Synthesis and Model of Formation-Water Flow, Alberta Basin, Canada 1

Transcription

1 Synthesis and Model of Formation-Water Flow, Alberta Basin, Canada 1 Stefan Bachu 2 ABSTRACT Based on a large amount of publicly available data, several studies have previously examined the flow of formation waters in different parts of the Alberta basin, offering various interpretations as to the causes of the observed pressure regime and flow pattern; however, there has been no synthesis of these diverse studies on a basin-wide basis. Accordingly, these studies are critically reviewed in this paper and synthesized in a new basin-scale model of the flow of formation waters in the Alberta basin. The proposed regional-scale model has significant implications for understanding hydrocarbon migration pathways, ore genesis, the geothermal regime, and deep waste disposal in the Alberta basin. Several flow systems, each one driven by a different mechanism, are identified, together with the main processes leading to the nonhydrostatic pressures observed in the basin. Two megahydrostratigraphic successions and associated flow systems are recognized. The first succession corresponds to the pre-cretaceous passive-margin stage of basin development, and consists of thick, carbonate-dominated aquifer systems separated by shaly aquitards and evaporitic aquicludes. A northeastward basin-scale flow system is driven by basin topography, with recharge in Montana and discharge in northeastern Alberta. Southwest-to-northeast regional-scale flow adjacent to the fold and thrust belt is probably the result of past tectonic processes. The salinity of formation water in pre-cretaceous aquifers is high, and generally increases both northward and with depth. This variability is the result of increased water-rock reactions (mineral solubility) with increased temperature, and of incomplete flushing by meteoric water. Copyright The American Association of Petroleum Geologists. All rights reserved. 1Manuscript received January 26, 1994; revised manuscript received August 18, 1994; final acceptance November 3, Alberta Geological Survey, ADOE, Street, Edmonton, Alberta T5K 2G6, Canada. The author wishes to express his thanks and gratitude to Brian Hitchon for useful discussions and critical review of the manuscript. As a result of salinity variations, flow-retarding buoyancy effects can be important. The second megahydrostratigraphic succession corresponds to the post-jurassic foreland stage of basin evolution, and consists of thick, shaly aquitard systems and relatively thin sandstone aquifers. In the southwestern part of the basin, flow in isolated aquifers is driven southwestward by erosional rebound in the thick intervening shales, downdip toward the fold and thrust belt. The salinity of formation waters in the post-jurassic aquifers is low. Mixing of waters and interference between the major flow systems takes place along the pre-cretaceous unconformity where successively older pre-cretaceous aquifers subcrop from west to east. Flow in shallow local systems at the top of the sedimentary succession is driven by local topography. INTRODUCTION Sedimentary basins are of great economic importance because of their energy and mineral resources. In recent years it has been widely recognized that the flow of formation water may play an important role in the redistribution of terrestrial heat and minerals, in the migration and accumulation of hydrocarbons, and in the genesis of ore deposits. Particularly in the last decade this recognition led to numerous observational and numerical studies of fluid flow in sedimentary basins, with the purpose of identifying flow mechanisms and developing basin-scale exploration tools. Because of their complexity, numerical studies generally have focused on simulating individual rather than a combination of the processes driving the flow of formation waters in various sedimentary basins. Observational studies also had the tendency to emphasize one process or another, such that, in many basins, more than one flow-driving mechanism has seldom been recognized. In some basins the data are scarce, in which case the resulting image of fluid flow is inherently broad and sometimes lacks the necessary detail for identifying all pertinent features. In the opposite situation, the amount of existing information for basins like the AAPG Bulletin, V. 79, No. 8 (August 1995), P

2 1160 Flow of Formation Water, Alberta Basin Gulf Coast and western Canada is overwhelming, making the data processing and interpretation a Herculean task even in this day of computer-based technology. Superimposed on the issue of existing information is the problem of data availability. Unless theoretical and numerical in nature, most studies rely on data collected by industry, which usually retains ownership and restricts availability. The Alberta basin in western Canada is in a mature stage of exploration. With more than 150,000 wells drilled so far, it is probably the only basin in the world with so much publicly available information; however, the basin size, its complexity, and the huge database have made it practically impossible, so far, to study in detail the flow of formation water in the entire basin. Using a limited amount of data, Hitchon (1969a, b) interpreted the flow in the basin as being in equilibrium with and driven by the present-day topography. In the quarter century since then, the flow of formation waters has been analyzed and described for various parts of the Alberta basin. Depending on the knowledge at the time and the particular part of the basin being studied, various models and mechanisms driving fluid flow in the basin have been proposed to explain the observed flow patterns. These models are sometimes different, even contradictory. By analyzing all the previous studies and reinterpreting some of the observed flow characteristics, this paper presents a synthesis into a consistent basin-scale model of the flow of formation waters in the Alberta basin. The resulting new model shows that a number of equally important flow systems driven by different mechanisms are active in the basin. This has significant implications with regard to hydrocarbon migration and accumulation, ore genesis, geothermal regime, and fate of wastes disposed of in the basin. First, to understand the flow in the Alberta basin better, a review of the various mechanisms driving fluid flow in sedimentary basins and of the processes leading to nonhydrostatic pressure distributions is essential. FLOW-DRIVING MECHANISMS AND PRESSURING PROCESSES Several mechanisms drive the flow of formation waters, and various processes lead to nonhydrostatic pressures in sedimentary basins. Bredehoeft and Hanshaw (1968) and Bradley (1975) provided short reviews of factors contributing to the origin and maintenance of nonhydrostatic fluid pressures, and Neuzil (1986) provided a more extensive mathematical analysis of some of these factors together with a discussion of flow-driving phenomena. Notwithstanding these and other similar works, the following brief review attempts to establish their relative importance in relation to basin characteristics, to understand better the complex flow systems identified in the Alberta basin and their origins. Fluid flow in porous media is described by the equation of momentum conservation expressed empirically by Darcy s law. The generalized Darcy s law written in terms of hydraulic head rather than pressure explicitly shows that the flow of formation water is driven by hydraulic gradients (hydraulic flow), by density differences (buoyancy flow), and by variations in fields other than hydraulic, such as thermal and chemical concentration (osmotic flow) (Bachu, 1995). Laboratory experiments reviewed by Neuzil (1986) show that the osmotic hydraulic conductivity coefficients are exceedingly small. Thus, high temperatures or concentration gradients are needed to induce osmotic flow comparable in strength with hydraulic- and buoyancy-driven flows; these are seldom found in sedimentary basins. Only for very tight clays and shales are chemo-osmotic and hydraulic conductivities comparable. Osmotic flow across shales was interpreted to cause nonhydrostatic pressures in several cases (Hitchon, 1969b; Kharaka and Berry, 1974; Marine and Fritz, 1981). However, for largescale flow systems, osmotic flow across shale layers is generally deemed unimportant by comparison with other driving mechanisms because geological media are neither very efficient nor uniform membranes (Bradley, 1975; Phillips, 1983). Hydraulic flow may be caused by gradients resulting from position (elevation) differences in the earth s gravitational field. This is the case of topography-driven flow, where the formation waters in a sedimentary basin flow in local, intermediate, and regional systems driven by topographic relief differences at the water table (Toth, 1963). Important features of this type of flow are that the flow is from high to low topographic elevations, and pressures are higher everywhere than the pressure corresponding to the lowest topographic point in the basin. Steady-state, regional-scale topography-driven flow was identified in the Palo Duro basin (Senger and Fogg, 1987) and in the Denver basin (Belitz and Bredehoeft, 1988). Subhydrostatic pressures in these basins were explained as being caused by extensive low-permeability strata in recharge areas and high-permeability strata at depth and in discharge areas (drain effect). Hydraulic gradients also may be the result of pressure differences caused by factors other than topography. Geochemical processes that release or use water, like gypsum and montmorillonite dehydration, may produce above-hydrostatic pressures in low-permeability environments (Hanshaw and Bredehoeft, 1968). However, Bradley (1975) and Magara (1975) questioned the

3 Bachu 1161 relative importance of geochemical processes in the generation and maintenance of nonhydrostatic pressures in sedimentary basins. Based on qualitative reasoning and order-of-magnitude analysis, Bradley (1975) and Hunt (1990) suggested that water thermal expansion is significant enough to produce nonhydrostatic pressures when heated during sediment burial or cooled during uplift and erosion. However, numerical simulations of compacting basins (Bethke, 1985, 1986; Shi and Wang, 1986; Corbet and Bethke, 1992) showed that changes in water density caused by temperature variations are negligible in the generation of nonhydrostatic pressures. According to these simulations, porosity changes caused by variations in mechanical stresses during compaction are the main factor in generating and maintaining above-hydrostatic pressures in geological media characterized by high compressibility and low permeability (e.g., shales). Nonhydrostatic pressures are rapidly dissipated in more permeable and less compressible rocks like sandstones and carbonates. Magara (1976) and Bredehoeft et al. (1988) showed that compaction of interbedded sand-shale successions induces significant outward lateral flow in the intervening sandy layers. In a reverse process, subhydrostatic pressures may be generated and maintained for long periods of time by erosional stress unloading, as shown by several theoretical and numerical studies (Neuzil and Pollock, 1983; Neuzil, 1986; Corbet and Bethke, 1992) and measurements (Neuzil, 1993). Pore-space rebound in thick, highly compressible low-permeability shales creates a drop in pressure, which generates a suction effect in adjacent aquifers. If these aquifers are isolated by shales, pressures may become subhydrostatic, with inward lateral flow from the permeable boundaries (Neuzil and Pollock, 1983). Erosional unloading seems ineffective in creating large-scale underpressuring in basins where thick shale layers are absent, like the Palo Duro basin in Texas (Senger et al., 1987). Loading, compaction, and erosion at the top of a sedimentary succession induce vertical stress changes that cause porosity variations leading to nonhydrostatic pressures. Thrusting and tectonic compression at a basin margin may induce lateral stress variations that cause fluids to be expelled from the margin sediments and travel into the foreland basin (Oliver, 1986; Bethke and Marshak, 1990). Numerical simulations show that tectonic compression induces flow or increases the velocity of formation waters flowing in the basin by an amount of the same order of magnitude (1 10 cm/yr) as the tectonic plate movement (Ge and Garven, 1989; Deming et al., 1990). Compaction, erosion, and tectonic compression generate external stresses. Phase changes of a substance other than water, partially or totally filling the pore space, may also induce fluid flow and nonhydrostatic pressures because the volume change associated with this process generates internal stresses. This is the case for organic material undergoing hydrocarbon generation (Hedberg, 1974). Similarly, solution or precipitation of minerals in formation water affects the porosity and permeability of the rocks, theoretically inducing flow and, potentially, nonhydrostatic pressures. However, unless the rock has very low permeability to start with, these pressure variations dissipate rapidly. Finally, nonhydrostatic pressures in a basin may be the result of transient effects of a paleoflow system in the process of adjusting to new boundary conditions created by topographic changes. Bradley (1975) called these fossil pressures. Toth and Millar (1983) and Toth and Corbet (1986) explained nonhydrostatic pressures in the Alberta basin as being transient paleoflow systems in an incompressible rock matrix. However, Neuzil (1985, 1986) questioned the validity of the incompressibility assumption, and Corbet and Bethke (1992) showed that flow directions in deep aquifers cannot reflect topographic features that have been eroded. Observational evidence, laboratory measurements, and mathematical and numerical modeling previously reviewed indicate that the main mechanisms driving the flow of formation waters in sedimentary basins are (1) hydraulic gradients resulting from topographic (elevation) differences; (2) hydraulic gradients resulting from pressure variations generated by various processes, the most important being related to changes in mechanical stresses; and (3) buoyancy caused by significant differences in formation-water density. Not all these mechanisms and processes are currently active in any given sedimentary basin; however, based on observational evidence and interpretation, it seems that most of them are presently driving different flow systems in various parts of the Alberta basin. GEOLOGY AND HYDROSTRATIGRAPHY OF THE ALBERTA BASIN The Alberta basin is located in western Canada east of the Rocky Mountains and consists of a northeasterly tapering wedge of sedimentary rocks (Figure 1). The basin is bordered to the east and northeast by the exposed Canadian Precambrian Shield and to the west and southwest by the Cordilleran foreland fold and thrust belt. The Alberta basin is separated from the intracratonic Williston basin to the southeast by the Bow Island arch, and from the Mackenzie corridor leading to the Mackenzie basin in the north by the Tathlina arch. The buried Precambrian basement dips to the

4 (a) 62 o N 126 o W 96 o W 62 o N 60 o YUKON Tathlina Arch Great Slave Lake NOTHWEST TERRITORIES 60 o BRITISH COLUMBIA ALBERTA Lake Athabasca SASKATCHEWAN MANITOBA Athabasca Oil Sands Peace River Arch E Canadian Shield W Alberta Basin Cold Lake Oil Sands Rocky Mountains Lake Winnipeg 49 o N 126 o W Bow Island Arch Williston Basin 49 o N 96 o W CANADA U.S.A. MONTANA 110 o U.S.A. (b) W E Tertiary Cretaceous Elevation (m) Jurassic Triassic Mississippian Cambrian Devonian Precambrian Scale 0 50 km Well Locations Figure 1 Characteristics of the Alberta basin: (a) location and main tectonic elements, and (b) geological dip cross section, based on actual wells.

5 Bachu 1163 (a) 62 o 126 o 108 o W 62 o N (b) 62 o 126 o 108 o W 62 o N Tathlina Arch Great Slave Lake Canadian Shield Tathlina Arch Great Slave Lake Canadian Shield Lake Athabasca Caribou Mtns Birch Mtns. Lake Athabasca Foreland Fold and Thrust Belt Foreland Fold and Thrust Belt 800 Swan Hills o 126 o Bow Island Arch o N 108 o W 49 o 126 o Cypress Hills Bow Island Arch o N 108 o W Figure 2 Bounding surfaces of the Alberta basin: (a) Precambrian crystalline basement, and (b) topography (isolines in meters). southwest (Figure 2a) and consists mainly of Archean crystalline rocks and Aphebian supracrustal rocks that were modified by deformation, metamorphism, and magmatism during the early Proterozoic Hudsonian orogeny. Recent erosion of the overlying Phanerozoic wedge created a relatively mild topographic relief ranging between more than 1400 m in the west and southwest along the fold and thrust belt and less than 200 m in the northeast near the Great Slave Lake (Figure 2b). Major isolated upland areas are the Caribou and Birch mountains in northern Alberta, the Swan Hills in west-central Alberta, and the Cypress Hills in southern Alberta just east of the Bow Island arch. Based on Porter et al. (1982) and Ricketts (1989), a short geological description of the Alberta basin is presented to explain the hydrostratigraphic delineation and the flow of formation waters in the basin. Geology The Alberta basin was initiated during the late Proterozoic by rifting of the North American craton. Following initial rifting, thermal contraction led to the transgressive onlap of the North American cratonic platform from Middle Cambrian to Middle Jurassic time. The corresponding sedimentary succession (Figure 3) is dominated by shallow-water carbonates and evaporites. A basal diachronous quartz sandstone is overlain by Middle Cambrian to Lower Ordovician strata dominated by marine shale deposits. As a result of pre-middle Ordovician erosional beveling and of major pre- Middle Devonian erosion, Ordovician Silurian strata are absent in the basin, whereas Cambrian strata are present only in the southern half of the basin and in an area in the northwest, north of the Peace River arch. Everywhere else Devonian strata directly overlie the Precambrian basement. The Lower and Middle Devonian strata of the Elk Point Group, consisting of a succession of red beds and massive halite units overlain by carbonates and evaporites, filled the topographic depressions formed in the east-central and northern parts of the Alberta basin. The Upper Devonian depositional succession records shallow-marine carbonate platforms and associated reef complexes. Widespread carbonate platforms (Beaverhill Lake, Woodbend, Winterburn, and Wabamun groups, and Cooking Lake and Grosmont formations, Figure 3) constitute the major components of the succession. Basin-filling limestones and shales are minor components, except for the Ireton Formation. Mississippian deposition began with a thin, widespread veneer of

6 1164 Flow of Formation Water, Alberta Basin Period Tertiary Cretaceous Jurassic Lower Middle Upper Lower Upper L Triassic Permian Pennsylvanian Mississippian Cambrian Devonian U M Silurian Ordovician U Pre - cambrian M L Edmonton Colorado Stratigraphic Nomenclature Group Formation Elk Point Mannville Woodbend Upper Lower Paskapoo Scollard Battle Whitemud Horseshoe Canyon Bearpaw Belly River Lea Park Cardium Wabamun Viking Winterburn Ireton Beaverhill Lake Banff Exshaw Prairie Winnipegosis Not deposited Milk River Second White Speckled Sandstone Stoddart Rundle Leduc Clearwater Grosmont Cooking Lake Basal Sandstone Not deposited post - Colorado aquifer - aquitard system Hydrostratigraphy Scollard - Paskapoo aquifer Battle aquitard Horseshoe Canyon aquifer Bearpaw aquitard Belly River aquifer system Lea Park aquitard Milk River aquifer Colorado aquitard system Upper Mannville aquifer Clearwater aquitard Lower Mannville aquifer Jurassic aquitard Mississippian - Jurassic aquifer system Exshaw - Banff aquitard Upper Devonian aquifer system Ireton aquitard Middle - Upper Devonian aquifer system Prairie aquiclude - aquitard system Winnipegosis aquifer Elk Point aquiclude system Cambrian aquitard system Basal aquifer aquiclude Figure 3 Relevant basin-scale stratigraphic and hydrostratigraphic delineation and nomenclature, Alberta basin.

7 Bachu 1165 organic shale (Exshaw Formation, Figure 3), followed by an interbedded shale-to-carbonate succession (Banff Formation). This trend was continued by the overlying carbonate succession of the Rundle and Stoddart groups. Pennsylvanian to Jurassic strata are present only in the northwest in the Peace River arch region and consist of interbedded sandstones, siltstones, and carbonates. As a result of accretion to the western margin of North America of oceanic terranes and island arches, isostatic flexure of the lithosphere formed the foreland basin, which filled with synorogenic clastics derived from the Cordillera. Because of westward dipping, erosion exposed, from west to east, successively older strata from Middle Jurassic to Lower Devonian. The overlying Cretaceous strata are divided into three depositional successions: the Mannville, Colorado, and post-colorado groups. The Mannville Group, which corresponds to the Columbian orogeny, consists of fluvial and estuarine valley-fill sediments, and sheet sands and shales deposited by repeated marine transgressive-regressive events. The Colorado Group corresponds to a lull in tectonic plate convergence characterized by a widespread marine transgression. Colorado strata consist predominantly of thick shales within which are isolated, thin, laterally extensive sandy units (Viking, Second White Speckled Sandstone, and Cardium formations and their equivalents, Figure 3). Post-Colorado Cretaceous and Tertiary strata correspond to the Laramide orogeny and subsequent tectonic relaxation, and consist of eastward-thinning nonmarine clastic wedges intercalated with argillaceous sediments. This cyclicity is best developed in the southern part of the basin, where the Milk River, Belly River, Horseshoe Canyon, and Scollard-Paskapoo formations refer to the clastic wedges, and the Pakowki (Lea Park), Bearpaw, and Battle formations refer to the intervening shales. In the central and northern parts of the basin many of these cycles are absent. Following the end of Laramide deformation in the Paleocene, Tertiary, and Holocene, erosion removed between 1 and 4 km of sediment (Nurkowski, 1984; Bustin, 1992). A variety of glacial and postglacial surficial deposits of Quaternary age overlie the bedrock. Basin-Scale Hydrostratigraphy The sedimentary rocks in the basin constitute the matrix through which the formation fluids move. Depending on their permeability, the sedimentary strata can be classified from a hydrogeological point of view into aquifers, aquitards, and aquicludes. Various rock types have different characteristics with regard to fluid flow (e.g., sandstones and carbonates generally constitute aquifers, shales are aquitards, and evaporites are aquicludes). A hydrostratigraphic unit (aquifer, aquitard, or aquiclude) comprises one or more geological units that are in contact and exhibit similar characteristics with regard to fluid flow. Hydrostratigraphic units are usually defined individually at the local scale. Hydrostratigraphic systems are complex groups of hydrostratigraphic units that exhibit common overall characteristics at a regional scale. Thus, an aquifer system behaves generally like an aquifer even if minor aquitards are present, whereas an aquitard system behaves generally like an aquitard even if individual, isolated aquifers are present. The regional-scale hydrostratigraphy of the Alberta basin (Figures 3, 4) is described in the following paragraphs in terms of major, basin-scale hydrostratigraphic systems, based on the general lithology of the sedimentary strata and on their behavior with respect to the flow of formation waters as identified by previous hydrogeological studies in various parts of the basin (Hitchon, 1969a, b; Toth, 1978; Schwartz et al., 1981; Toth and Millar, 1983; Bachu et al., 1986; Toth and Corbet, 1986; Hitchon et al., 1989, 1990; Bachu and Underschultz, 1993, 1995). The Precambrian crystalline basement is considered to be an aquiclude. The overlying Basal Cambrian Sandstone in the southern half of the basin, and Granite Wash detritus in the northern half, form the relatively thin Basal aquifer. The remainder of the predominantly shaly Cambrian strata constitute the Cambrian aquitard system, present in the southern half of the basin. In the central, eastern, and northeastern parts of the basin, the thick, predominantly halite Lower Elk Point subgroup forms the Elk Point aquiclude system. The carbonates of the Winnipegosis Formation and its equivalents comprise the regionally extensive Winnipegosis aquifer, which overlies the Cambrian aquitard system in the southern part and the Elk Point aquiclude system in the east-central and northern parts. The overlying thick evaporitic beds (Prairie Formation and its equivalents) and thin shales constitute the Prairie aquiclude-aquitard system. Over most of the basin where the Prairie salt is present, this hydrostratigraphic unit has aquiclude characteristics; everywhere else it has aquitard characteristics. The overlying platform carbonates and associated reef complexes of the Beaverhill Lake Group and Cooking Lake and Leduc formations form the Middle-Upper Devonian aquifer system. Hydrostratigraphically, the thick overlying Ireton shales are an aquitard. The thick platform carbonates of the basinwide Winterburn and Wabamun groups, together with the Grosmont Formation present in central and northeastern Alberta, form the Upper Devonian aquifer system. In the western and

8 1166 Flow of Formation Water, Alberta Basin W5M W4M Elevation (m) Paskapoo Battle Scollard Horseshoe Canyon Belly River Colorado Mississippian -Jurassic Bearpaw Cambrian Basal Exshaw-Banff Jurassic Lea Park Upper Devonian Ireton Upper Mannville Middle-Upper Devonian Elk Point Lower Mannville Prairie Clearwater Winnipegosis Aquifer Aquitard Aquiclude Scale 0 50 km Figure 4 Basin-scale hydrostratigraphic dip cross section, Alberta basin (see Figure 1 for location and Figure 3 for geological and stratigraphic reference). southern parts of the basin, the thin Exshaw shales and the shaly lower part of the Banff Formation constitute the Exshaw-Banff aquitard. Although minor shale layers are present in the remainder of the Mississippian, Permian, Triassic, and Lower Jurassic, the entire succession constitutes a single regional aquifer system, the Mississippian Jurassic. The thin, predominantly shaly remaining Jurassic strata form the Jurassic aquitard. Because of significant pre-cretaceous erosion, all the Devonian-to-Jurassic hydrostratigraphic units and systems, except the Elk Point aquiclude, subcrop from east to west at the pre-cretaceous unconformity. The Lower Mannville strata covering the entire basin have aquifer characteristics, except in northeastern Alberta where they have aquitard characteristics because of the bitumen filling the pore space in the areas of the Athabasca and Cold Lake oil sand deposits (Figure 1), and in northern Alberta where the entire Mannville Group is shaly. Because of their change from nonmarine in the south to marine in the north, the Upper Mannville strata have varying hydrogeological characteristics. In the southern part of the basin they form a single aquifer overlying the Lower Mannville aquifer. Over most of the basin, from south-central to north-central, the shaly Clearwater aquitard separates the Upper Mannville and Lower Mannville aquifers. In the northern part of the basin, the Upper Mannville is an aquitard. The thick, shale-dominated Colorado succession forms the Colorado aquitard system, which practically blankets the basin except for the northeastern corner where it has been removed by Tertiary to Holocene erosion. The thin Viking, Second White Speckled Sandstone (Dunvegan in the central-western part), and Cardium sandstone aquifers are isolated and encapsulated within the Colorado aquitard system. The remainder of the stratigraphic succession forms the post-colorado aquifer-aquitard system, which consists of an alternating succession of aquifers and aquitards that are better developed in the southern part of the basin. Because of Tertiary to Holocene erosion, the post- Colorado aquifer-aquitard system is present only in the central-western and southwestern parts of the Alberta basin. PREVIOUS FLOW ANALYSES Despite the publicly available wealth of data for the Alberta basin (more than 150,000 wells with associated drill-stem tests and formation-water analyses), only Hitchon (1969a, b) attempted to analyze the flow at the scale of the entire basin using nearly 3000 pressure readings. Subsequent studies treated all or parts of the hydrostratigraphic succession in various sedimentary blocks chosen more for ease of geographic delineation and data extraction than for hydrogeological considerations (Figure 5). Based on Toth s (1963) theoretical model, Hitchon (1969a, b; 1984) assumed that the flow in the basin is at steady state, in equilibrium with and driven by the present topography, and controlled by basin geology (lithology). According to his model, major flow recharge takes place in the Foothills, Caribou Mountains, Swan Hills, and Cypress Hills, whereas the major basin discharge takes place in the northeast at the

9 Bachu o 54 o 52 o 50 o 62 o 60 o 58 o G 126 o 122o 1 G 3 Peace River Arch Red Earth Foreland Fold and Thrust Belt 118 o Southwest Alberta Cold Lake Taber lowest point in the basin (Figure 6). Intermediate and local flow systems are superimposed over this regional system. Hitchon (1969a, b) identified low (subhydrostatic) hydraulic heads in the Upper Devonian and Mississippian carbonate aquifers, which he attributed to the drain effect of these highly permeable strata that channel the regionalscale flow in the basin toward the Athabasca area where Upper Devonian rocks are exposed at the surface. Closed low hydraulic heads and anomalous salinities in the Viking and Mannville aquifers in south-central Alberta were attributed to osmotic effects by Hill et al. (1961) and Hitchon (1969a, b). However, Dickey and Cox (1977) speculated that the subhydrostatic pressures observed in the Viking aquifer in central Alberta might be related to mechanical and aquathermal effects caused by erosion. The flow of formation water in the centralnorthern part of the basin was studied for the entire sedimentary succession by Toth (1978), Hitchon et al. (1989, 1990), and Bachu and Underschultz (1993) (Figure 5), using about 27,700 drill-stem tests and 36,000 water analyses. Toth (1978) concluded that the flow in the Upper G 2 Canadian Shield Northeast Alberta W 108 o N 62 o G o 110 o o 49 N 108 W Figure 5 Location of previous regional-scale hydrogeological studies in the Alberta basin: Red Earth (Toth, 1978), Taber (Schwartz et al., 1981; Toth and Corbet, 1986), Cold Lake (Hitchon et al., 1989), Peace River (Hitchon et al., 1990), northeastern Alberta (Bachu and Underschultz, 1993), and southwestern Alberta (Bachu and Underschultz, 1995). Lines G 1 G 2 and G 3 G 4 denote the cross sections of Garven (1985, 1989). o Devonian-to-surface hydrostratigraphic succession is at steady state, in equilibrium with and driven by the present topography, whereas the flow in the Middle-Upper Devonian and Winnipegosis aquifers is in a transient state of adjustment from Pliocene to present topography (fossil pressures), being cut off from continued recharge in the area by the flow along the pre-cretaceous unconformity. In a subsequent one-dimensional mathematical analysis, Toth and Millar (1983) showed that it is possible to simulate the observed hydraulic heads (pressures) in the pre-ireton aquifers as being in a transient state if the rock matrix is assumed to be incompressible (an assumption questioned by Neuzil, 1985). Toth s (1978) conclusions regarding the flow in the post-ireton aquifers is probably correct, but may be questionable for the pre-ireton aquifers because of his assumption regarding rock incompressibility and because in his analysis he did not consider the buoyancy effects of the very saline Lower and Middle Devonian formation waters. Hitchon et al. (1989, 1990) and Bachu and Underschultz (1993) reached similar conclusions with Toth (1978) regarding the division of the hydrostratigraphic succession into three major entities (pre-ireton, post-ireton to Clearwater, and post-clearwater), but differed from Toth (1978) mainly in the interpretation of the flow in the pre- Ireton aquifer systems. According to Hitchon (1989, 1990) and Bachu and Underschultz (1993), the flow in pre-ireton aquifers is regional, modified in places by geological features and buoyancy effects caused by enhanced salinity due to halite dissolution near the Prairie aquiclude. The flow in the pre- Cretaceous aquifer systems successively changes character from regional in the west to intermediate and local in the east because these aquifers become progressively shallower eastward. They are in contact with the Lower Mannville aquifer along the pre- Cretaceous unconformity as a result of a combination of pre-cretaceous erosion and salt dissolution of the Prairie aquiclude. The Upper Devonian aquifer, which crops out and is a discharge region north of these areas, exerts a drain effect on other aquifers in contact with it. Local flow systems dominate the flow in post-clearwater aquifers. Water salinity is low in Mesozoic aquifers and high in Paleozoic aquifers, but particularly high (more than 300,000 mg/l) in the vicinity of the evaporite-dominated Elk Point and Prairie aquicludes. In the southern part of the Alberta basin, Bachu et al. (1986) studied the flow in the entire Basal aquifer based on information from only 76 wells that reach this unit. Although evaporite beds are absent, the formation-water salinity is very high, in the 300,000 mg/l range. The flow of formation waters is generally eastward, with a slight northeastward turn toward the northern edge of the

10 1168 Flow of Formation Water, Alberta Basin (a) ROCKY MOUNTAINS FOOTHILLS INTERIOR PLAINS regional recharge regional lateral flow regional discharge local discharge local recharge local discharge fault Precambrian Basement (b) o o Tathlina Arch Canadian Shield 108 o W 62 o N 58 o 56 o 54 o 52 o 50 o 49 o 126 o 122 o Foreland Fold and Thrust Belt 118 o 114 o Bow Island Arch 49 N 108 o o 110o W Figure 6 Diagrammatic model of steady-state topography-driven flow of formation waters in the Alberta basin (after Hitchon, 1984): (a) dip cross section, and (b) plan view (modified). aquifer. The flow of formation waters in the post- Devonian succession (post Exshaw-Banff aquitard) was studied in the extreme southern part of the basin by Schwartz et al. (1981) and Toth and Corbet (1986), and in the entire southwestern part of the basin by Bachu and Underschultz (1995) (Figure 5), using about 16,000 drill-stem tests and 13,000 water analyses. Schwartz et al. (1981) noted

11 Bachu 1169 that in the extreme southern part of the basin the generalized flow direction in almost all the units is downdip northward from recharge areas occurring in Montana where these units crop out, and that the salinity of formation waters increases northward within the same unit. Isotope analyses (Schwartz et al., 1981) suggested that the waters are the product of long-term mixing of meteoric and so-called connate waters. Toth and Corbet (1986) distinguished three flow systems in the same area and hydrostratigraphic succession. According to them, the flow in the shallow post- Colorado succession is adjusted to the present topography, whereas the flow in the Colorado aquitard system is driven by erosional rebound of thick shales. Flow in Mississippian Jurassic and Mannville aquifer systems was considered by Toth and Corbet (1986) to be a relict (fossil) of a topography-driven flow system that existed prior to Pliocene Pleistocene erosion. According to Bachu and Underschultz (1995), the flow in the Mississippian Jurassic aquifer system is regional in nature. The flow is directed northward in southern Alberta from recharge areas in Montana. In southwestern and central Alberta recharge is from the southwest close to the fold and thrust belt, and the flow direction shifts northeastward in the general basin-scale direction suggested by Hitchon (1969a, b) for the flow in the Upper Devonian aquifer. The flow in the adjacent Mannville aquifers is influenced by the regional-scale flow in the Mississippian Jurassic aquifer system. In the southwestern part of the basin, flow in the Viking, Second White Speckled Sandstone, Belly River, and Horseshoe Canyon aquifers is driven by erosional rebound, and is directed inward, downdip, and westward toward the fold and thrust belt. The only exceptions to this general pattern are the Cardium and Milk River aquifers. Localized hydrostatic, subhydrostatic, and above-hydrostatic pressures are observed in distinct areas in the Cardium aquifer (Pendergast, 1969; Bachu and Underschultz, 1995). Flow in the Milk River aquifer is generally in equilibrium with the present topography (Schwartz et al., 1981; Toth and Corbet, 1986; Bachu and Underschultz, 1995). In conclusion, the flow of formation waters in the Alberta basin was studied to date for a very large area (approximately 500,000 km 2 ) and, depending on location, for all or significant portions of the sedimentary succession, using a huge amount of data (about 42,000 pressure determinations and 48,000 water analyses). The observed pressure distributions and flow patterns were variously attributed to topography, drain effects by highly permeable strata, chemical osmosis, fossil pressures, and erosional rebound. Various flow systems were identified in the basin, assumed to be either in equilibrium (steady state) or in the process of adjustment (transient) with the present boundary conditions. The image of the flow of formation waters in the basin evolved as new studies were performed; however, the interpretations are sometimes contradictory and lack a general, unifying view. In addition, the topography-driven flow model has been used to explain the observed geothermal regime in the basin (e.g., Majorowicz and Jessop, 1981; Hitchon, 1984), and the genesis of the Pine Point Pb-Zn ore deposit at the northern edge of the basin (Garven, 1985) and of the Athabasca oil sand deposit in the northeast (Garven, 1989). Based on previously published information and without examining new data, in the following section I attempt to reconcile previous interpretations into a basin-scale conceptional model of the flow of formation waters in the Alberta basin. SYNTHESIS MODEL OF FORMATION-WATER FLOW It is obvious from the previous review that the flow of formation waters in the Alberta basin is quite complex, changing character from one area and hydrostratigraphic interval to another. The various analyses and interpretations presented so far can be synthesized for the entire basin in the context of flow-driving mechanisms and processes at work in the basin. In general, the flow of formation waters in the basin is driven by hydraulic-head gradients caused both by elevation (topographic) differences and by pressure differences induced by erosional rebound. In places, buoyancy plays an important role mainly by retarding the flow of formation waters driven by hydraulic-head gradients. This is because the density of formation waters generally increases with depth, hence downdip, whereas the flow of formation waters is generally updip. For illustration, Figures 7 and 8 present the variation with depth of water salinity, temperature, and density in two wells, one ( W5M) located in the deep part of the basin toward the fold and thrust belt, and the other one ( W4M) situated toward the basin feather edge in the east where many hydrostratigraphic units are absent because of erosion at various times in the basin s history (see Figure 4 for location). The salinity and temperature profiles are based on measured and interpolated data, whereas the water density profiles were calculated from salinity and temperature using relationships published by Rowe and Chou (1970). The variation of salinity, and hence water density, along bedding in various aquifers is also significant (Bachu et al., 1986; Hitchon et al., 1990; Bachu and Underschultz, 1993, 1995). Thus, buoyancy cannot

12 1170 Flow of Formation Water, Alberta Basin T ( o C) Edmonton Figure 7 Variation with depth of aquifer-average formation-water salinity, temperature, and density in well W5M in the Alberta basin (for well location see Figure 4) Belly River Cardium 1400 Second White Speckled Ss. Depth (m) Viking Upper Mannville Clearwater Lower Mannville Jurassic Mississippian-Jurassic Exshaw - Banff Upper Devonian 2800 Ireton C (10 3 mg/l) ρ (kg/m 3 ) Middle - Upper Devonian Prairie Cambrian Basal be everywhere neglected as a mechanism playing an important role in the flow of formation waters in the Alberta basin (Bachu, 1995). However, chemical osmosis, although considered in the past to be an important driving mechanism for flow in Cretaceous aquifers (Hitchon, 1969b), is negligible because formation-water salinity in Mesozoic and Cenozoic aquifers does not differ significantly, being basically close to fresh water (Figures 7, 8) (Hitchon et al., 1990; Corbet and Bethke, 1992; Bachu and Underschultz, 1993, 1995). Two megahydrostratigraphic groups and associated flow systems can be distinguished at the scale of the Alberta basin: the pre-cretaceous and the post-jurassic, which correspond to units deposited, respectively, during the initial platformal phase of basin evolution and during the foreland-basin stage. The flow of formation waters in the two megahydrostratigraphic systems is presented diagrammatically in Figure 9. Post-Jurassic Hydrostratigraphic Group This succession is dominated by aquitards and aquitard systems consisting of thick shales alternating with sometimes isolated and thin sandstone aquifers. The salinity of formation waters is generally low, close to fresh water (Figures 7, 8) (Schwartz et al., 1981; Hitchon et al., 1990; Bachu and Underschultz, 1993, 1995). The mild increase of salinity with depth (Figures 7, 8) is probably due to increased solubility of various minerals with increased temperature (water-rock reactions). Nevertheless, the density variations are small, such

13 Bachu 1171 Depth (m) Temperature (C ) Colorado Viking Upper Mannville Clearwater Lower Mannville Upper Devonian Ireton Middle-Upper Devonian Figure 8 Variation with depth of aquifer-average formation-water salinity, temperature, and density in well W4M in the Alberta basin (for well location see Figure 4) Prairie 1200 Winnipegosis Elk Point C (10 mg/l) ρ (kg/m 3) that buoyancy effects most probably are negligible. The flow of formation waters in shallow aquifers is driven by local topography in local flow systems. In confined aquifers found at greater depths in the southwestern part of the basin near the fold and thrust belt, flow is driven by erosional pore-space rebound in the thick intervening shales (Neuzil and Pollock, 1983), and is in a transient stage of adjustment to the present topography and overburden. This is the case of the Viking, Second White Speckled Sandstone, Belly River, and Horseshoe Canyon aquifers in the Colorado and post-colorado systems (Toth and Corbet, 1986; Bachu and Underschultz, 1995). The erosional rebound-driven flow manifests itself at a regional scale in southwestern Alberta. The flow is west-southwest, downdip toward the fold and thrust belt, in a direction opposite to the regional topographic drop. Pressures reach values lower than the values corresponding to the lowest elevation in the basin, actually found km to the northeast at the Great Slave Lake. Toward the eastern edge of these aquifers, flow is oriented eastward, and in general is driven by pressure distributions in or close to equilibrium with the present relief (topographic drive). The flow in these areas is already adjusted to the new boundary conditions (topography) because these aquifers are shallower as a result of Tertiary to Holocene erosion. This flow pattern is consistent with Neuzil and Pollock s (1983) analysis, which showed that flow driven by erosional unloading is inward from the permeable boundaries. Although the rebound takes place in the thick shales, inward lateral flow occurs as a result in the intervening sandstone aquifers in a reverse process to the outward flow from compacting sand-shale successions noted by Magara (1976) and Bredehoeft et al. (1988). The different areal distributions of subhydrostatic pressures in these aquifers indicate spatial variability in shale thickness and hydraulic diffusivity, as well as permeability barriers within the aquifers themselves (Bachu and Underschultz, 1995). The only significant exceptions to this pattern are the Cardium and Milk River aquifers. Compartmentalized above-hydrostatic pressures in the Cardium aquifer (Pendergast, 1969; Bachu and Underschultz, 1995) most probably are caused by phase change of organic material, a process described by Hedberg (1974). Hydrostatic pressures in the Milk River aquifer (Schwartz et al., 1981; Toth and Corbet, 1986; Bachu and Underschultz, 1995) are due both to aquifer recharge at outcrop and to high permeability (hence diffusivity), which allowed relatively fast (on a geological time scale) adjustment to the current boundary conditions. Figure 9 shows diagrammatically the main features of the flow driven by topography and erosional rebound in the post- Jurassic succession in the Alberta basin. Pre-Cretaceous Hydrostratigraphic Group The flow of formation waters in this succession has different characteristics and is driven by different mechanisms and processes than in the post- Jurassic succession. The salinity of formation waters is high, regardless of relative depth (e.g., compare salinities at 800 and 1200 m depth in the two wells of Figures 7, 8). Only a small part of the salinity increase with depth is attributable to temperature-dependent water-rock reactions. Extremely high salinity in the vicinity of evaporitic beds can be explained by salt dissolution

14 1172 Flow of Formation Water, Alberta Basin W E Mixing and interference zone between flow systems in the pre-cretaceous and post-jurassic megahydrostratigraphic successions Flow normal to the plane of cross section 62 o 126 o 108 o W N 62 o Great Slave Lake Canadian Shield 4 Lake Athabasca Post-Jurassic hydrostatic succession 1. Local-scale topography-driven flow 2. Regional-scale flow driven by erosional rebound Pre-Cretaceous hydrostratigraphic succession Basin-scale south-northeastward topographydriven flow 4. Basin-scale flow of probable tectonic origin with strong buoyancy effects o 126 o Foreland Fold and Thrust Belt Approximate area of flow driven by erosional rebound o 49 o W N Figure 9 Diagrammatic model of the flow of formation waters in the Alberta basin: (a) dip cross section, and (b) plan view. (e.g., Winnipegosis aquifer in Figure 8) (Hitchon et al., 1990; Bachu and Underschultz, 1993); however, very high salinities were measured in water samples from all pre-cretaceous aquifers, even where evaporitic beds are absent (Figure 8) (Bachu et al., 1986; Hitchon et al., 1990; Bachu and Underschultz, 1993, 1995). The salinity in the pre- Cretaceous aquifers decreases significantly only

15 Bachu 1173 near outcrop in recharge areas in the south (Schwartz et al., 1981; Bachu and Underschultz, 1995) where meteoric water enters the system and mixes with so-called connate water (Schwartz et al., 1981), and in the east-northeast where these aquifers progressively subcrop at the pre- Cretaceous unconformity and mix with fresher waters from post-jurassic aquifers driven in local flow systems by local topography (Hitchon et al., 1990; Bachu and Underschultz, 1993). Even if pre- Cretaceous aquifers are in direct contact with post- Jurassic aquifers, fresh meteoric water cannot penetrate deeper into the former because of buoyancy effects that seem to be stronger than the topographic drive. This feature can be deduced by comparison of isosalinity contours at the pre- Cretaceous unconformity in the Peace River arch area (compare figures 6b; 7a, b, in Hitchon et al., 1990). The high salinity of formation waters in pre- Cretaceous aquifers indicates also that, unlike other basins (e.g., the Llanos basin in Colombia; Villegas et al., 1994), the passive-margin succession has been flushed only partially by meteoric water, and that this process is still taking place. No local systems are present in pre-cretaceous aquifers for four main reasons. First, in the southwestern part of the basin, no recharge can take place through the thick post-jurassic succession because of the sink effect in the thick Colorado and post-colorado shales undergoing erosional rebound. Thus, the pre-cretaceous aquifers are cut off from topographic recharge in this area. Second, buoyancy effects at the contact between pre- Cretaceous and post-jurassic systems oppose the descent of water flowing in local systems. Third, lateral drainage along the pre-cretaceous unconformity in the northeastern part of the basin does not allow extension of local, post-jurassic flow systems into pre-cretaceous aquifers (Toth, 1978; Bachu and Underschultz, 1993). Fourth, the geometry and basin-scale extent of the pre-cretaceous aquitards and aquicludes preclude penetration of local flow systems into the pre-cretaceous succession. Topography drives flow in Upper Devonian and Mississippian Jurassic aquifers in a south-northeastward basin-scale system, with recharge at outcrop in northern Montana at the southern edge of the basin, and discharge in northeastern Alberta (Hitchon, 1969a, b; Bachu and Underschultz, 1993, 1995). Pressures are subhydrostatic in most of the downstream parts of these aquifers; still, pressures are higher than those corresponding to the discharge point at the aquifer outcrop in northeastern Alberta (Hitchon, 1969a, b; Bachu and Underschultz, 1993, 1995). The flow in this system is similar to the deep topography-driven flow in the Denver and Palo Duro basins (Senger and Fogg, 1987; Belitz and Bredehoeft, 1988). The subhydrostatic pressures are due to the high permeability of mainly Upper Devonian Wabamun and Grosmont carbonates, particularly along subcrop on a south-northeast axis, which creates a drain effect (Hitchon, 1969a, b; Belitz and Bredehoeft, 1988). This south-northeastward topography-driven flow system is laterally fed in the southwestern, west-central, and northwestern parts of the basin by east-northeastward updip flow from the fold and thrust belt (Hitchon et al., 1990; Bachu and Underschultz, 1995). This east-northeastward flow from the fold and thrust belt cannot be driven by topography as previously thought because it is completely cut off from topographic recharge by the thick Colorado and post-colorado aquitard systems. Similar east-northeastward flow has been noted in the deeper Middle-Upper Devonian, Winnipegosis, and Basal aquifers (Bachu et al., 1986; Hitchon et al., 1990; Bachu and Underschultz, 1993). Toth (1978) and Toth and Millar (1983) hypothesized that in the Red Earth region this could be the expression of a relict (fossil) flow system in a transient stage of adjustment from pre-pliocene to present topography. Neuzil (1985) argued against the assumptions made by Toth and Millar (1983) to demonstrate the possibility of fossil flow, whereas Corbet and Bethke (1992) showed numerically that the influence of past topography would have rapidly dissipated. Moreover, the hypothesis stated in Toth (1978) and Toth and Millar (1983) seems highly improbable because of the very large (basin) scale of this flow pattern and because it takes place in various aquifer systems with no apparent recharge areas (e.g., the Basal, Winnipegosis, and Middle-Upper Devonian aquifers do not subcrop or crop out except at discharge in the northeast). Another possibility would be that these flow systems are recharged through fault systems in the Rocky Mountains beyond the fold and thrust belt; however, this is also highly improbable because (1) the salinity of formation waters should be much lower for a system recharged by meteoric water; (2) the fault system should be continuous and open for hundreds of kilometers along the fold and thrust belt and mountain range; and (3) some aquifers, such as the Winnipegosis, do not reach the fold and thrust belt (Figure 4). Isotope analyses (Hitchon and Friedman, 1969) showed that hot spring waters in the Canadian Rockies have a meteoric origin, supporting Hitchon s (1969a) hypothesis that the groundwater flow in the Rocky Mountains has a cellular pattern and that flow systems in the mountains discharge locally. The meteoric-water circulation cells probably do not penetrate below the thrust sheets to feed regional-scale flow systems in the Alberta basin. Recent isotope analyses show indeed that deep penetration of surface waters into the subsurface has not occurred in the western and main ranges of the Rocky Mountains and that west-east

16 1174 Flow of Formation Water, Alberta Basin flow occurred before initiation of thrusting in the Rockies (Nesbitt and Muehlenbachs, 1993a, b). The most probable initial flow-driving mechanism and origin of these high-salinity northeastward-flowing waters is past tectonic compression along the western basin margin. Tectonic processes can expel brines from deep strata under thrust sheets into the basin, as discussed by Oliver (1986), at rates estimated (Bethke and Marshak, 1990) and calculated numerically (Ge and Garven, 1989; Deming et al., 1990) on the order of 10 2 m/yr. Similar rates were estimated using dimensional analysis for the flow in these aquifers in the Cold Lake and Peace River arch areas (Bachu, 1985, 1988; Bachu and Cao, 1992). Although the excess pressure generated by tectonic compression appears to dissipate in yr (Ge and Garven, 1989), the residence time for the tectonically expelled waters is much larger (i.e., it would take 5 50 m.y. for a fluid particle to travel 500 km, with the calculated and inferred velocities of 1 10 cm/yr). The difference between dissipation and residence time is that dissipation depends on permeability and compressibility, whereas residence time depends on permeability only. Figure 9 shows diagrammatically the main features of flow in the pre-cretaceous succession in the Alberta basin. As mentioned previously, the density of formation waters in the pre-cretaceous succession is variable, generally increasing with depth. As a result, buoyancy plays a role not in driving the flow, but rather in opposing (retarding) the flow. This is because the updip hydraulic drive is opposed by the downdip increase in formation-water density. Thus, buoyancy effects cannot generally be neglected when analyzing the flow of formation waters in these aquifers (Bachu, 1995). Several shaly aquitards and evaporitic aquicludes are found in the pre-cretaceous succession; however, no erosional rebound-driven flow is apparent for the aquifers in the pre-cretaceous succession for several reasons. Some of these shale aquitards, such as the Exshaw-Banff and Jurassic, are relatively thin compared to the adjacent aquifers and thus have limited available pore space. The shales in this succession underwent, depending on depth, at least two and as many as five loading-unloading cycles. Given the hysteretic character of shale hydraulic properties with varying stress (Neuzil, 1993), the porosity of these shales decreased after each cycle, such that no more rebound could have taken place as a result of Tertiary to Holocene erosion, unlike shales in the Cretaceous Tertiary strata, which underwent only one loading event and are in the middle of the unloading part of the first cycle. Also, with the exception of the Basal aquifer, the aquifers adjacent to these pre-cretaceous shales are thick, have generally high permeability, and are in communication at subcrop. Thus, any disequilibrium pressure caused in the past by loading and unloading would have rapidly propagated throughout the system, reaching equilibrium as a result of high hydraulic diffusivity. The pre-cretaceous and post-jurassic megaflow systems interact at their contact. In the southern part of the basin, the entire Mannville Group is a single aquifer system, with shale lenses present throughout. In the central part of the basin, the Clearwater aquitard divides the Mannville strata into the Lower and Upper Mannville aquifers. In the northern part of the basin, depositional facies changes mean that the Upper Mannville is an aquitard. The Mannville aquifers overlie the pre- Cretaceous hydrostratigraphic succession, coming progressively in contact eastward with the Mississippian Jurassic, Upper Devonian, and Middle- Upper Devonian aquifer systems. As a general observation then, and throughout the Alberta basin, the flow of formation waters in aquifers in the Mannville Group is influenced by both overlying topographydriven local flow systems and by the regional-scale flow in underlying aquifer systems (Toth, 1978; Hitchon et al., 1990; Bachu and Underschultz, 1993, 1995). As mentioned previously, mixing of fresh meteoric water with deep basinal brines takes place at the contact between the (Lower) Mannville aquifer and underlying aquifer systems. Regarding cross-formational flow, the thickness and very low permeability of the pre-cretaceous aquitards and aquicludes, combined with increased formationwater density with depth and in the vicinity of evaporitic beds, impede cross-formational flow in this succession, probably significantly. Some cross-formational flow is present between the Middle-Upper Devonian and Upper Devonian aquifer systems in places where carbonate reef complexes breach through the intervening Ireton aquitard (Toth, 1978; Bachu and Underschultz, 1993). Cross-formational flow in the post-jurassic succession is present in areas where topography drives the flow (Toth, 1978; Hitchon et al., 1990; Bachu and Underschultz, 1993), but is absent in the southwestern areas where the flow is driven by erosional pore-space rebound in the intervening aquitards. IMPLICATIONS The previously described model of formationwater flow in the Alberta basin shows a very complex, three-dimensional flow pattern and structure. The flow changes character across the basin, and no single mechanism or process is dominant at the basin scale. Various flow-driving mechanisms and processes are active in different parts of the basin (both areally and stratigraphically). The complex flow pattern has implications in several domains.

17 Bachu 1175 Hydrocarbon Migration and Accumulation After primary expulsion from source beds, hydrocarbons migrate along bedding in more permeable strata (aquifers), driven by their own buoyancy and entrained by the flow of formation waters (Hubbert, 1940, 1953). The present model of formation-water flow indicates that the Athabasca and Cold Lake oil sand deposits could not have been formed as a result of basin-scale topography-driven flow of formation waters from southwestern and western Alberta to northeastern Alberta, as conceptualized by Hitchon (1984) and numerically simulated by Garven (1989). Specifically, from the migration point of view, the Cretaceous shales of southwestern Alberta cannot be the source of the oil that is now found as biodegraded oil in the Cold Lake and Athabasca areas. The downdip erosional rebound-driven flow in the Colorado and post- Colorado aquitard systems could not have driven the hydrocarbons updip toward the basin feather edge. Hydrocarbons generated in Cretaceous shales in southwestern Alberta are hydrodynamically trapped in the intervening, isolated aquifers. Regarding the Athabasca and Cold Lake oil sand deposits, their source is probably in Jurassic and Devonian shales, based on possible migration paths alone. Hydrocarbons migrated updip in the Middle- Upper Devonian and Devonian aquifers in the pre- Cretaceous flow system until they reached the pre- Cretaceous unconformity and were trapped in stratigraphic and structural traps in Mannville strata, where they were subsequently biodegraded in place, as found today. It is significant that both the Athabasca and Cold Lake oil sand deposits are found in Mannville strata at the contact with Middle-Upper Devonian and Upper Devonian aquifer systems. This hydrodynamic model of water flow in the Alberta basin supports and in turn is supported by the geochemical analysis of source rocks and oils in the Western Canada sedimentary basin (Creaney and Allan, 1990; Allan and Creaney, 1991), which shows that only Mississippian-, Jurassic-, and Mannville-generated oils have contributed to the Athabasca and Cold Lake oil sand deposits. Terrestrial Heat Flow The steady-state, topography-driven, basin-scale flow model of Hitchon (1969a, b) (Figure 6) was used by Majorowicz and Jessop (1981, 1993) and Hitchon (1984) to explain the observed geothermal pattern in the Alberta basin. According to this model, low geothermal gradients and heat flow in southwestern Alberta are due to cooling by meteoric water recharge, whereas high geothermal gradients and heat flow in northern and northeastern Alberta are caused by discharge of warm water carrying heat in an advective flow system. Based on dimensional analysis and flow patterns, Bachu (1985, 1988), Bachu and Burwash (1991), and Bachu and Cao (1992) argued that the flow of formation waters in the basin is too slow (due to low permeability) to influence the geothermal field significantly and that the geothermal regime in the basin is dominated by conduction, shown also by the numerical simulations of Corbet and Bethke (1992). Using numerical modeling of brine migration by topographically driven recharge, Deming and Nunn (1991) showed that a basin would be completely flushed of brines by meteoric water should the flow field be strong enough to carry heat from recharge to discharge areas. This obviously is not the case for the Alberta basin. The present flow model suggests that actually no basin-scale recharge takes place in southwestern Alberta, where the Colorado and post-colorado aquitard systems act as sinks, and that topographydriven flow takes place generally in local flow systems. Thus, the fluid-flow model in the Alberta basin reinforces the conductive heat-flow model of the basin, with direct application in the study of basin thermal history and hydrocarbon generation. Ore Genesis Garven (1985) showed numerically that, given the right combination of rock and water properties, regional-scale topography-driven flow of formation waters could have formed the Pb-Zn Pine Point ore deposit at the northern edge of the Alberta basin. Beside the fact that Garven s (1985) model assumed unrealistically high permeability values (Fowler, 1986), the present flow model shows that the formation waters in the Winnipegosis aquifer (named Keg River in the area) have no meteoric origin and the flow was and is not driven by topography. Rather, these brines either originated in Paleozoic shales of the western North American margin and migrated to the east (Nesbitt and Muehlenbachs, 1993a, b), or have a deep crustal origin and their flow was driven by tectonic processes (Hitchon, 1993). Deep Waste Disposal Because of the very low velocity of formation waters in the Alberta basin, particularly in pre- Cretaceous aquifers (about 10 2 m/yr; Bachu, 1985, 1988), liquid wastes injected into deep aquifers will be hydrogeologically trapped in the basin for geological-scale periods of time (Bachu et al., 1994). The subhydrostatic pressures and downdip

18 1176 Flow of Formation Water, Alberta Basin flow generated in Colorado and post-colorado aquifers by erosional rebound in the adjacent shales offer another mechanism for geological timescale underground waste isolation in the Alberta basin, as noted previously in a generic sense by Neuzil and Pollock (1983). Wastes disposed of in these deep aquifers will be practically captive until the system reaches equilibrium with the evolving topography, after which a slow, outward flow will start with velocities comparable with the ones in underlying aquifers. CONCLUSIONS Fluid flow in sedimentary basins is driven mainly by hydraulic-head gradients and buoyancy of formation waters. Although osmotic processes may be present, they are generally negligible in comparison with the other two mechanisms. Hydraulichead gradients may result from differences in elevation (topography-driven flow) or from changes in the fluid and pore volume. The latter can be the result of mechanical processes like loading during compaction, erosional unloading, and tectonic compression, or of thermal and physico-chemical processes like phase changes and mineral dissolution, precipitation, and dehydration. Of these, mechanical processes seem to be dominant in inducing nonhydrostatic pressures and driving the flow of formation waters. No single mechanism and process is fully responsible for fluid flow in a sedimentary basin, although a single mechanism can be dominant in various places and at different times. Because the Alberta basin in western Canada is in a mature stage of exploration, a wealth of data exists and is available in the public domain. Several studies have been performed in various parts of the basin, each analyzing and describing the flow of formation waters in the respective area and stratigraphic succession. These studies were critically reviewed in this paper in the context of flow-driving mechanisms and processes, and synthesized in a new model of the flow of formation waters in the Alberta basin. Based on flow characteristics and driving mechanisms, and on formation-water characteristics, two megahydrostratigraphic successions were identified in the basin: pre-cretaceous and post-jurassic. The pre-cretaceous succession corresponds to the initial platformal phase in the basin s development and consists of carbonate-dominated aquifer systems separated by shaly aquitards and evaporite- (halite-) dominated aquicludes. The salinity, hence density of formation waters, increases generally with depth, reaching more than 300,000 mg/l in very deep aquifers and in the vicinity of evaporitic beds. Topography drives flow in a basin-scale system in the Upper Devonian and Mississippian Jurassic aquifer systems, from recharge in Montana in the south to discharge in northeastern Alberta. High permeability of rocks in the Upper Devonian aquifer system creates a drain effect on flow. East-northeastward regional-scale flow systems in the pre-cretaceous aquifers and aquifer systems are most probably the expression of past tectonic processes in the west. Salinity generally increases both with depth and northward. The increase with depth is partly explained by increased water-rock reactions with increased temperature. The difference between pre-cretaceous and post-jurassic waters and the northward increase in salinity is explained by a different origin of formation waters and incomplete flushing and mixing with meteoric water. Because of salinity variations, buoyancy probably plays an important role, opposing and retarding the flow induced by other processes. The post-jurassic succession corresponds to the foreland phase of basin evolution and consists of thick shale-dominated aquitards and sandstonedominated aquifers. In the southwestern part of the basin, where the sedimentary succession is thickest, the flow in isolated, relatively thin aquifers embedded in the thick Colorado and post- Colorado aquitards is driven westward downdip, toward the fold and thrust belt, by erosional rebound in the intervening shales. Elsewhere in the basin, where much of the succession has been removed by Tertiary to Holocene erosion, the formation waters flow in local systems, driven by local topographic differences. The salinity of formation waters in the post-jurassic succession is low, such that buoyancy effects are most probably negligible. In topography-driven flow systems the water is of meteoric origin. In rebound-driven flow systems, the water has no meteoric component. Cross-formational flow in the pre-cretaceous succession is probably impeded by both buoyancy effects and the thick intervening aquicludes and aquitards. Some cross-formational flow probably takes place between the Middle-Upper Devonian and Upper Devonian aquifer systems where reef complexes breach the intervening Ireton aquitard. In the post-jurassic succession, cross-formational flow takes place in areas where the flow is driven by topography, but does not exist in the southwest where the flow is driven by shale erosional rebound. Mixing and interference between flow systems in the pre-cretaceous and post-jurassic hydrostratigraphic successions take place along the pre-cretaceous unconformity where pre- Cretaceous aquifer systems subcrop and are overlain by the Cretaceous Mannville aquifers.

19 Bachu 1177 This model of formation-water flow in the Alberta basin has implications for understanding the (1) origin, migration, and accumulation of hydrocarbons, particularly heavy oils and oil sand deposits in northeastern Alberta; (2) geothermal regime and thermal history and maturation of the basin; (3) genesis of the Pine Point Pb-Zn ore deposit in the Northwest Territories; and (4) optimum, environmentally safe, deep disposal of toxic wastes. REFERENCES CITED Allan, J., and S. Creaney, 1991, Oil families of the Western Canada basin: Bulletin of Canadian Petroleum Geology, v. 39, p Bachu, S., 1985, Influence of lithology and fluid flow on the temperature distribution in a sedimentary basin: a case study from the Cold Lake area, Alberta, Canada: Tectonophysics, v. 120, p Bachu, S., 1988, Analysis of heat transfer processes and geothermal pattern in the Alberta basin, Canada: Journal of Geophysical Research, v. 93, no. B7, p Bachu, S., 1995, Flow of variable-density formation water in deep sloping aquifers: review of methods of representation with case studies: Journal of Hydrology, v. 164, p Bachu, S., and R. A. Burwash, 1991, Regional-scale analysis of the geothermal regime in the Western Canada sedimentary basin: Geothermics, v. 20, p Bachu, S., and S. Cao, 1992, Present and past geothermal regimes and source rock maturation, Peace River arch area, Canada: AAPG Bulletin, v. 76, p Bachu, S., and J. R. Underschultz, 1993, Hydrogeology of formation waters, northeastern Alberta basin: AAPG Bulletin, v. 77, p Bachu, S., and J. R. Underschultz, 1995, Large-scale erosional underpressuring in the Mississippian Cretaceous succession, southwestern Alberta basin: AAPG Bulletin, v. 79, p Bachu, S., B. Hitchon, and P. Mortensen, 1986, Preliminary analysis of transport processes in the basal Cambrian aquifer of south-central Alberta, in B. Hitchon, S. Bachu, and C. M. Sauveplane, eds., Hydrogeology of sedimentary basins: application to exploration and exploitation: National Water Well Association, Dublin, Ohio, p Bachu, S., W. D. Gunter, and E. H. Perkins, 1994, Aquifer disposal of CO 2 : hydrodynamic and mineral trapping: Energy Conversion and Management, v. 35, p Belitz, K., and J. D. Bredehoeft, 1988, Hydrodynamics of the Denver basin: an explanation of subnormal fluid pressures: AAPG Bulletin, v. 72, p Bethke, C. M., 1985, A numerical model of compaction-driven groundwater flow and heat transfer and its application to the paleohydrology of intracratonic sedimentary basins: Journal of Geophysical Research, v. 80, p Bethke, C. M., 1986, Inverse hydrologic analysis of the distribution and origin of Gulf Coast type geopressured zones: Journal of Geophysical Research, v. 81, p Bethke, C. M., and S. Marshak, 1990, Brine migrations across North America the plate tectonics of groundwater: Annual Reviews Earth Planetary Sciences, v. 18, p Bradley, J. S., 1975, Abnormal formation pressure: AAPG Bulletin, v. 59, p Bredehoeft, J. D., and B. B. Hanshaw, 1968, On the maintenance of anomalous fluid pressures, 1. Thick sedimentary sequences: Geological Society of America Bulletin, v. 79, p Bredehoeft, J. D., R. D. Djevanshir, and K. R. Belitz, 1988, Lateral fluid flow in a compacting sand-shale sequence, south Caspian basin: AAPG Bulletin, v. 72, p Bustin, R. M., 1992, Organic maturation of the Western Canadian sedimentary basin: International Journal of Coal Geology, v. 19, p Corbet, T. F., and C. M. Bethke, 1992, Disequilibrium fluid pressures and groundwater flow in the Western Canada sedimentary basin: Journal of Geophysical Research, v. 97, no. B5, p Creaney, S., and J. Allan, 1990, Hydrocarbon generation and migration in the Western Canada sedimentary basin, in J. Brooks, ed., Classic petroleum provinces: London, Geological Society Special Publication 50, p Deming, D., and J. A. Nunn, 1991, Numerical simulations of brine migration by topographically driven recharge: Journal of Geophysical Research, v. 96, no. B2, p Deming, D., J. A. Nunn, and D. G. Evans, 1990, Thermal effects of compaction-driven ground water flow from overthrust belts: Journal of Geophysical Research, v. 95, p Dickey, P. A., and W. C. Cox, 1977, Oil and gas in reservoirs with subnormal pressures: AAPG Bulletin, v. 61, p Fowler, A. D., 1986, The role of regional fluid flow in the genesis of the Pine Point Deposit, Western Canada sedimentary basin a discussion: Economic Geology, v. 81, p Garven, G., 1985, The role of regional fluid flow in the genesis of the Pine Point Deposit, Western Canada sedimentary basin: Economic Geology, v. 80, p Garven, G., 1989, A hydrogeologic model for the formation of the giant oil sands deposits of the Western Canada sedimentary basin: American Journal of Science, v. 289, p Ge, S., and G. Garven, 1989, Tectonically induced transient groundwater flow in foreland basins, in A. E. Beck, G. Garven, and L. Stegena, eds., The origin and evolution of sedimentary basins and their energy and mineral resources: American Geophysical Union Geodynamics Series Monograph 48, p Hanshaw, B. B., and J. D. Bredehoeft, 1968, On the maintenance of anomalous fluid pressures, 2. Source layer at depth: Geological Society of America Bulletin, v. 79, p Hedberg, H. D., 1974, Relation of methane generation to undercompacted shales, shale diapirs, and mud volcanoes: AAPG Bulletin, v. 58, p Hill, G. A., W. A. Colburn, and J. W. Knight, 1961, Reducing oilfinding costs by use of hydrodynamic evaluations, in Economics of petroleum exploration, development, and property evaluation: Proceedings of the 1961 Institute of the International Oil and Gas Educational Center, Englewood, New Jersey, Prentice-Hall, p Hitchon, B., 1969a, Fluid flow in the Western Canada sedimentary basin, 1. Effect of topography: Water Resources Research, v. 5, p Hitchon, B., 1969b, Fluid flow in the Western Canada sedimentary basin, 2. Effect of geology: Water Resources Research, v. 5, p Hitchon, B., 1984, Geothermal gradients, hydrodynamics, and hydrocarbon occurrences, Alberta, Canada: AAPG Bulletin, v. 68, p Hitchon, B., 1993, Geochemistry of formation waters, northern Alberta, Canada: their relation to the Pine Point ore deposit: Alberta Research Council Open File Report , 99 p. Hitchon, B., and I. Friedman, 1969, Geochemistry and origin of formation waters in the Western Canada sedimentary basin, I. Stable isotopes of hydrogen and oxygen: Geochimica et Cosmochimica Acta, v. 33, p Hitchon, B., S. Bachu, C. M. Sauveplane, A. Ing, A. T. Lytviak, and J. F. Underschultz, 1989, Hydrogeological and geothermal regimes in the Phanerozoic succession, Cold Lake area, Alberta and Saskatchewan: Alberta Research Council Bulletin 59, 84 p. Hitchon, B., S. Bachu, and J. R. Underschultz, 1990, Regional subsurface hydrogeology, Peace River arch area, Alberta and British Columbia: Bulletin of Canadian Petroleum Geology, v. 38A, p Hubbert, M. K., 1940, Theory of ground-water motion: Journal of Geology, v. 48, p

20 1178 Flow of Formation Water, Alberta Basin Hubbert, M. K., 1953, Entrapment of petroleum under hydrodynamic conditions: AAPG Bulletin, v. 37, p Hunt, J. H., 1990, Generation and migration of petroleum from abnormally pressured fluid compartments: AAPG Bulletin, v. 74, p Kharaka, Y. K., and F. A. F. Berry, 1974, The influence of geological membranes on the geochemistry of subsurface waters from Miocene sediments at Kettleman North Dome in California: Water Resources Research, v. 10, p Magara, K., 1975, Reevaluation of montmorillonite dehydration as cause of abnormal pressure and hydrocarbon migration: AAPG Bulletin, v. 59, p Magara, K., 1976, Water expulsion from clastic sediments during compaction directions and volumes: AAPG Bulletin, v. 60, p Majorowicz, J. A., and A. M. Jessop, 1981, Regional heat flow patterns in the Western Canadian sedimentary basin: Tectonophysics, v. 74, p Majorowicz, J. A., and A. M. Jessop, 1993, Relation between basement heat flow and thermal state of the sedimentary succession of the Alberta Plains: Bulletin of Canadian Petroleum Geology, v. 41, p Marine, I. W., and S. J. Fritz, 1981, Osmotic model to explain anomalous hydraulic heads: Water Resources Research, v. 17, p Nesbitt, B. E., and K. Muehlenbachs, 1993a, Synorogenic fluids of the Rockies and their impact on paleohydrogeology and resources of the Western Canadian sedimentary basin, in G. M. Ross, ed., Alberta basement transect workshop: LITHO- PROBE Report 31, LITHOPROBE Secretariat, University of British Columbia, p Nesbitt, B. E., and K. Muehlenbachs, 1993b, Crustal hydrogeology of the Rockies: implications to the origins of brines and Pb-Zn mineralization in the Western Canadian sedimentary basin: Geological Association of Canada Annual Meeting, Program with Abstracts, p. 18. Neuzil, C. E., 1985, Comment on possible effects of erosional changes of the topographic relief on pore pressures at depth by J. Toth and R. F. Millar: Water Resources Research, v. 21, p Neuzil, C. E., 1986, Groundwater flow in low-permeability environments: Water Resources Research, v. 22, p Neuzil, C. E., 1993, Low fluid pressure within the Pierre Shale: a transient response to erosion: Water Resources Research, v. 29, p Neuzil, C. E., and D. W. Pollock, 1983, Erosional unloading and fluid pressures in hydraulically tight rocks: Journal of Geology, v. 91, p Nurkowski, J. R., 1984, Coal quality, coal rank variation and its relation to reconstructed overburden, Upper Cretaceous and Tertiary plains coals, Alberta, Canada: AAPG Bulletin, v. 68, p Oliver, J., 1986, Fluids expelled tectonically from orogenic belts, their role in hydrocarbon migration and other geologic phenomena: Geology, v. 14, p Pendergast, R. D., 1969, Correlating Cardium stratigraphy using static bottom hole pressures: Canadian Well Logging Society Journal, v. 2, p Phillips, F. M., 1983, Comment on Chemical osmosis, reverse chemical osmosis, and the energy of subsurface brines by D. L. Graf: Geochimica et Cosmochimica Acta, v. 47, p Porter, I. W., R. A. Price, and R. G. McCrossan, 1982, The Western Canada sedimentary basin: Philosophical Transactions of Royal Society of London, Series A, v. 305, p Ricketts, B. D., ed., 1989, Western Canada sedimentary basin a case history: Calgary, Canadian Society of Petroleum Geologists, 320 p. Rowe, A. M., and J. C. S. Chou, 1970, Pressure-volume-temperature-concentration reaction of aqueous NaCl solutions: Journal of Chemical Engineering Data, v. 15, p Schwartz, F. W., K. Muehlenbachs, and D. W. Chorley, 1981, Flowsystem controls of the chemical evolution of groundwater: Journal of Hydrology, v. 54, p Senger, R. K., and G. E. Fogg, 1987, Regional underpressuring in deep brine aquifers, Palo Duro basin, Texas, 1. Effects of hydrostratigraphy and topography: Water Resources Research, v. 23, p Senger, R. K., C. W. Kreitler, and G. E. Fogg, 1987, Regional underpressuring in deep brine aquifers, Palo Duro basin, Texas, 2. The effect of Cenozoic basin development: Water Resources Research, v. 23, p Shi, Y., and C. Y. Wang, 1986, Pore pressure generation in sedimentary basins: overloading versus aquathermal: Journal of Geophysical Research, v. 91, no. B2, p Toth, J., 1963, A theoretical analysis of groundwater flow in small drainage basins: Journal of Geophysical Research, v. 68, p Toth, J., 1978, Gravity-induced cross-formational flow of formation fluids, Red Earth region, Alberta, Canada: analysis, patterns, and evolution: Water Resources Research, v. 14, p Toth, J., and T. Corbet, 1986, Post-Paleocene evolution of regional groundwater flow-systems and their relation to petroleum accumulations, Taber area, southern Alberta, Canada: Bulletin of Canadian Petroleum Geology, v. 34, p Toth, J., and R. F. Millar, 1983, Possible effects of erosional changes of the topographic relief on pore pressures at depth: Water Resources Research, v. 19, p Villegas, M. E., S. Bachu, J. C. Ramon, and J. R. Underschultz, 1994, Flow of formation waters in the Cretaceous Miocene succession of the Llanos basin, Colombia: AAPG Bulletin, v. 12, p ABOUT THE AUTHOR Stefan Bachu Stefan Bachu received an engineering degree in hydraulics, an M.Sc. degree in hydrogeology, and a Ph.D. in transport processes from the Technion-Israel Institute of Technology. After doing research at Cornell University on a Lady Davis Fellowship, he joined the Alberta Research Council (ARC) in Canada in In 1987, he became head of the petroleum geoscience section, Alberta Geological Survey, ARC. Since April 1995 he has been senior advisor in the Alberta Geological Survey of the Alberta Department of Energy. His research interests include fluid flow, heat transfer and transport in sedimentary basins, modeling basin evolution, and reservoir analysis.