the presence of a steel casing at GDC-1 and GDC-2B and a low water table at GDC-6 (sonic only).
The variable natural gamma responses observed within the bedrock between 288 and 307 masl in GDC-2B, and between 292 and 309 masl in GDC-6, indicate increased stratigraphic subdivision; here, the high gamma combined with low resistivity support the presence of a clay-rich unit consistent with the regionally discontinuous Eramosa Formation (Brunton 2009;Cole et al. 2009), which regionally separates the underlying unsubdivided Amabel (Goat Island and Gasport formations as per Brunton 2009) from the overlying Guelph Formation (absent in this area) (Brunton 2009).
The sonic logs show variable changes in P-wave velocity (generally ranging between 2000 and 7000 m/s) superimposed over more gradual trends.Here, the variable fluctuations reflect changes in mechanical properties (e.g., fracture frequency), while the gradual trends demarcate the lithostratigraphic sequences (e.g., Eramosa, Goat Island, and Gasport formations).P-wave velocities of the shallow bedrock were approximately 4500 m/s.
FLUTe hydraulic transmissivity profiling, as described by Keller et al. (2013), performed in GDC-2B, GDC-1, and GDC-6, was used to compare hydraulic conditions at each of these locations with respect to the BBV.These data illustrate a decrease in the hydraulic transmissivity of the borehole below the bottom of the descending liner, which is expected as the liner seals flow features and reduces remaining open-hole transmissivity as it descends.The transmissivity T (and bulk hydraulic conductivity, K b ) of the bedrock between the base of the valley (below the bottom of the casing at GDC-1 or 288 masl) and the top of the Cabot Head shale was highest in GDC-2B, and lowest in GDC-1 (Fig. 5).The bedrock transmissivity south of the valley at GDC-2B was relatively high at the base of the BBV (T = 1 × 10 −3 m 2 /s; K b = 3 × 10 −5 m/s), with few changes in transmissivity, i.e., few inflections in the profile and nearly all the reduction in T occurred at once near the top of the Cabot Head shale.The profile at GDC-1 within the valley was comparatively low (T = 1 × 10 −5 m 2 /s; K b = 3 × 10 −7 m/s) and exhibited more noticeable declines in transmissivity throughout the bottom of the Gasport compared with GDC-2B before T diminished to a nonmeasurable value.Whereas the transmissivity at GDC-6 (north of the buried valley) exhibited an intermediate value (T = 3 × 10 −4 m 2 /s; K b = 1 × 10 −5 m/s), the characteristics of the profile were distinct; for instance, a significant drop in T was observed in the middle of the Gasport at 273 masl, approximately 18 m above the bottom of the hole and well above the depth of the other two holes.This suggests that the valley bottom did not receive much groundwater discharge from below but rather from the flanks of the valley, with the bedrock on either side of the valley showing more transmissivity.

Surface electrical resistivity profiles
The Quaternary-bedrock contact was defined across the study area using inverted ERT models.This approach is illustrated with the results from line 10 (ERT-10A-ERT-10B) (Fig. 6).The interpreted position of bedrock surface (dotted and dashed lines) shown on the interpreted geo-electrical model was set according to the largest upward electrical gradient in the inverted model section (i.e., approximately where the resistivity exceeded 600 ⍀•m).Here, the upper zone is interpreted as less resistive Quaternary deposits, whereas the lower zone represents the more resistive bedrock.It is important to note that because the ERT model defines the spatial extent of "electrically similar" material, it is possible for unconsolidated materials to exhibit similar electrical properties to bedrock and vice versa (Reynolds 2011).Further, the inversion process can result in resistivity artifacts, particularly around the sides and bottom of the section that limits interpretability in those regions.
Line 10 was collected using a 10 m electrode spacing and orientated such that it was orthogonal to the suspected valley structure noted in lines 1-5 (Figs.7a-7e).These electrical data indicate a wide bedrock depression or valley structure between 480 and 780 m, with a shallower valley wall on the north slope; the interpreted bedrock surface is consistent with observations at GDC-1, GDC-2B, and GDC-6.However, it is apparent that the interpreted bedrock elevation of the valley floor is slightly elevated relative to GDC-1.This is likely the result of a layer of bedrock rubble logged along the valley floor, which is electrically more similar to the underlying competent bedrock than the overlying diamict.In addition, inverted apparent resistivity values obtained between 0 to 250 m along line 10 within the interpreted bedrock zone are lower than expected based on adjacent areas.These low values could be interpreted as a decrease in bedrock elevation and (or) another valley feature; however, available borehole data from Old #8 (Table 1), approximately 138 m eastward intercepts a shallow bedrock surface in that area.
A closer inspection of the measured apparent resistivity data reveals a resistivity decrease between 120 and 160 m, which extends down through the pseudo-section (upper image in Fig. 6).This response coincides with the position of a perennial storm water pond (refer to Fig. 4), which collects runoff from a parking Fig. 6.ERT results from line 10 using a 10 m electrode spacing show measured apparent resistivity pseudo-section (top), inverted resistivity model (middle), and interpreted geo-electrical model (bottom).Long dotted line on interpreted model (250-950 m position) depicts largest resistivity gradient (i.e., interpreted as Quaternary-bedrock boundary), while the short dotted line (0-250 m position) represents an interpreted boundary based on geo-electrical trends and available borehole logs.Given the lateral offset of Old #8, it is possible that the resistivity response from 0 to 250 m identifies the edge of a valley feature that extends southwestward from lines 8 and 9. Core logs are superimposed to show measured unconsolidated sediment thickness (white) and depth of bedrock (black); orthogonal offsets of core logs from resistivity transect are shown as "x" for into and "o" for out of the 2D resistivity plane.Fig. 7. Summary of 10 m ERT resistivity images with interpreted single layer over half-space models.Nearby logs with measured depth to rock are shown with corresponding orthogonal offsets represented as "x" for into and "o" for out of the 2D plane.Results include (a-e) lines 1-5 orientated in a southwest-to-northeast direction and (f-l) lines 7-11, 14, and 15 orientated in a south-to-north direction.bm, bedrock mound.
Pagination not final (cite DOI) / Pagination provisoire (citer le DOI) 10 Can.J. Earth Sci.Vol.00, 0000 Published by NRC Research Press lot and nearby roadway ditches during spring snowmelt.This pond might represent an area of focused surface water infiltration with possibly higher total dissolved solids (e.g., road salt), which might explain the relatively lower bedrock resistivity between 0 and 250 m (middle image in Fig. 6).Given the large lateral offset of Old #8 east of the transect, it is also possible that the resistivity response along the south end of line 10 is indicative of a bedrock valley, corresponding to the southern edge of the valley identified at the southern extent of lines 8 and 9; however, data coverage is limited in this area.
The bedrock surface can generally be identified across the study area using the 600 ⍀•m threshold in the 10 m spaced resistivity profiles (Fig. 7).Occasionally, the bedrock surface was interpolated in areas of low model sensitivity near the boundaries of the model (e.g., Figs. 7b,7f,7g,7h) and (or) where there was slight disagreement between the delineation of the resistivity threshold and a nearby borehole (e.g., Figs. 7d,7i,7k,7l).In those instances, available borehole information and careful interpretation of subtle electrical trends in the apparent resistivity data sets were used to define the bedrock surface.Lower-resistivity trends, marked by the green-coloured zones in the inverted model, appear to delineate a valley floor consistent with expected bedrock elevation; here, interpretation of the bedrock surface was more tentative given the low electrical contrast likely attributed, in part, to a change in the Quaternary sediment (i.e., transition from predominantly conductive infill to relatively resistive infill or bedrock fragments).
The interpreted bedrock surface reveals numerous bedrock mound (bm) features (Fig. 7).These smaller-scale (tens of metres) features suggest variable bedrock relief outside the valley.To better image these features, a series of higher resolution resistivity surveys were conducted using 5 m electrode spacings (Fig. 8).Resulting models also improved characterization of Quaternary deposits outside and within portions of the valley structure, and led to a more accurate depiction of the bedrock mounds.Highly variable Quaternary deposition is evident across these two profiles with sand and gravel-type units (glaciofluvial) deposited within broader clay to sandy-silt clay-rich units (till, debris flow, or glaciolacustrine deposits) within the upper part of the succession.These higher resolution data also define the geometry of the bedrock mound (bm) features (<50 m in diameter).Although reef mound facies are present in the lower half of the Guelph Formation and the upper Gasport formation (Brunton 2009), it is not known whether these apparent bedrock mounds represent exposed reef lithofacies (residing above the Eramosa), or are simply elongated glacial erosional features.Nevertheless, these mound features were identified throughout the broader study area outside the BBV (Figs. 7, 8); thus, these features are not considered to be an artifact of the inversion process.

Seismic refraction surveys
Bedrock depth estimates based on seismic refraction across the study area ranged from 9 to 20 m bgs (Table 2), and were broadly consistent with ERT-based bedrock elevations.Results from two surveys carried out on the north side of the study site are presented in Fig. 9 as a standard delay-time analysis with off-end shots used to extrapolate the bedrock refraction to zero offset (Redpath 1973), and as an inverse model (Sandmeier 2012).Reported P-wave velocities in dolostones can range from 2600 to 6500 m/s (Reynolds 2011).Based on the range of velocities and smoothed nature of the tomographic models, and measured sonic velocities within the Upper Eramosa Formation (ϳ4500 m/s), a bedrock threshold velocity of 3200 m/s was used to delineate the bedrock surface.The position of the interpreted bedrock surfaces differs significantly between the two methods, with the standard delay-time method consistently providing ϳ5 m shallower depths to bedrock (Table 2) and the tomographic inversion providing a smoothed 2D velocity model (Fig. 9).Although tomographic analyses will have a tendency to smooth sharp interfaces (e.g., the contact between unconsolidated sediments and bedrock), the benefit of this method is its ability to represent actual velocity gradients, which, if significant, could impact the top of rock calculated using the delay-time method.However, the absence of distinct boundaries in the 2D model means that a more subjective analysis of the data is needed to mark the depth and thickness of geologic layers.
Refraction data along line 1 (SR-1A-SR-1B; Fig. 9a) and line 2 (SR-2A-SR-2B; Fig. 9b) (locations shown in Fig. 4) were collected over an interpreted bedrock mound (bm) feature.The velocity models indicate a zone of relatively lower velocity (3200-3800 m/s) extending downward within the bedrock between 308 and 316.5 masl (between 30 and 50 m along line 1 and between 35 and 50 m along line 2).Its geometry is consistent with the mounded feature identified in the resistivity data (Fig. 8a) over the same area.These lower velocities suggest that the bedrock mounds are composed of less dense material compared with adjacent rock and that they Pagination not final (cite DOI) / Pagination provisoire (citer le DOI) may be exposed reef mounds; however, further study is required to better constrain their composition and origin.In summary, based on the corroborating evidence provided by the seismic refraction survey and geological borehole data, the resistivity method has effectively identified the bedrock surface across the study site, highlighting complex bedrock morphology and sediment architecture.

Interpolated buried bedrock valley model
We constructed a high-resolution, local-scale bedrock elevation model using ERT-derived bedrock elevation data.The 2D elevation profiles were interpolated using the kriging method with a coarse 125 m grid discretization; this grid size corresponds to approximately one-half the maximum ERT line separation.Figures 10a-10c show the topography from a regional 10 m × 10 m digital elevation model (DEM), spatially interpolated bedrock elevations from the ERT data, and calculated overburden thicknesses (DEM elevation minus ERT-derived bedrock elevation distribution); these data were used to construct a quasi-3D conceptualization of the BBV (Fig. 10d).While the coarser grid discretization ensured a smooth interpolation between adjacent transects reducing potential interpolation artifacts, the coarser discretization ignored small-scale variations in bedrock elevation (e.g., bedrock mounds interpreted in Figs. and 6 are not captured in this model).
These data indicate an east-west-trending valley approximately 150->300 m wide and 27-51 m deep (Fig. 11a).The valley geometry is narrow with an undulating channel base.The valley floor rises in elevation from west to east, displaying an abrupt bend 150 m east of GDC-1 with a locally elevated bedrock floor, before terminating east of the Taylor Center well.The valley incises the upper Goat Island formation, exposing the full Eramosa Formation to the Quaternary infill (Fig. 11b).Based on the regional-scale bedrock elevation data (Gao et al. 2006;Figs. 3, 4), this bedrock valley is a tributary of the regional Guelph valley mapped by Cole et al. (2009).

Buried bedrock valley: morphology and origin
The development of BBVs can be affected by the regional tectonic setting, structural elements and basin evolution, preglacial paleoenvironmental conditions and the nature of glaciation in the area, and associated changes in glacial hydraulics and relative base level.In previously glaciated terrains, BBVs can be the result of a combination of preglacial, subglacial, and meltwater (proglacial or subglacial) erosional processes.In areas where BBVs are thought to have been created in a subglacial environment (i.e., tunnel valleys), the relative importance of direct glacial erosion and subglacial meltwater erosion as well as the steady-state versus catastrophic nature of the meltwater erosion is often the subject of considerable debate (Jørgensen and Sandersen 2006;Kehew et al. 2012).
Buried bedrock valleys in southern Ontario have been identified along broad valleys below escarpments, as escarpment reentrant and as individual or dendritic networks of channels developed across the Paleozoic bedrock surface, with a combination of preglacial and glacial erosional processes and regional tectonic stresses contributing to their development (Eyles et al. 1997;Russell et al. 2007;Gao 2011).As elsewhere in Europe and North America, catastrophic meltwater erosion has been proposed for some bedrock erosional forms associated with escarpments northwest of the study area (Tinkler and Stenson 1992;Kor and Cowell 1998).Recent review papers have highlighted the controversial aspects of these interpretations (Jørgensen and Sandersen 2006;Kehew et al. 2012), which are difficult to resolve for BBVs in Ontario with current data sets.Additional studies of BBVs and their infill will help to more clearly identify the relative importance of various erosional processes on the landscape.
The BBV in this study is an example of an individual channel segment developed on the Paleozoic bedrock surface that connects to a dendritic network of channels, namely the Guelph and Rockwood BBV west of the Niagara Escarpment as indicated by the bedrock surface of Gao et al. (2006), and to the Dundas valley further to the south (Eyles et al. 1997) (Fig. 3).The undulating morphology of the valley floor, sudden change in valley axis direction, and abrupt termination (Figs. 8,9) suggest Pleistocene subglacial meltwater erosion was a significant process in the development of this BBV (Jørgensen and Sandersen 2006;Gao 2011).It is difficult to assess the relative importance of other erosional processes (proglacial meltwater, direct glacial, and nonglacial) at this site with the current dataset, though all of these processes are likely to have contributed at least in part, considering the dendritic appearance of the channel network (Eyles et al. 1997), the multiple advances and retreats of ice in the area, the presence of 6 m of bedrock rubble at the base of the channel in GDC-1 (see next subsection), and the long period of geological time represented by the regional unconformity at the sediment-bedrock interface.For personal use only.

Nature and origin of Quaternary valley infill
The valley in this study is primarily infilled with a thick (tens of metres) package of diamict, and occasional laterally discontinuous coarse-grained lithologies at the top and bottom of the succession; this is evident in the detailed log of GDC-1 (Fig. 5) and the borehole record of the Arboretum Center well (Table 1), and is further supported by the natural gamma log in GDC-1 (Fig. 5) and the nature of the inverted modeled resistivity in the multiple ERT transects .Most transects across the valley are characterized by well-defined low-resistivity zones (approximately <250 ⍀•m; Fig. 7), with several transects showing laterally discontinuous units of higher resistivity attributed to coarser grained materials (>700-2000 ⍀•m;Figs. 7c, 7k, 7l) in the upper 10 m of the succession.Higher resolution ERT surveys (Fig. 8) captured more variability in the resistivity of the infill, with slightly higher resistivity values of up to 500 ⍀•m; perhaps recording some of the textural variability within the thick package of diamict noted in the detailed log of GDC-1, or the spatial variability in the infill sediment as noted in Figs.7f, 7g, and 7i.
Valley infills of BBVs tend to be variable (Kehew and Boettger 1986;Jørgensen and Sandersen 2006;Russell et al. 2007;Cummings et al. 2012;Szabo et al. 2013;Seyoum and Eckstein 2014).Russell et al. (2004) classified valley infills as either (i) complex with variable lithologies and unconformities recording a series of depositional and erosional events before and during several glaciations, or (ii) simple with one or two dominant lithologies (e.g., basal gravel capped with diamict or silt-clay), primarily recording one event or a relatively short period of time.Some valley infills contain significant coarse-grained colluvium along the sides of the valley, which interfinger with relatively finer-grained infill (Kehew and Boettger 1986;Jørgensen and Sandersen 2006;Kehew et al. 2012;Lajeunesse 2014), suggesting mass movement was significant during a pro-  Brunton (2009).The buried valley cuts into the upper Goat Island formation.Undiff., undifferentiated.[Colour online.]tracted period of valley infilling.In other instances, complex valley infill of BBVs can include sediment hosted buried valleys (Russell et al. 2004), exhibiting lateral variability in sediments and architecture along the valley axis (Russell et al. 2004;Zweirs et al. 2009;Burt 2011;Cummings et al. 2012).Recent studies in Ontario have shown that many BBVs contain complex successions of sediments recording periods of erosion and associated unconformities, multiple changes in depositional conditions over time, including changes from stadial to interstadial conditions and in some cases multiple glacial periods (Meyer and Eyles 1997;Davies et al. 2008;Zweirs et al. 2009;Bajc et al. 2012).The nearby Rockwood BBV contains a record of two separate ice advances (Catfish Creek and Canning Till) and a thick package of Lower Erie phase predominantly sandy stratified sediments (Burt 2011(Burt , 2012)).
Although this is perhaps due to its size and the limited highresolution continuous core data available at this time, the BBV infill in this study is comparatively simple (Fig. 11b), as the valley infill is dominated by one thick till (28 m thick out of a total borehole depth of 43 m) attributed to Catfish Creek ice (Nissouri phase).The 6 m of basal gravel that underlies this till consists of black shale fragments or rubble, which most likely came from the Vinemount Member (lower Eramosa Formation) during preglacial times following fluvial incision and (or) during glacial times as a result of glacial erosion and weathering of exposed bedrock, prior to its infill with Catfish Creek Till.Part of this rubble may be colluvium from adjacent valley walls (Kehew and Boettger 1986), and (or) part of it may result from in situ weathering of the bedrock valley floor as is evident in the modern Eramosa River.Additional subsurface data within the valley to delineate the spatial distribution of this rubble are needed to confirm which of these origins is most likely, as well as the timing of rubble deposition.At this time, the deposition of 6 m of rubble and the erosional event that created the BBV is interpreted to predate the Catfish Creek ice advance in the region prior to the Michigan Subepisode (Nissouri phase), as is the case in the nearby Rockwood BBV (Fig. 2; Burt 2011Burt , 2012)).
Other minor lithologies overlying the thick Catfish Creek Till, both within the valley and adjacent to it, record ice-marginal and glaciofluvial conditions.Stratigraphically, the clay and clay-rich diamict between 4 and 8.5 m at the top of the succession in GDC-1 is tentatively correlated to the Maryhill Till of Karrow (1987) based on its clay-rich texture and its stratigraphic position.Its depositional origin is difficult to establish with this limited borehole data; the Maryhill Till has been interpreted as a till or a product of sediment gravity flows in a glaciolacustrine environment (Karrow 1987).Although the clay and clay-rich diamict were both stiff, which could result from overconsolidation by overlying ice, these sediments are laterally associated with sand and gravel interpreted as glaciofluvial deposit (Figs. 6, 7), such that accumulation in a subaqueous environment as a result of ponding in an icemarginal environment is also plausible.Further studies that characterize the distribution of these fine-grained sediments and the interface between them and the adjacent coarse-grained facies are needed to resolve the depositional origin of the clay-rich interval in GDC-1.
The uppermost gravel unit in GDC-1 is associated laterally with the upper brown-coloured diamict in GDC-2B (Fig. 5, 2-8 m) and in the other water supply wells in the area (Table 1).The upper diamict in GDC-2B has been interpreted as a debris flow.Laterally, the ERT transects (e.g., Figs. 7, 8) provide evidence that the upper 10 m of valley infill are locally composed of discontinuous sand and gravel, which likely record glaciofluvial conditions.This lateral facies association of debris flow and glaciofluvial sediment in the valley infill, together with evidence of similar facies in laterally equivalent sediments outside of the valley (Fig. 5, GDC-2B and GDC-10A), and ERT transects (line 3 in Fig. 7c and lines 12 and 13 in Fig. 8) are all consistent with an ice-marginal setting associated with either Maryhill or Port Stanley ice (Port Bruce phase; Fig. 11b).

Buried bedrock valleys and underlying aquifer conceptualization
A conceptual model for karst formation around BBVs in the Guelph area was proposed by Cole et al. (2009).They observed a spatial relationship between productive municipal water supply wells completed at ϳ60 m depth and proximity to the regional BBV.In their model, valley incision resulted in steep hydraulic gradients, allowing (chemically) "aggressive" water to reach greater depths, enhancing dissolution of vertical joints, fractures, and conduits along flow paths, and ultimately discharging that water at the base of the valley (Cole et al. 2009).Deposition of lowpermeability glacial sediments during subsequent ice advances reduced groundwater flow and discharge along the valleys, limiting the rate of bedrock dissolution.They hypothesized that these dissolution features were responsible for the increased production capacity of municipal supply wells near the BBV.
If we apply the Cole et al. (2009) model to this study, dissolutionenhanced features would have developed as shallow and intermediate groundwater flowed through the Guelph and Eramosa formations towards the valley floor (Goat Island formation), prior to the valley being infilled with low-permeable clay and clay-rich diamict attributed to Catfish Creek ice (pre-Nissouri phase of the Michigan Subepisode) (Fig. 11b).Whereas some of these pathways may be dissolution-enhanced features as hypothesized by Cole et al. (2009), most of the outcrop and cored hole data to date suggest well-connected fracture networks are more significant (Perrin et al. 2011).Determining the relative importance, and lateral and vertical connectivity of these pathways, is the subject of ongoing research, as it will likely impact flow paths and time of travel through the Guelph, Eramosa, and Goat Island formations, and consequently, the spatial distribution of recharge to the primary Gasport formation aquifer.
In a separate study of GDC-6, dissolution-enhanced channelling features (vuggy conduits and fractures) were observed in core and video logs primarily within two intervals (267-268 and 272-275 masl, Fomenko 2015); these zones lie within the Gasport formation, roughly 17-25 m below the interpreted bedrock valley floor.During the development of GDC-6 and adjacent boreholes, when pressurized air was released from a compressor at the base of the Gasport formation (top of Cabot Head shale), circulation was observed at the surface at other wells located 7.5-15 m away.This strong transmissivity supports the existence of karst features and well-connected open fractures within the Gasport formation that has been documented locally and regionally (e.g., Brunton 2009;Cole et al. 2009).The average production zone interval across the City of Guelph based on data from 18 municipal supply wells (as presented in Cole et al. 2009) was 268-278 masl; this broad regional zone encompasses the two karst zones observed in GDC-6.The bulk of the dissolution that occurred within the Gasport may be the result of relatively deeper groundwater flow discharging into the floor of the regional BBV system (Fig. 4), if and where it intersects the Gasport formation (i.e., conduit feature elevation) further down the BBV thalweg.The presence of rubble at the base of the valley, if laterally extensive, would also be an important transmissive feature in this 3D flow system.
Further study of the distribution, morphology, and nature of the Quaternary infill of these BBVs relative to the underlying bedrock stratigraphy and hydrogeology is required to more fully understand the relationship between BBVs, preferential pathways, recharge distribution, and high-capacity municipal production wells.A better understanding of these aspects of the groundwater flow system will in turn improve our characterization of groundwater residence time and capture zone geometries that are central to groundwater resource management and source water protection.For personal use only.

Conclusions
This study demonstrates the synergistic integration of several surface geophysical methods combined with sedimentological analysis of continuous cores and hydraulic data for characterization of the impact of BBVs and their Quaternary infills on the present-day bedrock groundwater flow system.Similar to larger regional valleys, these local-scale features can be defined by complex channel morphology and Quaternary infill, requiring a suite of tools at scales pertinent to localized studies of municipal production well recharge zones and contaminant transport and fate.
The primary goal of this study was to demonstrate the capabilities and limitations of the ERT technique for mapping a BBV at a relatively local scale.Here, the method was used to delineate a bedrock surface beneath Quaternary deposits ranging from a few metres to >50 m thick.The ERT-derived bedrock depths were confirmed through refraction surveys and available borehole logs.Although the ERT provided reliable estimates of bedrock depth, the existence of a rubble layer along the valley floor may have led to an underestimation of the valley-floor extent.While the ERT was generally effective at differentiating between the clay-rich diamict and underlying dolostone at a local scale, available borehole information did aid our interpretation of the bedrock interface in areas near the edges and base of the models, or where Quaternary infill changed from clay-rich sediment to coarser grained sand-gravel deposits near the bedrock surface.
The bedrock valley within the study area is thought to be a tributary to the Guelph and Rockwood valleys west of the Niagara Escarpment and the Dundas valley to the south.The valley appears to originate in the northeast corner of the study site, where it trends southwestward, widening and deepening as it connects to the Eramosa River (current exposed section of the Guelph BBV).The width of the valley ranged from 150 to >300 m in the southwest corner.The highly uneven and complex nature of this valley morphology is consistent with subglacial meltwater erosion, though it is likely that other erosional processes (proglacial meltwater, direct glacial erosion, or preglacial fluvial incision) contributed to its development.
A comprehensive analysis of unpublished geotechnical reports and hydraulic data sets within the Guelph area by Gautrey (2004) revealed a statistical relationship between high-capacity groundwater production wells and proximity to BBVs.This increase in production was attributed to the presence of dissolution-enhanced fractures and karst conduit features that formed through progressive bedrock incision prior to being covered by low-permeability Quaternary deposits (Cole et al. 2009).Our study showed consistently higher bedrock transmissivity outside the BBV and relatively lower transmissivity beneath the valley between the base of the valley floor and the top of the regional aquitard.These differences in bulk hydraulic property may be explained by an increase in fracture connectivity and (or) karst conduit formation as deep groundwater discharged in the valley floor prior to being infilled with Quaternary deposits.
The valley in this study is primarily infilled with a thick (tens of metres) package of diamict attributed to Catfish Creek ice (Nissouri phase), and occasional laterally discontinuous coarse-grained lithologies in the top and bottom of the succession.One other diamict unit within the succession is tentatively correlated to the Maryhill Till (Port Bruce phase).Observed lateral facies association of debris flow and glaciofluvial sediment in the uppermost valley infill, together with evidence of similar facies in laterally equivalent sediments outside of the valley, and nearby ERT transects are all consistent with an ice-marginal setting associated with the ice retreat phase.The spatial distribution and geometry of BBVs and their variable infill may play a role in the transmissivity characteristics of the incised bedrock formations.The lateral and vertical connectivity of fractures and their 3D connectivity to karst conduit and dissolution-enhanced features will likely impact the spatial distribution of recharge and groundwater flow lines, ultimately affecting groundwater travel times and capture zone delineation.

Fig. 1 .
Fig. 1.Lower and Middle Silurian traditional stratigraphic nomenclature for Guelph area based on Armstrong and Carter (2010) and with proposed revisions by Brunton (2009).*Cabot Head shale of the Cataract Group.

Fig. 4 .
Fig. 4. Study area showing extent of geophysical transects, research boreholes, and water supply wells owned and operated by the University of Guelph.Inset map shows areal extent of geophysical study area southeast of the Eramosa River within the City of Guelph.Cole et al. (2009) identified multiple buried bedrock valleys within the city limits, which they denoted as the Guelph buried bedrock valley; the primary Guelph valley connects to the Rockwood buried valley to the northeast.Coordinate system: NAD83 UTM zone 17N.

aFig. 5 .
Fig.5.Stratigraphic Quaternary and bedrock logs from three locations across the buried valley feature: GDC-2B (south of valley); GDC-1 (within valley); GDC-6 and GDC-10A (north of valley).Downhole geophysical logs include natural gamma, induced electrical resistivity, and P-wave velocity; additional FLUTe transmissivity profiles were performed within the open bedrock coreholes.The triangle symbol along the transmissivity profiles indicates the transmissivity (T) of the bedrock between bottom of the valley (ϳ288 masl) and the top of the Cabot Head shale.The term broken rock* refers to either mechanically pulverized bedrock (GDC-10A) or broken rock fragments (GDC-1).Note that the Amabel Formation is shown here with local informal subdivisions of Irondequoit, Gasport, and Goat Island formations as proposed byBrunton (2009).b, unsealed bedrock interval; cl, clay; cps, counts per second; g, gravel; mbgs, metres below ground surface; s, sand; si, silt; Vf, very fine.

Fig. 8 .
Fig. 8. Higher resolution ERT transects with 5 m electrode spacing corresponding to (a) line 13 and (b) line 12.Resistivity models show spatially variable Quaternary deposits and an undulating bedrock surface with abrupt ridges denoted as bedrock mounds (bm).

Fig. 9 .
Fig. 9. Results of seismic refraction measurements collected along (a) SR-1 and (b) SR-2 with delay-time interpretation (top) and tomographic inversion models (bottom).The + symbols on the tomographic inversion correspond to estimated bedrock depths using delay-time technique, while the dotted line represents the interpreted bedrock interface based on a velocity (v) threshold of 3200 m/s.[Colour online.]

Fig. 10 .
Fig. 10.Conceptualization of buried bedrock valley based on (a) 10 m × 10 m provincial digital elevation model (DEM), (b) ERT-derived bedrock elevation model (black lines represent the location of the ERT transects), and (c) calculated Quaternary isopach map.(d) Vertically exaggerated bedrock valley shows a complex and hummocky channel bottom with abrupt changes in valley strike, which connects to the Guelph buried valley to the southwest and an abrupt termination towards the east.Coordinate system: NAD83 UTM zone 17N.

Fig. 11 .
Fig. 11.(a) Resistivity-derived bedrock elevations showing extent of buried bedrock valley within the study area (coordinate system: NAD83 UTM zone 17N) with (b) cross-sectional view of the bedrock morphology and Quaternary infill along transect A-A= based on resistivity and stratigraphic logs.Note that the Amabel Formation is shown here with local informal subdivisions of Irondequoit, Gasport, and Goat Island formations as proposed byBrunton (2009).The buried valley cuts into the upper Goat Island formation.Undiff., undifferentiated.[Colouronline.] Pagination not final (cite DOI) / Pagination provisoire (citer le DOI) Steelman et al. 15 Published by NRC Research Press Can.J. Earth Sci.Downloaded from www.nrcresearchpress.com by UNIV GUELPH on 05/11/17

Table 2 .
Summary of seismic refraction results.
a Based on a Quaternary-bedrock velocity transition of 3200 m/s.