Walker 1990

PERSPECTIVE FACIES MODELING

AND SEQUENCE

STRATIGRAPHY 1

ROGER G. WALKER Department of Geology McMaster University Hamilton, Ontario L8S 4MI, Canada From my perspective, it seems useful to think again about the relationship between sedimentology and stratigraphy, or more specifically, between aspects of facies, architectural elements, and facies modeling (Walker 1984a; Miall 1985) and the new genetic stratigraphic schemes (sequence stratigraphy, genetic stratigraphic sequences, allostratigraphy). Smith (1985) suggested that a JSP Perspective might he "opinion-oriented," challenge "existing viewpoints" and "take a hard l o o k . . , at some popular model and conventional wisdom." I will attempt to do all of the above, showing (not surprisingly) that many basic ideas have been recycled over the years and clothed in new terminology. My ideas have been stimulated particularly by Plint's work on Cardium Formation erosion surfaces (PEnt et al. 1986), Bhattacharaya's work on sequences in the Dunvegan Formation (Bhattacharaya 1988), the work of several people from Exxon on sequence stratigraphy (Posamentier and Vail 1988; Posamentier et al. 1988), and a commentary on the Exxon work by Galloway (1989). Sequence stratigraphy, as presented by the Exxon group (Van Wagoner et al. 1988; Posamentier and Vail 1988; Posamentier et al. 1988), is a theoretical concept which was introduced without specific worked-out examples. I will apply their ideas in some detail to a Cretaceous situation in order to examine whether interpretations are driven by the theoretical model, or whether specific situations highlight weaknesses in the model (and thus can be used as feedback, with a view to strengthening the model in the real world of applications). THE STRATIGRAPHICRECORD Mother Nature presents us with layer upon layer of stratified r o c k - - m o r e than we can comprehend at once. In the most general sense, stratigraphy involves the subdivision of these rocks into bite-sized pieces. In a measured section, it is obvious that each measured piece is in some way different from the pieces above and below. These pieces of different aspect are in effect facies (Walker 1984b); each piece is separated from the next by a sharp or gradational contact. These simple pieces can be reassembled in various ways, serving the needs of correlation, sedimentological interpretation, facies modeling, study of relative sea-level fluctuations, etc. The bite-sized pieces are normally descriptive, but their re-assembly usually involves interpretation. ' Manuscript received 30 November 1989; revised 22 January 1990.

Subdivision I have discussed at length the problems involved in the scale of facies subdivision (Walker 1984b, p. 1-2), assuming that a different scheme appropriate for each geological situation would be created by each worker (here termed a local scheme, as in Fig. 1; see de Raaf et al. 1965 and Walker 1983 for examples). However, there are now two facies schemes that are being applied universally; the turbidite scheme o f Mutti and Ricci-Lucchi (1972) and the fluvial scheme of Miall (1978). Universal schemes ideally reflect our collective increased understanding of the deposits of certain environments, and universal schemes for other environments are probably not far away; the disadvantage of such schemes is the possibility of "force-fitting" particular local situations into the scheme. Regardless of whether an existing universal scheme is used or an appropriate local scheme devised, we subdivide the stratigraphic record into a collection of units (facies) separated by sharp or gradational contacts (Fig. 1; de R a a f e t al. 1965). Contacts may be sharp for purely sedimentological reasons (e.g., fluvial crevasse splay sandstones on floodplain mudstones) or because of major changes in depositional environment caused, for example, by relative sea level fluctuations. The significance accorded the contacts is one of the main problems in stratigraphy and sedimentology. Re-assembling the Stratigraphic Pieces In many local studies, facies descriptions may be so complex that they go beyond our interpretive abilities. The first way to simplify a complex scheme is to group facies perceived to be similar and/or genetically related; this can be done by using local criteria or by reference to an existing universal scheme. A good example of a local facies scheme for deltaic sediments is that of Collinson (1969). He grouped 14 facies into five interpretive associations which consisted of "groups of facies genetically related to one another and which have some environmental significance" (Collinson 1969, p. 207). This concept o f facies associations has been recycled into the more general idea of architectural elements (Allen 1983), discussed below. Individual facies can also be re-assembled into existing universal associations, such as the slope, fan, and basin floor associations of Mutti and Ricci-Lucchi (1972) or the fluvial architectural elements of Allen (1983). The strength of recognizing facies associations or architectural elements is that each facies may now be placed in context

JOURNAL OV SEDnde~rARY P e ' r R o ~ v , VOL. 60, No. 5, SeYre~meR, 1990, r. 777-786 Copyright © 1990, SEPM (Society for Sedimentary Geology) 0022-4472/90/0060-777/$03.00

ROGER G. WALKER

LAGOONAL CLAM BEDS PARALLEL LAM. SWALEY CROSS STRAT.

HCS SANDSTONES INTERBEDDED WITH BIOTURBATED MUDSTONES

WAVE RIPPLED V. FINE SANDSTONES

BIOTURBATED MUDSTONES

FIG. 1.- Hypothetical vertical facies sequence, with interpretations suggested to right.

with others perceived to be genetically related. Each facies therefore contributes to the interpretation of the others. FACIES SUCCESSIONS

The term "sequence" now appears to be used in a large scale context, that of seismic sequences and sequence stratigraphy. It was first used in a large-scale sense by Sloss (1963) for the six craton-wide stratigraphic sequences of the North American continent. However, Teichert (1958, p. 2723) points out that "in the German literature successions of vertical facies became known as Faziesreihen . . . this concept will be known in this paper as facies sequence" (Teichert's italics). Teichert does not specify a scale, but it appears to be closer to that of an outcrop rather than an entire craton. Thus Sloss has changed the meaning of a pre-existing useful term; however, because of current usage, I reluctantly cede the term "sequence" to the seismic and sequence stratigraphers and will define it later. Thus what used to be termed "facies sequences" will hereafter be called "facies successions." Facies can be placed in relative context by the recognition of progressive facies successions-those in which some or many rock properties change systematically upsection. In Figure 1, most of the facies contacts are shown as gradational except for the base of the shoreface, which is shown as sharp and loaded (e.g., McCrory and Walker 1986; Rosenthal and Walker 1987).

The building of vertical successions has been discussed by Johannes Walther (in Middleton 1973) and more recently emphasized by de Raaf et al. (1965) and Visher (1965). The ideas were reiterated by Busch (197 1) under the heading of "genetic increments of strata" (GIs). These were defined as "intervals of strata representing one cycle of sedimentation in which each lithological component ["facies" in my terminology] is related genetically to all others." Busch's GIs take on a genetic implication that was not present in the facies successions of de Raaf et al. (1965). It is implicit (de Raaf et al. 1965) or explicit (Busch 1971) that any progressive facies succession with dominantly gradational facies contacts is the result of one related set of depositional conditions. Experience has shown that in a broad sense there are only a limited number of progressive facies successions. Historically, the comparison of many different examples of one type of succession (ancient and modern) has led to generalizations that form the bases of facies models. FACIES MODELS

The term "facies model" continues to give trouble; it is used in the literature both for a summary of a local situation and for a conscious attempt at generalization using the combined features of many local examples (Walker 1984b). Both usages are well established; I prefer the generalization and will use the term in this way

PERSPECTIVE throughout the paper. G. V. Middleton (pers. comm., 1990) has questioned some of the purposes of facies modeling; stimulated by his comments, I suggest that the process of facies modeling is useful as a deliberate attempt to synthesize information in a particular system. The resulting model is perhaps of most use to students (using this term in a very broad sense). A model is of less use to a "professional"--one who has worked in the system, and understands both its beauty and its warts. In my discussion of general facies models (Walker 1984b), I emphasized that good models can only be constructed by the careful comparison of many modern and ancient examples. The examples must be from a homogeneous population; a poor model with limited predictive power will result if end-member examples of (say) meandering and braided rivers are carelessly or unthinkingly mixed together to make a "generic river" model. I also commented that models could be expressed "as idealized sequences of facies, as block diagrams, and as graphs and equations" (Walker 1984b, p. 5). I considered it obvious that three-dimensional control (block diagrams) was preferable to two-dimensional control (vertical sequences). Most workers in the field have tried to record vertical and lateral facies changes (given appropriate outcrop or ability to correlate), and I do not feel as strongly as Miall does (1985, p. 263) that vertical profiles have been overemphasized. There are many problems involved in the construction of facies models, including the choice of environment to model and the scale at which to model it. I speculated it 1984 that "as very large scale systems are studied in more detail (e.g., submarine fans), models for sub-components of the system may emerge" (Walker 1984b, p. 7). What constitutes a system and a sub-environment are matters of opinion (and may depend on the scale of the study), but it must be emphasized that a model can only predict within the framework of the model. Thus a model for the channel-levee complexes on submarine fans cannot predict beyond the channel-levee complexes, for example to basin plain facies. It may therefore be preferable to try and model entire depositional systems rather than subcomponents, despite my 1984 comments cited above. I strongly retain the idea that facies models are absolutely necessary in stratigraphy and sedimentology, and hence that efforts to improve existing models and construct new ones are worthwhile. Models alone give us a norm, without which we are unable to assess the significance of a new example. Models alone embody the predictive capability ofsedimentology (Walker 1984a). However, before suggesting new approaches to facies modeling, I will consider the topic of "architectural element analysis" (Miall 1985).

779

analysis, and claims that this analysis, unlike facies modeling, "reverts to the purely descriptive." Architectural elements are associations of facies, or individual facies, separated by bounding surfaces. In the case of fluvial deposits, Miall (1985, p. 269) suggests eight basic elements which he described carefully; however, the descriptions also embody some interpretation. These basic elements were given genetic names; "'channels, gravel bars and bedforms, sandy bedforms, foreset macroforms (now termed downstream accretion, Miall 1988), lateral accretion deposits, sediment gravity flows, laminated sand sheets, and overbank fines." The analysis of these elements involves the "ways in which they may combine and interbed with each other"; these ways are "almost infinitely variable" (Miall 1985, p. 297). This type of analysis used to be called "vertical and lateral facies relationships" and is an extremely valuable and necessary part of stratigraphy and sedimentology. Miall (1985, p. 299) concludes that "the architectural elements have become the norm for purposes o f comparison, the framework and guide for future observations, the predictor in new geological situations, and the basis for hydrodynamic interpretations.'" In my opinion, this sentence reveals two fundamental flaws in architectural element analysis. First, if the combinations of elements are "almost infinitely variable," and the elements have become the basis for prediction, it follows that prediction can only be attempted within an element. Recognition of a downstream accretion element, for example, allows prediction of facies only within the downstream accretion element, but it is impossible to predict where to look for channels, gravel bars and bedforms, or lateral accretion deposits. Second, AEA offers no overall point of reference (norm) for a depositional system as a whole. Each combination of architectural elements (each individual example) is treated as unique, and in the absence of a norm, there is no way of knowing whether the individual example is similar to, or greatly different from, other examples. This is sedimentological anarchy. Finally, it is obvious that AEA requires three-dimensional outcrop of a kind rarely encountered; it is almost impossible to do in the subsurface where bounding surfaces are hard or impossible to define in cores. The fact that certain facies, facies associations, or architectural elements occur universally allows comparison and interpretation of many local examples--classical turbidites form a good example. The description of architectural elements embodies the search for the basic building blocks of sedimentology, world-wide, Archean to Recent. This is the strength of architectural elements, not their "analysis.'" FUTURE DIRECTIONSIN FACIESMODELING

ARCHITECTURALELEMENTANALYSIS(AEA) Alien (1983) first introduced the concept of architectural elements, which can be considered as a generalization of the concept of facies associations (Collinson 1969). Miall (1985, p. 297) has gone one step farther in suggesting that the elements can be used in stratigraphic

In the past, the basic approach to facies modeling has consisted of 1) subdividing all depositional environments into a relatively small number of basic types (sandy rivers, deltas, shelf etc.) and then 2) describing those environments as carefully as possible by comparing recent examples with those interpreted to be their ancient coun-

780

R O G E R G. W A L K E R

terparts (Walker 1984a). This comparison results in general statements about the environment that form the basis for a model. In attempting to divide up environments, Galloway (pers. comm. 1989) doubts that a few models "can reasonably reflect the diversity of depositional systems/facies that e x i s t . . . [and thus he has] emphasized the system approach -- a few basic geomorphic and process-defined depositional systems, each with a spectral array of styles (tide, wave, river dominated deltas), and each composed of a predictable suite o f facies." However, although an environment is a geomorphic entity influenced by internal sedimentological processes, its preservation in the geological record is strongly influenced by outside controls such as the rate and type of sediment supply, tectonics, and relative sea-level fluctuations. Facies models as presently formulated attempt to recognize those outside controls, but generally in an inadequate way. For example, in my submarine fan model (Walker 1984c), static plan views are emphasized. Only one hopelessly overgeneralized and unrealistic vertical section attempts to show what might happen during fan progradation. New submarine fan models must attempt to relate in time and space all of the autocyclic and allocyclic elements, such as basin floor sheet sands, channel-levee complexes (with or without depositional lobes at the end), rates and types of sediment supply, and fluctuations o f relative sea level. In other environments, the need for more responsive, less static, facies models will even influence the choice of environment modeled. For example, the stratigraphic sequence shown in Figure 1 is increasingly commonly recognized in the geological record. Facies are described next to the graphic log, and interpreted in the second column (where the term "shelf" is an abbreviation for any open, shallow marine environment). The facies succession represents the progradation of a storm-dominated shoreface into a storm-dominated offshore area. It now seems appropriate to model this as one system, rather than setting up models for barrier islands (incorporating some discussion of the shoreface) and separate models for the shelf (subdivided into storm- and tide-dominated). Similarly, tide-dominated deltas and offshore tidal sand bars should probably be considered as part of one system, rather than modeling tide-dominated deltas as one third of the overall deltaic system (river- and wave-dominated deltas make up the other two thirds), and offshore tidal sand bodies as one half of the shelf and shallow marine system (the other half being storm-dominated) (Walker 1984a). This idea can be traced back a long way--at least to the depositional systems of Fisher and McGowen (1967). Depositional systems are "three-dimensional assemblages of process-related facies that record major paleogeomorphic basin elements" (Galloway 1989, p. 126). A "linkage of contemporaneous depositional systems" can be termed a "systems tract" (Brown and Fisher 1977). They took the idea beyond that of depositional environments, recognizing regional or basinwide unconformities, and stating that "[seismic] reflection-bounded units composed of contemporaneous depositional systems (systems tracts) a r e . . , called 'seismic-stratigraphic units'" (Brown and Fisher 1977, p. 215).

Future facies modeling must emphasize these contemporaneous, linked depositional environments, and their response to tectonics and changes of relative sea level. This will combine the strengths of classical facies modeling with the recognition that widely spaced and "distinct" geographic environments (summarized as models) can be rapidly superimposed as part of one transgressive or regressive system. THE NEW STRATIGRAPHIES

There are currently at least four stratigraphies that attempt to subdivide rocks into genetic packages based on bounding unconformities or discontinuities. They are largely conceptual, with little or no consideration of 1) scale of application, 2) actual geological examples (although Galloway 1989, provides some), or 3) the relationship between the different schemes. They all derive from seismic stratigraphy (Vail and Mitchum 1977), in which large scale subdivision is based upon discordances between seismic markers (toplap, downlap, otflap). Seismic sequences between markers are "generally tens to hundreds of meters thick" (Mitchum et al. 1977, p. 56); examples from West Africa average 600-700 m (Todd and Mitchum 1977, p. 157; Mitchum and Vail 1977, p. 137). The concepts of seismic stratigraphy have evolved into those of sequence stratigraphy (Van Wagoner et al. 1988; Posamentier et al. 1988; Posamentier and Vail 1988) with additional geological rather than seismic emphasis. The scale implied by sequence stratigraphy is not explicitly discussed by the authors mentioned above, but Jervey (1988, p. 41, 56, 68) indicates thicknesses in the 100-300 m range. This glossing over of the scale problem by the Exxon group creates problems with terms such as "significant hiatus" and "relatively conformable," because these terms have very different connotations in different scales of study. The scale of a "genetic stratigraphic sequence" is not specifically discussed by Galloway (1989), although his Tertiary G u l f Coast examples are several hundred meters thick even before they expand downdip due to growth faulting. The differences between sequence stratigraphy and genetic stratigraphic sequences are subtle and will be discussed below. Seismic- and sequence-stratigraphic schemes suggest that the bounding unconformities result from widespread changes of relative sea level and have generated considerable discussion concerning the influence (dominance?) of global sea-level fluctuations. Problems concerning the global universality of coastal onlap curves have been succinctly reviewed by Miall (1986). Seismic stratigraphy, sequence stratigraphy and genetic stratigraphic sequences are not formally recognized by the North American Commission on Stratigraphic Nomenclature (NACSN 1983). The fourth scheme, allostratigraphy, is formally recognized. It is similar in concept to sequence stratigraphy in that the stratigraphic units are unconformity bounded. Specifically, allostratigraphy recognizes a body of sedimentary rock "defined and identified on the basis of its bounding discontinuities" (NACSN 1983). Formal names (Alloformations, Allomembers) can be assigned, and the subdivisions are sim-

PERSPECTIVE

IT " 5 ~ ~ ~ . ~ . . ~

781

HIGHSTAND .SYSTEMSTRACT

SYSTEMS TRACT FIo. Z--Definition of terms used in the text, based on facies geometries and relationships in the Cardium Formation. SE, subaerial erosion; IT, initial transgression; RT, resumed transgression; Mar.FS, marine flooding surface; MaxFS, maximum flooding surface. Stipple indicates coastal and nearshore marine facies in three sandier-upward prograding parasequences. Parasequences comprise floodplain rocks to left of stipple, and offshore marine rocks to right.

ilar in scale to lithostratigraphic formations and members (and are commonly much smaller than the subdivisions in the first three stratigraphies discussed above). The basic conceptual ideas of sequence stratigraphy have been outlined by Van Wagoner et al. (1988, p. 39; Fig. 2). The fundamental unit of sequence stratigraphy is the sequence, which is bounded by unconformities and their correlative conformities. A sequence can be divided into systems tracts, which are defined by their position within the sequence, and by the stacking patterns of parasequence sets and parasequences bounded by marine flooding s u r f a c e s . . , a parasequence is a relatively conformable succession of genetically related beds or bedsets bounded by marine flooding surfaces... [which are surfaces] across which there is evidence of an abrupt increase in water depth. It is clear that there are problems with the application of these essentially theoretical ideas of stratigraphy: no scale is suggested; "relatively unconformable" is undefined; the meaning of "genetically related" is not explained; many successions of marine rocks (e.g., thick Lower Paleozoic quartzites in many parts of the world) do not contain parasequences; and no guidance is given with respect to distinguishing parasequence bounding surfaces from sequence-bouding unconformities. Finally, sequence stratigraphy, with its emphasis on bounding unconformities and marine flooding surfaces, cannot easily be applied to non-marine rocks (if at all). Some of the concepts and terminology are shown in Figure 2. Here, three prograding parasequences make up a parasequence set, with stipple indicating shoreface and inner shelf sands. Subaerial erosion (SE) is illustrated by incised fluvial channels (also stippled). The sequence bounding unconformities that initially formed subaerially were subsequently modified by marine erosion--either by continuous transgression (IT = initial transgression) or by transgression punctuated by stillstand and shoreface incision ( I T / R T = initial t r a n s g r e s s i o n / r e s u m e d transgression). Marine flooding surfaces (Mar.FS) bound the parasequences and the m a x i m u m flooding surface (Max.FS) separates the two systems tracts.

It appears that a parasequence is similar in scale and concept to the facies sequence of Teichert (1958), to Busch's (197 l) GIS (genetic increment of strata), and to the facies (sequence) succession of de R a a f et al. (1965) (recycled terminology again). A parasequence (Fig. 2) is the result of a set of related depositional conditions--for example, a shoreface progradation, giving a facies succession of offshore mudstones, interbedded HCS sandstones and bioturbated mudstones, shoreface deposits (cross bedded and/or swaley cross stratified), and foreshore parallel laminations (Fig. l). Such a succession could vary from 5 to about 50 m thick and might be capped by marine transgressive strata above a Marine Flooding Surface (Mar.FS, Fig. 2) that initiates another set of depositional conditions. If so, the parasequence is limited by bounding discontinuities, and hence can be formally named in an allostratigraphic scheme (see, for example, Plint et al. 1986, 1987; Bergman and Walker 1988). GENETIC RELATIONSHIPS IN STRATIGRAPHY

Another problem with the Exxon stratigraphic schemes is that the exact meaning of the phrase "'genetically related strata" is never spelled out. Genetic relationships can be hypothesized on at least two scales, with different implications of the term genetic. On the small scale, for example, a 20-30-m-thick prograding, coarsening-upward succession may contain facies that are genetically related 1) by being laterally gradational before progradation stacked the facies vertically, and 2) by being deposited during one set ofprograding depositional conditions (Figs. 1, 2). Under these conditions, the term "genetically related" embraces sedimentological parameters such as the rate and type of sediment supply, the wave and tide climate of the basin, salinity, etc. These may remain fairly constant during the progradational event. On the large scale o f sequence stratigraphy or genetic stratigraphic sequences, the facies are genetically related only because they are assumed to have been deposited during one complete cycle of relative sea level fluctuation. The resulting sequence can be subdivided into systems tracts (Posamentier et al. 1988; Posamentier and Vail 1988; Fig. 2),

ROGER G. WALKER

782

but the sedimentological controls operating in the transgressive tract (for example, low or zero rate of sediment input, and reworking of older deposits into transgressive tidal sand wave complexes) may be quite different from those operating in the subsequent highstand systems tract (for example, high rates o f sediment supply, prograding river-dominated deltas, and offshore sand transport dominated by storms). There are at least three reasons for studying stratigraphic relationships on the large scale of sequence stratigraphy and genetic stratigraphic sequences. The first is to attempt inter-regional or global correlations ofstratigraphic packages using global eustatic sea level curves as a basis for the correlation. The second is to develop a time-stratigraphic framework for the rocks under study. The third is to treat the geometry of the unconformities in the sense of a broadly defined predictive model, such that (for example) a type 1 sequence boundary (Van Wagoner et al. 1988) on the shelf predicts a lowstand fan and lowstand wedge at the base of slope. In this large scale sense, the depositional systems are genetically related within systems tracts, and some of the systems tracts might also be genetically related within the cycle of relative sea level fluctuation (for example, transgressive overlain by highstand systems tracts). However, I emphasize the word might here because it is very important to emphasize the extent to which depositional patterns can and will change across unconformities and across m a x i m u m flooding surfaces. BOUNDARIES OF GENETIC PACKAGES

I will use a well worked out, but relatively small scale, example to examine how sequence stratigraphy and/or genetic stratigraphic sequences can be used to study rocks in the Cretaceous Western Interior Seaway (Fig. 3, Cardium Alloformation; Plint et al. 1986, 1987; Bergman and Walker 1987, 1988; Pattison 1988; Walker and Eyles 1988; Leggitt el al. 1990). I will look at the choice of boundaries of genetic packages, particularly contrasting the approach associated with workers from Exxon (sequences bounded by unconformities; Vail and Mitchum 1977; Posamentier and Vail 1988; Posamentier el al. 1988) and the Galloway (1989) approach (genetic stratigraphic sequences bounded by flooding surfaces; Figs. 2, 3). Erosion surface E5 (Bergman and Walker 1988) rests exclusively on marine rocks, with no evidence ofsubaerial exposure. Surface E4 rests partly on marine (Pattison 1988), and partly on non-marine rocks (Plint et al. 1986). The regional development of both erosion surfaces suggests that marine rocks have been subaerially exposed by a fall of relative sea level and that the subaerial erosion surface has subsequently been transgressed; where marine rocks rest on older marine rocks, all evidence of exposure has been eroded away, including any incised streams. It is therefore not clear whether they are type 1 or type 2 sequence boundaries (even if this terminology can be used in basins which do not have distinct shelf/slope breaks). Transgression was interrupted by short periods of relative stillstand, when the northeast-facing asymmetrical scours (Fig. 3), interpreted as shorefaces, were incised (Bergman

and Walker 1988; Walker and Eyles 1988). Sand and gravel were supplied to the shorefaces during stillstand. When transgression resumed, the upper parts of the shoreface succession, including the beach, were truncated, and coarse sediment was spread southwestward as a lag. Two transgressive surfaces can therefore be recognized; an initial transgression (IT) and a resumed transgression (RT). The preserved shoreface deposits between IT and RT can be up to 18 m thick in the Carrot Creek Allomember (Fig. 3), which is defined by the "bounding discontinuities" IT and RT. These were originally termed E5 (the initial erosion surface) and T5 (implying transgression, which we now recognize is also erosive in many places). Away from the incised shorefaces, IT and R T become the same surface (IT/RT; location 1 in Fig. 3), which is overlain by a thin transgressive pebble lag. This lag is physically continuous with the Carrot Creek shoreface, and hence (despite being so thin) it is also termed the Carrot Creek Allomember. Eagle-eyed readers will notice that in the shoreface (location 2, Fig. 3), the allomember is defined by IT below and RT above; in the transgressive veneer, IT and R T are the same surface and the pebbles overlie I T / R T (location 1, Fig. 3). If allostratigraphie concepts are strictly applied, the pebble veneer should belong to a separate and younger allomember from the shoreface conglomerate. Here, the scale problem becomes important again, because in practical terms, this violation of principles is at such a small scale that little is served by giving the veneer a separate allomember name, so long as the violation is understood and appreciated. At both the Burnstick and Carrot Creek horizons (Fig. 3), the shoreface conglomerates and transgressive veneers are abruptly overlain by mudstones at locations 1 and 2 in Figure 3. Without chemical or micropaleontological analysis, we cannot easily determine from cores exactly where the m a x i m u m flooding surface (MFS) occurs. It is stratigraphically convenient to assume that it occurs at the top of the coarse conglomeratic layer, because in subsurface applications of these ideas, correlation depends on well log picks. The conglomerate/mudstone contact can easily be picked from resistivity and g a m m a ray logs, and thus for practical purposes it can be equated with the MFS in this Cardium example. Indeed, in some areas the g a m m a ray log shows a "'hot" (radioactive) marker that can represent the MFS. One added complication can be seen at location 3 (Fig. 3), where pebbly mudstones occur stratigraphically adjacent to shoreface conglomerates. The pebbly mudstones are interpreted to have formed at about the time that renewed transgression (RT) began. During the initial stages of RT, some of the shoreface gravel was moved southwestward as transgressive lags, but some gravel was moved by storms offshore (northeastward), where it mixed both physically and biologically (by bioturbation) with transgressive m u d to form pebbly mudstones. Thus at location 3 (Fig. 3) the RT surface separates conglomeratic shoreface from pebbly mudstone, and MFS is drawn immediately above the last pebbly horizon. The pebbly mudstones thin and disappear basinward, where the RT and MFS surfaces blend together. The progradational systems tract (Raven River Allo-

PERSPECTIVE

MFS

78 3

] AIIomember

,l

~..--N

E5

Roven River

~'~

RT (T4) -.-.

o-.-..._.

~

m

-

"°b'::"

IT

o:~ z -

- -.

RT (T4) I * "~'2-~:-°°~" ~" ~.> : 2~

Burnstick

AIIornember Hornbeck AIIomember ,

-z.

MFS T4 ¢ ° - °- °- ° - °- . . . . . 1, .

I Burnsfick AIIomember

IT/RT

¢

E4

I

Fro. 3.--Relationship of depositional facies, erosion surface geometry and sequence boundaries, based on the Cardium Formation. MFS, maximum flooding surface; RT, resumed transgression; IT, initial transgression. Note that the RT surface in the northeast becomes the IT surface as transgression proceeds to the southwest. E4, T4, E5, T5 refer to original Cardium terminology (Hint et al. 1986). Cardium Allomembers are named (after Plint et al. 1986) and sequence boundaries according to Galloway (1989) and Exxon workers are shown.

member) comprises a sandier-upward succession of black mudstones, bioturbated silty mudstones, and HCS sandstones interbedded with bioturbated mudstones. In the scheme associated with Exxon, log markers within this prograding package should downlap onto MFS (Fig. 2), but this cannot be demonstrated in the Cardium. The Cardium depositional scheme has been explained briefly above, in order that the following points can be made. First, a stratigraphic sequence (in the Exxon sense) could be drawn from IT to IT (or E4 to E5), whereas a Galloway (1989) genetic stratigraphic sequence could be drawn from MFS to MFS. ! quote Galloway (1989, p. 132) extensively to illustrate a problem I perceive in his scheme: Distinct shelf-system deposits, including sand-rich facies, are most likely formed during transgression and flooding (Swift and Rice 1984). Because shelf deposits are derived from reworked transgressed or contemporary retrogradational deposits, their distribution commonly reflects the paleogeography of the precursor

depositional episode. These deposits are best included in and mapped as a facies element of the underlying genetic stratigraphic sequence. Up to this point, I have deliberately avoided the term "depositional episode." It was probably first used by Frazier (1974), who used both "depositional event" and "depositional episode." Galloway's episode appears to be equivalent to Frazier's event (compare Frazier's figure 2 with Galloway's figure 3), and Galloway's "'depositional episode" gives rise to a "genetic stratigraphic sequence" (caption of Galloway's (1989) figure 2). Because I am discussing Galloway's ideas below, I am forced to use this confusing terminology. The Cardium conglomerates are "contemporary retrogradational deposits," deposited in incised shorefaces, but these shorefaces do n o t appear to "reflect the paleogeography of the precursor depositional episode." If Galloway's scheme is applied to the Cardium, it is not even clear what the "precursor episode" represents; it could be the shelf HCS sandstones that underlie the conglom-

784

R O G E R G. W A L K E R

erates (upper part of Raven River Allomember, Fig. 3), or it could be the floodplain depositional episode (now eroded away) that immediately preceded transgression. Galloway (pers. comm., 1989) has suggested that the precursor episode would include "deposition of the shoreface units, as well as the fluvial or valleys fills that [you] infer to have been present." Finally, if I quote Galloway in a Cardium context, I cannot agree that "'these deposits [the Cardium shoreface conglomerates] are best included in and mapped as a facies element of the underlying genetic stratigraphic sequence [Raven River HCS shallow marine sandstones]"; they have absolutely nothing to do with the underlying deposits. The second point that follows from this brief consideration of the Cardium is a very simple and very general one, namely that there is unlikely to be a direct sedimentological genetic relationship between rocks below and above an unconformity, and below and above an MFS (where a "sedimentologically genetic" relationship implies c o m m o n rates and types of sediment input, common wave, tidal and current conditions, salinities etc.). The stratigraphic sequence (in the Exxon sense) is not sedimentologically genetic because it spans an MFS (the Burnstick conglomerates are n o t genetically related to the sandier-upward Raven River succession), nor is Galloway's sequence sedimentologically genetic because it spans an unconformity (the Raven River succession is not genetically related to the Carrot Creek shoreface conglomerates). The changes that take place both at unconformities and MFS's may be profound and involve changes in basin size and depth (affecting major factors such as tidal range and the fetch of waves), gradients (both basin floor and floodplain), rates of sediment supply, salinities (influx of fresh water, basin restriction), and many other factors. Galloway (pers. comm., 1989) agrees with the use of allostratigraphic units for the Cardium but maintains that " m a n y (most?) transgressive units reflect, with modification, in their facies distribution the depositional environments/facies transgressed.'" Our differences of opinion perhaps concern how much "modification" has taken place during transgression, and hence the extent to which the underlying systems and transgressive systems are sedimentologically genetically related. I therefore emphasize again that the only genetic factor that operates at the Exxon or Galloway scale is the assumed relationship between a sequence, or genetic stratigraphic sequence, and one cycle of relative sea level fluctuation. As defined above, a sedimentologically genetic package cannot extend farther than 1) unconformity to overlying MFS, or 2) MFS to overlying unconformity. In correspondence, Galloway (pers. comm., 1989) has raised another interesting point with respect to the choice of unconformity versus MFS in the Cardium (Fig. 3). He suggests that interestingly the Exxon sequence model can be used for the C a r d i u m . . . only if one is willing to have the principal stratigraphic boundary (subaerial unconformity) be a surface that no longer exists and

can never be conclusively proven to have existed by direct observation. In contrast, the surface of marine erosion is real and is a dominant bounding stratigraphic element . . . I think it is fair to argue that an Exxon sequence cannot be legitimately drawn in your figures [Fig. 3]. There is n o s u b a e r i a l u n c o n f o r m i t y and therefore no "depositional sequence" boundary p r e s e r v e d in the Cardium stratigraphic record as you have interpreted it. The problem here lies with the broad concept of sequence boundaries as expressed by Exxon geologists and the absence of a clear statement that subaerial erosion surfaces are sometimes (commonly? always?) extensively modified or destroyed by subsequent transgression. But regardless of the extent of modification, the Cardium erosion surfaces c a n conclusively be proven to exist; one c a n do Exxon stratigraphy in the Cardium if one so wishes (Fig. 3). Finally, D. J. Cant (pers. comm., 1990) has raised the interesting question as to whether one can distinguish different types of bounding surface given information (core or outcrop) only from the basin center. If far enough from the basin margin, in cratonic or foreland basins without lowstand submarine fans, bounding discontinuities may exist as (unrecognizable) correlative conformities, and m a x i m u m flooding surfaces may simply be pauses in deposition within an already-slowly-deposited fine grained succession. One is left with the uncomfortable feeling that sequence stratigraphy may be very difficult in such situations, just as it is in thick non-marine successions. SYSTEMS TRACTS AND FACIES MODELING The sediment packages between unconformities and MFS's, or MFS's and unconformities are essentially syst e m s tracts (Fig. 2). In Figure 3, the Burnstick and Carrot Creek Allomembers are both transgressive systems tracts, and the Raven River is a prograding or highstand systems tract. The systems tract concept can clearly be applied on m a n y different scales. The concept unites a group of contemporaneous depositional systems. Each system (delta, barrier island, etc.) could be individually modeled, but my suggestion is that under the influence of progressive relative sea-level change one system can quickly evolve into another. For example, in m a n y alleged tide-dominated "deltas," the bulk of the sandy deposits appear to consist of submerged tidal sand ridges--the Klang (Malaysia) and Ord (Australia) are good examples (Coleman and Wright 1975). These systems are not adequately modeled at the moment, because the deltaic part (the irregular progradation of the shoreline) falls into one model, and the submerged tidal ridges fall into another (tide-dominated shelves). I f the submerged tidal ridges began life as deltas, and have been submerged by Holocene transgression, it may be more appropriate to formulate a model for a "transgressive tidal shelf/tidal delta systems tract.'" Other examples spring to mind; for example, prograding storm dominated shoreface/offshore systems (Fig. 1;

PERSPECTIVE

McCrory and Walker 1986; Rosenthal and Walker 1987; Plint and Walker 1987), or transgressive barrier island systems that involve lagoonal muds and sands, washover deposits, and lower shoreface storm sands m o v e d seaward as the barrier transgresses landward (Rampino and Sanders 1980). CONCLUSIONS

785

unit to model, because I suspect that there are fewer basic systems tracts than there are depositional systems. Clearly, environments can be separated or linked at many different scales. Galloway (pers. comm., 1989) comments that shorezone and shelf systems . . . [commonly evolve] from progradational to transgressive. If overall paleogeography and marine process framework remain similar, I have found it useful to view the depositional product as that of a single, evolving system. If the changes are significant, it becomes useful to separate progradational vs. transgressive systems as separate entities. There is nothing intended in my sequence model to prevent one doing this where it is useful

Although local facies schemes will continue to flourish, it is clear that more and more facies and facies associations (architectural elements) are being recognized as having universal application--classical turbidites, fluvial lateral accretion deposits, and swaley cross-stratified shorefaces are three examples. Defining more of these universal architectural elements is an important goal. On The /f ("if overall paleogeography and marine process a broader scale, the sedimentological approach to straframework remain similar") is a big IF, and I suspect that tigraphy involves the search for genetically related groups one will normally wish to treat the progradational and of rocks. The new genetic stratigraphic schemes attempt transgressive systems separately. to do this, and, as emphasized by Posamentier (pers. Modeling involves the search for the items shared in comm., 1989), they can be applied at any scale. However, c o m m o n between similar ancient and modern systems because they are also enmeshed with concepts of cycles tracts, the recognition of the role of relative sea-level of sea-level fluctuation, they do not necessarily identify the establishment of a norm, and the articsedimentologically genetically related groups of rocks (as fluctuation, ulation of the model in such a way that it can be used defined above). The best descriptive working scheme is These are also the ways in which sedimenallostratigraphy, in which units bounded by discontinu- predictively. tology and genetic stratigraphy are converging, and they ities can be formally recognized on any scale (NACSN present challenges for the large-scale understanding of the 1983), even down to separating as distinct allo-units the sedimentary record. transgressive lag above I T / R T (Fig. 3) from the shoreface conglomerate between IT and RT (if this is perceived as ACKNOWLEDGMENTS desirable and geologically useful). But regardless of controis, be they sedimentological or related to cycles of seaThe ideas in this paper have been stimulated by dislevel fluctuation, both allostratigraphy and sequence stracussions with many friends. I particularly thank Janok tigraphy help to establish a relative time-stratigraphic framework for sedimentary rocks, and both recognize the Bhattacharaya, Bill Galloway, Andrew Miall, Dale Lecksignificance of hiatuses of various temporal and spatial ie, Guy Plint and Henry Posamentier for reading a first draft of the manuscript, and Doug Cant and Gerry Midextents. Choices of sequence boundaries may be constrained by dleton for subsequent comments. They have all contribthe demands of a particular investigation and by the con- uted valuable insights and suggestions. I have tried not ceptual preferences of an individual worker. Thus Posa- to misinterpret any of their ideas, and I am solely rementier (pers. comm., 1989) prefers to use the uncon- sponsible for the wording and emphasis given in the text formity over the MFS as the boundary, on the grounds except where direct quotations are given. Operating and that reservoir facies typically overlie the unconformity, Strategic Grants from the Natural Sciences and Engiwhereas the MFS is associated with the source of hydro- neering Research Council of Canada have funded work carbons, and the seal of the reservoir. In contrast, Gal- in western Canada, where some of the ideas have been loway (1989, p. 138-140) gives a series of reasons for his developed and tested. preference of the MFS instead of the unconformity. These reasons stem from the fact that he "emphatically [does REFERENCES not] think that sea level is the principal control of se- ALLEN,J. R. L., 1983, Studies in fluviatile sedimentation: bar complexes quence d e v e l o p m e n t . . . [hence] I prefer the condensed and sandstone sheets (low-sinuosity braided streams) in the Brownsection/MFS" (Galloway, pers. comm., 1989). My own stones (L. Devonian), Welsh Borders: Sed. Geol., v. 33, p. 237-293. suggestion is to recognize that both schemes have similar BEROmAN, K. M., AND WALKER, R. G., 1987, The importance of sea level fluctuations in the formation o f linear conglomerate bodies: strengths and weaknesses; for smaller scale sedimentoCarrot Creek Member, Cretaceous Western Interior Seaway, Alberta, logical purposes, neither is appropriate. Canada: Jour. Sed. Petrology, v. 57, p. 651-665. Concepts of depositional systems and facies models BERGMAN, K. M., AND WAtJ~R, R. G., 1988, Formation of Cardium erosion surface E5, and associated deposition of conglomerate; Carrot both involve an initial choice--what is to be modeled. Creek field, Cretaceous Western Interior Seaway, Alberta, in James, Will it be an entire delta, or is a distributary mouth bar D. P., and Leckie, D. A., eds., Sequences, Stratigraphy, Sedimentola big-enough and separate-enough entity to recognize and ogy; Surface and Subsurface: Can. Soc. Petrol. Geol., Mere. 15, p. model as a depositional system by itself? The choices may 15-24. be a little simpler if systems tracts are taken as the basic BnAr'rACH,~JtAyA,J., 1988, Autocyclic and allocyclic sequences in fiver-

786

R O G E R G. W A L K E R

and wave-dominated deltaic sediments of the U pper Cretaceous Dunvegan Formation, Alberta; core examples, in James, D. P., and Leckie, D. A., eds., Sequences, Stratigraphy, Sedimentolngy; Surface and Subsurface: Can. Soc. Petrol. Geol., Mem. 15, p. 25-32. BROWN, L. F., JR., AND FISHER, W. L., 1977, Seismic-stratigraphic interpretation ofdepositional systems: examples from Brazilian rift and pull-apart basins, in Payton, C. E., ed., Seismic Stratigraphy--Applications to Hydrocarbon Exploration: A.A.P.G., Mem. 26, p. 213248. BUSCH, D. A., 1971, Genetic units in delta prospecting: A.A.P.G. Bull., v. 55, p. 1137-1154. COLEMAN, J. M., AiqDWmorIT, L. D., 1975, Modern river deltas: variability of processes and sand bodies, in Broussard, M. L., ed., Deltas, Models for Exploration: Houston Geol. Soc., p. 99-149. COLLINSON,J. D., 1969, The sedimentology of the Grindslow Shales and the Kinderscout Grit: a deltaic complex in the Namurian of northern England: Jour. Sed. Petrology, v. 39, p. 194-221. DE RAAF, J. F. M., READING, H. G., Ai~rDWALKER, R. G., 1965, Cyclic sedimentation in the Upper Carboniferous of North Devon, England: Sedimentology, v. 4, p. 1-52. FISHER, W. L., ANn McGOWEN, J. H., 1967, Depositional systems in the Wilcox Group of Texas and their relationship to occurrence of oil and gas: Gulf Coast Assoc. Geol. Socs., Trans., v. 17, p. 105-125. FRAZmR, D. E., 1974, Depositional episodes: their relationship to the Quaternary stratigraphic framework in the northwestern portion of the Gulf basin: Austin, TX, Bureau of Economic Geology, Geol. Circular 74-l, 28 p. GALLOWAY, W. E., 1989, Genetic stratigraphic sequences in basin analysis I; architecture and genesis of flooding-surface bounded depositional units: A.A.P.G. Bull., v. 73, p. 125-142. J~RVEV, M. T., 1988, Quantitative geological modeling of siliciclastic rock sequences and their seismic expression, in Wilgus, C. K. et al., eds., Sea Level Changes: An Integrated Approach: SEPM Spec. Publ. 42, p. 47-69. L ~ r r r , S. M., WALKER, R. G., AND E'~tasS, C. M., 1990, Control of reservoir geometry and stratigraphic trapping by erosion surface E5 in the Pembina--Carrot Creek area; Upper Cretaceous Cardium Formation, Alberta, Canada: A.A.P.G. Bull., v. 74. McCRoRY, V. L. C., AiqD WALKER, R. G., 1986, A storm and tidally influenced prograding shoreline--Upper Cretaceous Milk River Formation of southern Alberta: Sedimentology, v. 33, p. 47-60. MInLL, A. D., 1978, Lithofacies types and vertical profile models in braided river deposits: a summary, in Miall, A. D., ed., Fluvial Sedimentology: Can. Soc. Petrol. Geol., Mere. 5, p. 597-604. --, 1985, Architectural element analysis: a new method of facies analysis applied to fluvial deposits: Earth-Sci. Rev., v. 22, p. 261308. --, 1986, Eustatic sea level changes interpreted from seismic stratigraphy: a critique of the methodology with particular reference to the North Sea Jurassic record: A.A.P.G. Bull., v. 70, p. 131-137. --, 1988, Facies architecture in clasfic sedimentary basins, in Kleinspehn, K. L., and Paola, C., eds., New Perspectives in Basin Analysis: New York, Springer Verlag, p. 67-81. MIOOL~rON, G. V., 1973, Johannes Walther's law of the correlation of facies: G.S.A. Bull., v. 84, p. 979-988. MITCnUM, R. M., JR., VAIL, P. R., AND THOMPSON,S., III, 1977, Seismic stratigraphy and global changes of sea level, part 2: depositional sequence as a basic unit for stratigraphic analysis, in Payton, C. E., ed., Seismic Stratigraphy--Applications to Hydrocarbon Exploration: A.A.P.G., Mem. 26, p. 53-62. MITCrIUM, R. M., JR., AND VAIL, P. R., 1977, Seismic stratigraphy and global changes o f sea level, part 7: seismic stratigraphic interpretation procedure, in Payton, C. E., ed., Seismic Stratigraphy--Applications to Hydrocarbon Exploration: A.A.P.G., Mere. 26, p. 135-143. Mufti, E., AND Ricci-Luccm, F., 1972, Le torbiditi dell'Appennino settentrionale: introduzione all'analisi di facies: Mem. Geol. Soc. Italiana, v. 1 I, p. 161-199. English translation by T. H. Nilsen, 1978, Intl. Geol. Review, v. 20, p. 125-166. NORTH AMERICAN COMMISSION ON STRAT1GRA.PHICNOMENCLATURE, 1983, North American stratigraphic code: A.A.P.G. Bull., v. 67, p. 841-875. PA'I-rISON,S. A. J., 1988, Transgressive, incised shoreface deposits of

the Burnstick Member (Cardium "B'" sandstone) at Caroline, Crossfield, Garrington and Lochend; Cretaceous Western Interior Seaway, Alberta, Canada, in James, D. P., and Leckie, D. A., eds., Sequences, Stratigraphy, Sedimentology; Surface and Subsurface: Can. Soc. Petrol. Geol., Mem. 15, p. 155-166. PLlr,rr, A. G., ~ WALKER, R. G., 1987, Cardium Formation 8. Facies and environments of the Cardium shoreline and coastal plain in the Kakwa field and adjacent areas, northwestern Alberta: Bull Can. Petrol. Geol., v. 35, p. 48-64. PUNT, A. G., WALKER, R. G., AND BEROMAN, K. M., 1986, Cardium Formation 6. Stratigraphic framework of the Cardium in subsurface: Bull. Can. Petrol. Geology, v. 33, p. 213-225. PUNT, A. G., WALKER, R. G., AND BEaGMAN, K. M., 1987, Cardium Formation 6. Stratigraphic framework of the Cardium in subsurface: Reply: Bull. Can. Petrol. Geology, v. 35, p. 365-374. POSAMENTmR, H. W., JERVEY, M. T., AND VAIL, P. R., 1988, Eustatic controls on elastic deposition I--conceptual framework, in Wilgus, C. K. et al., eds., Sea Level Changes: An Integrated Approach: SEPM Spec. Pubi. 42, p. 109-124. POSAME~rrlER,H. W., AND VAIL, P. R., 1988, Eustatic controls on elastic deposition II--sequence and systems tract models, in Wilgus, C. K. et at., eds., Sea Level Changes: An Integrated Approach: SEPM Spec. Publ. 42, p. 125-154. RAMPINO, M. R., AND SANDERS,J. E., 1980, Holocene transgression in South-Central Long Island, New York: Jour. Sed. Petrol., v. 50, p. 1063-1080. ROSENTHAL, L. R. P., AND WALKER, R. G., 1987, Lateral and vertical facies sequences in the Upper Cretaceous Chungo Member, Wapiabi Formation, southern Alberta: Can. Jour. Earth Sci., v. 24, p. 771783. SLOSS, L. L., 1963, Sequences in the cratonic interior of North America: G.S.A. Bull., v. 74, p. 93-114. SMITH, N. D., 1985, Editorial: new features in JSP: Jour. Sed. Petrology, v. 55, p. 455. SwIvr, D. J. P., AND RICE, D. D., 1984, Sand bodies on muddy shelves: a model for sedimentation in the Cretaceous western seaway, North America, in Tillman, R. W., and Siemers, C. T., eds., Siliciclastie Shelf Sediments: SEPM Spec. Publ. 34, p. 43--62. TEICNER'r, C., 1958, Concepts of facies: A.A.P.G. Bull., v. 42, p. 27182744. TODD, R. G., AND Mrrca-ruM, R. M., JR., 1977, Seismic stratigraphy and global changes of sea level, part 8: identification of Upper Triassic, Jurassic and Lower Cretaceous seismic sequences in Gulf of Mexico and offshore West Africa, in Payton, C. E., ed., Seismic Stratigraphy-Applications to Hydrocarbon Exploration: A.A.P.G., Mem. 26, p. 145-163. VAIL, P. R., Arm MiTCa-nJM,R. M., JR., 1977, Seismic stratigraphy and global changes of sea level, Part 1: overview, in Payton, C. E., ed., Seismic Stratigraphy--Applications to Hydrocarbon Exploration: A.A.P.G., Mere. 26, p. 51-52. VAN WAOONER,J. C., POSAMENTmR,H. W., MITer-tOM, R. M., VAIL,P. R., SARG,J. F., Lotrrrr, T. S., AND HARDm,raOL, J., 1988, An overview of the fundamentals of sequence stratigraphy and key definitions, in Wilgus, C. K. et al., eds., Sea Level Changes: An Integrated Approach: SEPM, Spec. Publ. 42, p. 39--45. VISHER, G. S., 1965, Use of the vertical profile in environmental reconstruction: A.A.P.G. Bull., v. 49, p. 41-61. WALKER, ~ G., 1983, Cardium Formation 3. Sedimentology and stratigraphy in the Caroline--Garrington area: Bull. Can. Petrol. Geology, v. 31, p. 213-230. ---, 1984a, Facies models: Geol. Assoc. Canada, Geosci. Canada Reprint Series 1,317 p. ---, 1984b, General introduction: facies, facies sequences and facies models, in Walker, R. G., ed., Facies models: Geol. Assoc. Canada, Geosci. Canada Reprint Series 1, p. 1-9. , 1984c, Turbidites and associated coarse elastic deposits, in Walker, R. G., ed., Facies Models: Geol. Assoc. Canada, Geosci. Canada Reprint Series 1, p. 171-188. WALKER,R. G., AND EYLF.S,C. H., 1988, Geometry and facies of stacked shallow marine sandier upward sequences dissected by erosion surface; Cardium Formation, Willesden Green, Alberta: A.A.P.G. Bull., v. 72, p. 1469-1494.