You are on page 1of 16



Department of Geology, University of Toronto, Toronto, Ontario M5S 3B1, Canada 2 School of Geosciences, University of Wollongong, Wollongong, New South Wales 2522, Australia email:

ABSTRACT: The Hawkesbury Sandstone has long been assumed to represent the deposits of a large braided river system, comparable in style and magnitude with the modern Brahmaputra River of Bangladesh. Such an interpretation is based mainly on the common occurrence of very large-scale crossbedding, but no architectural studies of the unit have hitherto been carried out. This paper represents a rst attempt to estimate the magnitude of Hawkesbury channels and bars on the basis of the preserved architectural evidence. Photomosaics were constructed of two cliff sections south of Sydney, one 5.6 km in length. On the basis of these proles we estimate that characteristic channel-scale architectural elements are at least 2.7 km wide, and individual macroforms are 510 m high, indicating the constructional depth of typical channels. Hollow elements (scoop-shaped units interpreted to have formed at channel conuences) are up to 20 m deep. These magnitudes are large, but measurably smaller that those of channels and bars in the modern Brahmaputra River of Bangladesh.

Facies models for uvial deposits are based on data from modern rivers and from ancient deposits, but it has been suggested that existing models are of little use because of a lack of primary three-dimensional data from modern rivers, and because of the limited data base (typically vertical prole data or 2-D outcrop data) available for the ancient rock record (Bridge 1993b). However, studies of the ancient record play an important role in the development of facies models because they necessarily focus on what is actually preserved, an emphasis that is only now becoming possible in studies of modern rivers with the use of high-resolution geophysical techniques, such as ground-penetrating radar. Documentation of the preserved architecture of uvial sandbodies is essential for the improvement of our understanding of reservoir heterogeneity (Miall 1988; Weber and Van Geuns 1990). The Triassic Hawkesbury Sandstone of the Sydney Basin in New South Wales, Australia, is an ancient uvial unit that has been little studied but has been cited as probably the deposit of a very large river system. It is particularly well known because of the distinctive, very large-scale crossbedding present in almost every outcrop, notably in cliffs and road cuts in and around Sydney. The sandstone is a craton-sourced unit that was deposited within the foreland basin adjacent to the New England Fold Belt (Cowan 1993). Regional transport directions were towards the northeast; the unit extends for 225 km in that direction, and occupies a belt 75100 km wide, across depositional strike. There has been some debate regarding interpretations of the depositional environment of the Hawkesbury Sandstone, including arguments for shallow-marine (Conolly 1969; Conolly and Ferm 1971) and eolian (Ashley and Duncan 1977) environments. However, it is now universally agreed that the Hawkesbury Sandstone is a uvial deposit. A braided-uvial environment has been suggested, and analogies have been made with the modern giant braided river, the Brahmaputra (also locally called the Jamuna), of India and Bangladesh (Conaghan and Jones 1975; Rust and Jones 1987). These interpretations have been based, in part, on the very largescale crossbedding, which has been interpreted as the deposits of large dunes, the scale of which is limited only by the sediment supply and the
Copyright 2003, SEPM (Society for Sedimentary Geology) 1527-1404/03/073-531/$03.00

size of the channel in which they form (Ashley 1990). A modern facies analysis of the Hawkesbury Sandstone in terms of a braided uvial model was carried out by Rust and Jones (1987). In a separate study they analyzed a particularly characteristic facies of the unit, beds of massive sandstone, and developed an interpretation involving bank collapse and liquefaction of the sands (Jones and Rust 1983). Herbert (1997) assigned the Hawkesbury Sandstone to the lowstand systems tract of a regionally extensive stratigraphic sequence within the Sydney Basin. Given the long-standing assumption that the Hawkesbury Sandstone represents the deposits of an unusually large river, there is a particularly obvious need to explore its architecture, and this paper is offered as a rst study of the large-scale features of the Hawkesbury river system. The paper builds on the facies analysis of Rust and Jones (1987), and is based on studies of the coastal cliff sections of the Kurnell Peninsula and the Royal National Park south of Sydney (Fig. 1), where the section is representative of the central part of the Hawkesbury depositional system. The focus of this paper is on the large-scale architecture. Our data permit an order-ofmagnitude comparison between the modern Brahmaputra River deposits and the Hawkesbury Sandstone, but the study also demonstrates that even where very long outcrops, such as that at Kurnell Peninsula are available, making comparisons between large-scale depositional systems is fraught with uncertainty.

This paper is based primarily on the mapping of architectural features using photographic mosaics of large outcrops as base maps. Photographs of the cliff sections of the Kurnell Peninsula (Fig. 2) were taken from a chartered shing boat. The peninsula is ringed by virtually continuous cliffs up to 40 m high showing nearly 100% exposure for a distance of at least 6 km. We have constructed a prole more than 5.6 km long from our photographs. The scale of this prole is variable, because of the varying distance of the boat from the cliff, as it traversed past the various promontories and bays in the cliff face. The cliffs are sheer, and accessibility for studies on the ground is poor except at the central part and northern end of this prole. For this reason, traditional eld data, such as paleocurrent readings, are sparse, but where it has been possible to obtain such readings they are internally consistent and help to clarify the architectural interpretations. Architectural features are dened mainly by photo interpretation, with some ground checking where the cliffs are accessible. A second prole is based on photographs taken from the air and from a boat off the Royal National Park near Curracurrong (Fig. 3). Access to the base of this prole, along the wave-cut platform at the base of the cliff, is generally good. Additional eld data were obtained at The Cobblers and The Waterrun, locations that were described in some detail by Rust and Jones (1987). Cliff sections continue south to near Stanwell Park, but they are largely inaccessible and have not been studied for this paper. Large road cuts ank the main highways inland from these coastal sections, and these formed the source for some of the details described by Rust and Jones (1987), but they are now somewhat weathered and have not been revisited for this study, In places, especially near Potter Point, wave spray and wind have maintained a strip of cliff top tens of meters wide completely bare of soil and



FIG. 2.Map of the Kurnell Peninsula area, showing the location of the cliff prole.

Bridge (1993a) microscale set (e.g., ripple) mesocsale set (e.g., dune) micro/mesoscale coset macroscale inclined stratum macroscale inclined strata set group of macroscale sets group of macroscale sets

Miall (1996) 1st-order unit (set) 1st-order unit (set) 2nd-order unit (coset) 3rd-order unit (macroform increment) 4th-order unit (macroform element) 5th-order unit (channel element) 6th-order unit (e.g., channel-belt; sequence)

FIG. 1.Project location map.

vegetation, offering superb bedding-plane exposures of the Hawkesbury Sandstone. Classications of facies and architectural features, including their letter codes, are adapted, with minor additions, from those of Miall (1996, Chap. 4). Paleocurrent readings have been precisely located on cliff photographs.

The concept of a hierarchy of depositional scales and enclosing bounding surfaces, and the relationship of this hierarchy to depositional processes on various time scales and physical scales, was rst made explicit by Allen (1983). From his work evolved several approaches to the architectural subdivision of uvial deposits. Two classications are compared below. In this study we employ the numerical ranking of Miall (1996). In this system, the rank of a unit is determined by the lowest rank of the surfaces that enclose it. Bounding surfaces may be of higher rank, where a unit rests on or is cut into by an erosion surface of higher rank, for example a macroform (fourth-order unit) resting on or eroded into by a channel scour surface (fth-order surface).

In the outcrop proles described in this paper the most prominent bounding surfaces, extending for hundreds of meters to a few kilometers along the outcrop face, are those that dene the major sandbodies (the macroscale sets, or major stories of other authors), and are classied as fthorder in rank. They are here termed channel elements. Fifth-order surfaces are typically visible in weathered cliff faces because of subtle weathering differences between coarser sandstone above the surface and ner sandstone below. In some cases, fth-order elements are capped by overbank mudstones. These may be thin (a few centimeters, or less), but the contrast in weathering characteristics between a well-cemented sandstone and even a thin, underlying mudstone accounts for the prominence of some fth-order surfaces in the cliff outcrops. In some instances a zone of groundwater seepage is indicated by a line of iron staining on the cliff surface, indicating the presence of a porositypermeability barrier in the sandstone, or the presence of a thin mudstone. Macroforms constitute the major subdivisions of the fth-order channel elements and are ranked as fourth-order architectural elements. The most distinctive macroforms are those that develop by horizontal accretion, as evidenced by the presence of accretionary cross-bedding extending from the top to the base of the unit. The fourth-order surfaces that dene the top surfaces of macroforms are commonly convex-up, curving in the direction of local ow downward to become surfaces dipping parallel to the



TABLE 1.Lithofacies types and lithofacies assemblages in the Hawkesbury Sandstone near Sydney, Australia.
Stratied sandstone assemblage Ss Pebbly sandstone, with siderite-cemented intraclasts. High-order erosion surface at base. St Trough crossbeds; sets are commonly 23 m thick (Fig. 4). Sp Large-scale planar crossbed sets; sets are commonly 23 m thick; sets up to 7 m have been recorded (Fig. 5). Sl Low-angle crossbedding, dip at less than the angle of repose Scp Compound planar crossbed sets with rhythmically spaced reactivation surfaces. Dip of foresets is commonly at less than the angle of repose. Set thickness 12 m. Sr Ripples and climbing-ripple cosets. Set thickness 5 cm. Massive sandstone assemblage Sm Structureless to faintly laminated sandstone (Fig. 6). Fine-grained assemblage Fm Mudstone. Fl Interlaminated mudstone and laminated to rippled sandstone (Fig. 7). Sr Rippled sandstone.

acteristics because they do not mark contrasts between distinctive lithofacies. Surfaces of rst- to third-order rank are shown but have not been distinguished on the cliff proles. During this study, only in rare cases could architectural analysis be carried out at the scale small enough to permit this kind of analysis.

On the basis of their studies of road cuts and the cliffs near Bundeena, Rust and Jones (1987) subdivided the Hawkesbury Sandstone into three lithofacies assemblages. Individual lithofacies within these assemblages are listed in Table 1. Most of these facies are common in uvial deposits (and are therefore not described in detail here); the only unusual features are the large size of some of the crossbed sets and, unusually for a high-energy uvial deposit, the lack of sandstone with parting lineation (Sh). The latter facies is more common in ne- to very ne-grained sands, except under conditions of very high stream power (Allen 1984); its absence may simply result from the predominance of medium- to coarse-grained sand in the Hawkesbury Sandstone. Some distinctive examples of the range of lithofacies present in the project area are illustrated in the accompanying photographs (Figs. 47).

Architectural-element analysis is most useful for identifying what Allen (1983) termed packets of genetically related strata, which dene depo-

FIG. 3.Map of the Royal National Park area, showing the location of A) The Cobblers and The Waterrun, and B) the Curracurrong cliff prole.

internal accretionary bedding of the macroform. However, many macroforms are truncated by within-channel erosion, and show at fouth- or fthorder upper bounding surfaces. The base of a macroform may be convexup where it coincides with the top of an underlying macroform, but it is typically at. Concave-up fourth-order surfaces occur at the bases of hollow elements and minor channels, in which case they dene fourth-order units. Fourth-order surfaces normally do not have distinctive weathering char-

FIG. 4.Bedding-plane exposure of a large trough-crossbed set (St) forming part of a SD element, south of Cape Solander. Much of the cliff top near Cape Solander and Potter Point is characterized by a complete lack of vegetation owing to wave spray and wind erosion.



FIG. 5.Large, simple planar-crossbed set (Sp) at center of view, overlain and underlain by smaller sets, constituting altogether a simple dune architectural element (SD). South of Cape Solander.

FIG. 6.Trough-shaped unit lled with massive sandstone (Sm) at center of eld of view (base indicated with arrows), overlain and underlain by crossbedded sandstone comprising SD elements. Near Potter Point.

sitional elements within uvial systems larger than individual bedforms and smaller than channels. The techniques are particularly valuable for mapping the products of bar formation and channel-reach migrationmacroforms, to use Jacksons (1975) term. The depositional products of these processes are termed architectural elements. In two-dimensional outcrops that are largely inaccessible to ground inspection, element interpretation relies on large-scale architectural features, primarily the conguration of external bounding surfaces and internal stratication. Paleocurrent data can provide some constraints on the third dimension of individual elements, where it is available. Architectural elements display one of two types of internal conguration. Bedding is either tabular, indicating growth by vertical aggradation, or characterized by a low-angle dip, indicating that it developed by horizontal accretion. Theoretical or model cross-section diagrams of braid bars, such as those provided by Willis (1989) and Bridge (1993b), provide a partial basis for element interpretations. A classication of uvial elements was provided by Miall (1996), based on earlier work by Allen (1983) and Miall (1985, 1988). Table 2 lists the architectural elements at the macroform scale that have been identied in the Hawkesbury Sandstone of the project area, and the following notes provide additional details. DA (Fig. 8) and LA elements are interpreted as downstream and laterally accreting compound bars, respectively, on the basis of the internal relative arrangement of accretion surfaces and Sp and St sets. These elements commonly grade laterally into each other. Distinction between them is made on the basis of the orientation of paleoow directions relative to the orientation of accretion surfaces (Miall 1996, g. 6.24). Compound dune elements (element CD) are a distinctive component of the Hawkesbury Sandstone. Several were observed on the cliff top near Potter Point, and another was recorded near The Waterrun. One of these is illustrated in Figure 9. This element type is characterized by downwardclimbing crossbed sets, resting on bounding surfaces that commonly are signicant erosion surfaces. The structure of these elements is comparable to that of class IV tidal sand waves in the series of models developed by Allen (1980). HO: The hollow element represents the rapid cut and ll of scour hollows at channel conuences downstream from bars or junctions with tributaries. The distinctiveness of this architectural element was rst documented in ancient uvial deposits by Cowan (1991). A facies model for this element has been provided by Bristow et al. (1993). The curved base, lacking a at oor, and a ll composed of a single coset of Sl dipping at an oblique angle to the margin of the hollow (Fig. 10), are characteristics that help to distinguish these deposits from channel-lls. TR(C): This element is not typical of the Hawkesbury Sandstone as a

whole. One example of it was observed, at the Cobblers (Rust and Jones 1987) (Fig. 11), and a thin unit is present at the base of the Curracurrong section (Fig. 12). FF(C): This element is interpreted as a variant of the channel element (CH) in which the channel is of low energy, possibly undergoing abandonment, and is lled by ne-grained deposits. Examples occur at The Cobblers (Fig. 11; see also g. 15 in Rust and Jones 1987) and at Curracurrong (Fig. 12). FF(O): Similar in composition to FF(C), but organized in extensive tabular sheets, representing overbank deposition. Most examples of this element are thin (a few centimeters), but an element 5 m thick and extending along the cliff for nearly 1 km was mapped near Cape Solander (Figs. 7, 13, 14), and another element occurs between The Cobblers and The Waterrun. TR(O): occurring as sandstone sheets in overbank settings, including single dune sets, interpreted to represent individual overbank ood events (e.g., at The Cobblers; Fig. 11).

Kurnell Peninsula A diagram showing the bedding traces interpreted from the outcrop prole is provided as Figure 13. Details of one of the more accessible portions

FIG. 7.Exposure of thin-bedded shale and ne-grained sandstone (Fl), forming part of a FF(O) element, near Cape Solander. Top of element E, near 3000 m point in Kurnell Prole.


TABLE 2.Architectural elements in the Hawkesbury Sandstone of the Sydney area, Australia.
Code Facies Composition Internal/External Architecture Flat base with concave-up margins to element. Upper surface may be gradational into overbank elements Interpretation Channel with preserved cutbanks


Channel element (fth-order) CH Any assemblage of Sp, St, Sr. Ss typically at base of unit. May include component SD, LA, DA units, etc. Within-channel elements (fourth-order) SD Single or stacked large-scale Sp sets, with or without St topsets (Fig. 5) LA Stacked St cosets DA CD HO SS TR(C) FF(C) Stacked Sp cosets. May be capped by St sets (Fig. 8). Stacked sets and cosets of Scp (Fig. 9). Single Sl coset dipping obliquely across or down the centre of the unit (Fig. 10) Sm (Fig. 6)

Tabular upper and lower bounding surfaces Tabular external form or with convex-up top. Contains accretion surfaces with dip directions at high angle to orientation of St trough axes Tabular external form or with convex-up top. Contains accretion surfaces dipping in same direction as Sp crossbed dip Tabular bounding surfaces. Rhythmically spaced reactivation surfaces. Scoop-shaped erosional base Irregular, concave-up erosional base Tabular sets and cosets. Channel form (concave-up erosional basal bounding surface)

Field of large simple dunes Lateral-accretion deposit Downstream-accretion deposit Compound dune element Formation and ll of erosional hollow formed at channel conuence Interdune or between-bar deposit, formed by sediment gravity ow Field of small- to medium-scale bedforms Fill or partial ll of abandoned channel

Trough and ripple (St and Sr) assemblage (Fig. 12) Interbedded Sr, Fm, commonly with intraclast mudstone breccias (Fig. 11) Overbank elements (fourth-order) FF(O) Mudstone blankets (Fm), with or without minor sandstone lenses (Fl) and/or sandstone load structures (Fig. 7). TR(O) Trough, ripple, and laminated assemblages (St, Sr, Fl)

Tabular unit

Low-energy oodplain deposit

Tabular unit

Minor eld of small- to medium-scale bedforms

of the cliff section are shown in Figure 14. Structural dip is virtually zero in the 0 to 5 km section of the prole, which exposes essentially the same 40 m of strata. Northward from the 5.1 km point the prole bends northwestward and reveals a gentle northwestward apparent dip. Interpretation of the rank of the bounding surfaces is dependent on their shape, lateral extent, and associated lithofacies. A series of prominent bounding surfaces, interpreted to be of fth-order rank, subdivides the outcrop into 16 major sand bodies. These are referred to here as channel elements. The 16 channel elements are lettered A to J and U to Z in Figure 13, from the base to the top of the prole. It is possible that some of these channel elements are laterally separated segments of a single element. Thus the boundary between A and B, at the base of the cliff, cannot be determined. Several major bounding surfaces plunge into the sea at the

base of the cliff, but not enough of each is exposed to indicate whether they are likely to be fth-order surfaces. Similarly, the boundary between channel elements F and J, at the top of the cliff, has been removed by erosion. D and E could be the same unit, separated into two by the deep incision of channel element J (although their paleocurrent directions are different). The order of lettering suggests an order of deposition, but in several cases the suggested order cannot be conrmed by architectural relationships and is based only on relative vertical position. For example, the relative timing of channel elements G and H cannot be known because they are separated by more than 2 km. The same uncertainty as to the time sequence is the reason for the separate lettering (UZ) for the channel elements in the 47005600 m interval of the prole. Channel elements A and U may

FIG. 8.Panorama of a downstream-accretion element (DA), near Curracurrong. The element rests on a fth-order bounding surface, and below this is a ner grained sandstone bed, which accounts for the erosional overhang of the base of the DA unit. The element is capped by a fourth-order surface. Elements are labeled as in Figure 17. The scale of this outcrop is indicated by the person, at center (arrow), but the photomosaic was constructed from four photographs all taken from the same point, so the scale is reduced at either end of the view.



FIG. 9.Panorama of a compound-dune element (CD), showing the rhythmically-spaced reactivation surfaces. On clifftop, near Cape Baily Lighthouse, Kurnell Peninsula.

FIG. 10.Two examples of hollow elements (HO). In each case the base is shown by white arrows. A) Hollow approximately 30 m wide, with cosets of Sl dipping to the left, oblique to the lineline of the hollow. Note the sharp erosional truncation of the beds below. Curracurrong. B) A smaller hollow, 18 m wide, lled with Sl sets showing faint lamination, and dipping to the left. 5100 m interval in Kurnell Prole.

Tabbigai Gap Cape Baily Cape Baily Blue Hole Gap

bend in

Cape Solander

Yena Gap bend in profile

Sutherland Point

Fig. 14













































42y 4300












orientation of profile

-1st- to 3 d a d e r surfacas

orientation of profile

4thmder surfaces 5th-and higher-oder surfaws

orientation of profile

orientation of profile

orientation of profile

orientation of profile

Fig. 13

FIG. 13.-The Kurnell Peninsula profile. shown with X5 vertical exaggeration. A horizontal scale in 100 m increments is provided to facilitate reference to specific points along the profile (measured from an arbitrary zero point at the south end of the profile). The orientation of the profile varies somewhat along the cliff, as indicated by the orientation diagrams at the base of the profile and by the course of the boat shown in Figure 2. The top and bottom of the cliff are indicated by the dotted lines. Fifth-order architectural elements are labeled with capital letters in circles. Two- and three-letter codes refer to the architectural-element classification in Table 2. Paleocurrent arrows indicate local mean directions calculated from readings within individual elements. Orientations are plotted with respect to north directed vertically towards the top of the page, and are shown with number of readings.



FIG. 11.Channel lled by ne-grained deposits of element FF(C), The Cobblers. The photographs from which this panorama was constructed were all taken from the same location, and the scale diminishes to the right, as the cliff recedes into the distance. The cliff is curved, concave to the viewer, and the prole intersects the cut-bank of the channel on the left and a bank-attached bar at the right end of the eld of view. Note the large gutter casts at the base of the overlying channel sandstone.

be part of the same sand body, as could channel elements C and V. Given these caveats, the prole reveals much useful information about the construction of the Hawkesbury depositional system. The scales and, where available, the paleocurrent data for each element are given in Table 3. The last column in this table indicates the interpreted orientation of the outcrop relative to depositional dip (down-ow) and strike (across-ow). It is suggested that the 16 channel elements each correspond to the temporary position of a major channel within the Hawkesbury depositional system. More than one of these channels may have been active at a given time. None of the channel elements is completely dened within the available outcrop, despite the unusual length of the prole. This attests to the size of the Hawkesbury river system. Measured outcrop lengths range from 500 m for channel element C to 3200 m for channel element B, but it is known that element C continues for some distance farther to the north beyond the end of the prole, and, as noted above, the bounds of element B are difcult to determine, so these gures are not reliable as indicators of maximum and minimum widths. Channel element H probably terminates a short distance to the south of the end of the prole (it thins to a few meters at point 0 m), and at 600 m is the smallest of the ten elements A to J. Channel element J is bounded at both ends of the prole by relatively steeply dipping surfaces that are interpreted as channel cutbanks. The top of this element is not dened but, given the typical thickness of the channel elements, at between 18 and 22 m, it seems likely that the 18 m of channel element J represents most of it. Accordingly, the 1100 m width seems likely to represent nearly all of this element. Channel element D is 2700 m wide. This element thins near the south end of the prole and may be close to

FIG. 12.Oblique view along cliff showing sandstone bed consisting of trough crossbed cosets (St), constituting element TR(C). The sandstone has partly slumped into and partly scoured into underlying ne-grained ood deposits (Fl) of an abandoned channel-ll element FF(C).

its termination there. At the north end it is cut out by channel element J but, as noted above, it could be a continuation of channel element E (or even C), which would increase its maximum possible length to more than 4 km. The erosional relief on the fth-order surfaces provides an indication of the scour depth of the Hawkesbury channels, while the vertical height of identiable macroforms, such as DA and LA units, provides indications of the height of bars and indicates the minimum depth of channels undergoing depositional accretion. Deep scours commonly reect enhanced scour at channel conuences (Best 1987; Best and Ashworth 1997) and upstream from large bars (Cant 1976). Scour depths at such locations may be three to six times mean channel depth (Mosley 1976; Cant 1976; Best and Ashworth 1997). Cut-and-ll erosional relief on the fth-order surfaces in the Kurnell Peninsula prole is commonly as much as 10 m over lateral distances of a few hundred meters or less. The base of surface J rises 20 m between the 2820 and 3140 m points of the prole, and slightly more between 2820 and 2030. Both margins of this element are relatively steep (not forgetting the vertical exaggeration of Fig. 13), and may correspond to the cutbanks of the original channel. Even steeper cutbanks are present at the base of channel element G at 3400 and 3580 m. They indicate an erosional relief of up to 12 m on each margin of what appears to be a channel, some 180 m wide, cutting down into a thick overbank unit. This probably represents a channel in the process of avulsion across the oodplain (the FF(O) element at the top of channel element E) and is oriented at a high angle to the interpreted paleoow direction of element E. These lines of evidence suggest that the channel at the base of element G can be interpreted as a crevasse channel. Hollow elements (HO) are present in several places along the prole. Two have been identied in the 21002200 m interval. One, 10 m deep, appears to form an integral part of the base of channel element D, and the other, a short distance to the north, is truncated by the major surface at the base of channel element D and clearly predated the main constructional phase of that element. Two more hollow elements are present at 5100 (Fig. 10B). Where accretionary architectural elements (LA and DA) can be identied by stacking of third-order units or by the tracing of fourth-order surfaces they average about 10 m in thickness. One such element, at the 2330-m point in channel element D, is cut by a scour hollow about 4 m deep at its south end. In summary, the thicknesses of accretionary architectural elements (LA and DA units) indicate that channel bars ranged up to about 10 m in height. This height indicates the minimum constructional depth of the channels, although bankfull depth was undoubtedly somewhat greater. Scour depths ranged from 4 to 20 m. The relative arrangement of the major channel elements provides some insights into the larger-scale geomorphic history of the Hawkesbury river system. The order in which the channel elements were deposited indicates



FIG. 14.Example of detailed interpretation of the Kurnell Peninsula prole. Fifth-order elements are indicated by circled letters. Individual paleocurrent readings are shown, with the points of the arrows located at the point of measurement. Orientations are shown with respect to north towards the top of the page.

major lateral shifting in position of channels, although as noted above, the order in which the channel elements have been lettered on the prole may not correspond to the precise succession of events and, indeed, some of the channel elements may represent entirely different portions of the channel system, as the rivers underwent avulsion or river capture. The limited paleocurrent data available from the north end of the prole may reect channel sinuosity, or it may indicate that the channels underwent major shifts in transport direction as well as changes in position. For example, the vertical succession of channel elements A, C, E, and G in the 3500 4200 m portions of the prole is accompanied by shifts in mean channel direction of 40, 63, and 97 from one element to the next. A schematic interpretation of this succession of channel elements is shown in Figure 15. At the northwest end of the prole the succession of channel elements V, W, X, and Y is accompanied by shifts in mean channel direction of 206, 120, and 6, respectively. According to Cowan (1993) the Hawkesbury rivers were, in general, owing northeastward, from the craton across the Sydney Basin oblique to structural trends. In detail, however, the data from the Kurnell Peninsula
TABLE 3.Size and orientation of the major (fth-order) elements in the Kurnell Peninsula prole.
Paleocurrent Element A B C D E F G H I J U V W X Y Width (m) 600? 3200? 500 2700 900 1600 800 600 1300 1100 ? 1200 600 500 300 Max. Thickness (m) 8 20 8 20 22 20 18 11 20 18 10 10 13 15 11 Mean 257 297 082 360 097 n 10 5 3 9 12 Interpreted Outcrop Orientation strike ? strike strike dip ? strike ? ? strike ? dip dip oblique oblique

indicates that mean channel orientations ranged from west-southwest (channel element A) to east-southeast (channel elements J, V). The fth-order bounding surface at the base of each element represents channel scour, as a result of lateral channel shifting or avulsion. Both processes are autogenic and are interpreted to have occurred because of local slope advantages on the alluvial plain. Deposition within a given position of the channel resulted in accretion and aggradation up to a certain critical thickness, beyond which crevassing following a ood event or lateral channel-bank erosion led to ow diversion and the initiation of a new channel or channel belt (Jones and Schumm 1999). The fact that the fth-order channel elements all have maximum thicknesses within the 1822 m range suggests that there was a limit on the Hawkesbury alluvial plain beyond which one or more geomorphic thresholds was reached, one of which triggered the avulsion. The steep cutbanks at the base of channel element G (Figs. 13, 15), and the presence of the thick oodplain unit into which this element has incised, are both unusual features in the Hawkesbury Sandstone. Floodplain deposits tend to be minor components of braided uvial systems, but are not unknown (e.g., Reinfelds and Nanson 1993). Channel element CH (Fig. 13) is interpreted as a crevasse channel. Curracurrong The prole at Curracurrong is illustrated in Figure 16. An example of the photomosaic interpretation is shown in Figure 17. The prole is 770 m long and up to 35 m high. The cliff exposes a series of major channel elements bounded by fthorder surfaces. Ten of these have been dened and identied by the letters A to J. The large HO elements at the north end of the prole have not been given separate letters. Only one of the labeled channel elements, F, is completely exposed in cross section by the cliff. It is 450 m wide and up to 4 m thick. All the other elements are 770 m wide. Most of the elements are up to 10 m thick. Many consist of two or three macroforms superimposed vertically or laterally, as indicated by the fourth-order surfaces in Figure 16. Several of these surfaces are in part convex-up, indicating that they are preserved accretionary surfaces little modied by subsequent erosion. Ground observations along the base of the cliff indicate

112 135 342 101 095

14 6 15 7 5



FIG. 15.Schematic interpretation of the 35004000 m interval in the Kurnell Peninsula prole, showing schematic channel and bar reconstructions for the four successive elements at this location. The front face of each block diagram represents the exposure seen in the prole. Large arrows indicate paleoow direction.

that downstream accretion was an important process in the generation of these deposits. One of these is illustrated in Figure 8. In this case, individual crossbed sets indicate growth toward the north (right), in the same direction as the dip of the accretionary surfaces. Hollow elements (HO) are common in this prole. There is, in particular, a cluster of these elements at the north end of the prole. They are up to 10 m thick and 60 m wide, and they are superimposed on each other, indicating repeated scour at bar heads or channel conuences. In some cases fth-order channel elements are interpreted as being separated by thin mudstone units, as suggested by the prominence of some surfaces in the cliff sectionthey are indicated by the shadow of an overhang, where a prominent, resistant, basal sandstone unit overlies a recessive mudstone. Elsewhere, sandstone grain-size contrasts across the fth-order surface probably account for the weathering out of the surface. In a few cases these mudstones or grain-size contrasts are not present, and fthorder surfaces are then very difcult to trace. In a few cases, as indicated by the notation FF(O) in the upper parts of channel elements E and G in Figure 16, thicker oodplain units are indicated by darker, recessive intervals along the cliff. An example of the photomosaic from which Figure 16 was drawn is illustrated in Figure 17. Elements are labeled as in Figure 16, and the two lowest channel elements are subdivided into sub-elements based on their internal architectural characteristics. Channel element A rests on an extensive, recessive ne-grained unit interpreted as the ll of an abandoned

channel (FF(C)). The base of the channel element, sub-element A, consists of a relatively massive sandstone capped by a convex-upward surface, interpreted as a fourth-order surface capping a macroform unit, probably the deposit of a mid-channel bar form. Sub-element A onlaps the fourth-order surface and consists of stratication with a concave-up geometry. It is interpreted as an oblique cross section through a channel-ll succession, possibly developed as a result of convergent accretion of bars from opposite banks of the channel. Sub-element A is interpreted as a downstreamaccretion unit. Channel element B has been subdivided into four sub-elements. B is characterized by accretionary stratication, but the details of the bar type cannot be determined. Sub-element B is a small unit about 30 m long, with a concave-up base. It is probably a cross section through a chute channel that cut across the top of sub-element B. B is a large downstream-accretion unit, at least 330 m long. It extends across nearly half of the Curracurrong prole, from near the left end of Figure 17 to where it is cut out by the cluster of HO elements near the north end of the prole. Part of this DA unit is illustrated in Figure 8. Sub-element B, is bounded by a concave-up surface, and is interpreted as a hollow element formed at a site of secondary channel conuence.

Although the Hawkesbury Sandstone has long been interpreted as the deposit of a large braided river of the scale of the modern Brahmaputra,


FIG. 16.The prole at Curracurrong, shown with 5 vertical exaggeration. The prole is based on photographs of the cliff taken looking obliquely downward from a low ying aircraft. The downward bend in the prole at the right corresponds to an outward bend in the cliff face. Fifth-order elements are indicated by circled letters.



FIG. 17.Example of detailed photomosaic interpretation of part of the Curracurrong cliff prole. Fifth-order elements are labeled as in Figure 16. Sub-elements are shown with tick marks to facilitate discussion in the text. The DA element illustrated in Figure 8 is a part of element B, off this photograph to the north (right).

the basis for such an interpretation has always been limited, consisting mainly of remarks about the large scale of the crossbedding (Figs. 4, 5; see Conaghan and Jones 1975; Rust and Jones 1987). Early modern work on the bedforms, bars, and channels of the Brahmaputra River by Coleman (1969) provided some justication for this comparison. From echo-sounding proles, Coleman (1969) identied bedforms he termed dunes up to 7.6 m high, and sand waves up to 15 m high, although the latter are probably not simple trains of angle-of-repose bedforms (mesoforms) but the accretionary tops of bars (macroforms) which may be composed of stacked mesoforms. Ashworth et al. (2000), recorded dunes up to 3.5 m in height and they report records of other workers in the river that indicate dunes reaching 6 m in height. A comparison of the facies of the Hawkesbury Sandstone (Rust and Jones 1987) with those of the modern Brahmaputra river (e.g., Bristow 1987, 1993) reveals many similarities, but such a comparison is not denitive because the facies of most uvial systems are rather universal, reecting hydraulic conditions at and near the turbulent boundary layer that are not unique to any particular river planform or channel scale. More recent work on the morphology and dynamics of channels and large bar forms in the modern Brahmaputra River (e.g., Thorne et al. 1993; McLelland et al. 1999; Ashworth et al. 2000; Best et al. 2003) forms a better basis for the comparison, because these studies provide more information about the scale and dynamics of the large-scale features of the modern river, the large channels and giant bar forms. It was data of this kind for the Hawkesbury Sandstone that our research was primarily designed to provide. Nevertheless, the discussion presented here points up the limitations in the respective ancient and modern databases upon which a comparison between them must be based. Channels and bars in the Brahmaputra River (or Jamuna River, as it is locally called in Bangladesh) can be classied using a three-fold hierarchy (Bristow 1987). Bars of a given order are scaled to the channel of the same order. North of Dhaka the main channel belt (rst-order river) is up to 20

km wide, consisting of braided and anastomosing second-order channels typically 0.52 km wide (Fig. 18). Semipermanent islands (rst-order bars) up to about 8 km wide and 20 km long occur regularly along the river course. These are composed of amalgamated braid bars of second-order rank. Second-order channels are typically 1012 m deep, but scour depths of up to 50 m have been recorded (Best and Ashworth 1997). Maximum (bankfull) depths are reached for short periods during each monsoon ood. Bars typically range from half to slightly less than bankfull depth; therefore, in a channel 12 m deep, bars would typically be about 7 m high (C. Bristow, personal communication 2001). Second-order bars are 12 km wide and 36 km long. Ashworth et al. (2000) and Best et al. (2003) described the evolution of a second-order bar 4 km long, 1.5 km wide, and 12 m high. Width-to-depth ratios at bankfull ow approach 700 (Thorne et al. 1993). Observations by Bristow (1987, p. 68) suggest that it is the secondorder channels that control the internal stratigraphy of the resulting deposits. The major sand bodies that result would be fth-order units according to the architectural classication used here, and it is with these that the Hawkesbury Sandstone bodies should be compared. According to Bristow (1987) third-order channels are the product of ow across second-order bars at falling or low stage. Bristows (1993) facies model for bar-top sedimentation in the Brahmaputra River shows channels of this type approximately 100 m wide and up to 2 m deep. Sub-elements A and B in the Curracurrong prole (Fig. 17) may represent channels of this kind. The strongly seasonal discharge that characterizes the Brahmaputra River might be represented in the Hawkesbury Sandstone by the units of Sm, interpreted by Jones and Rust (1983) as the product of bank collapse following falls in river stage. Very little information on the architecture of preserved sand bodies in the Brahmaputra channel belt is available. A rst attempt to document the subsurface of a modern bar complex based on ground-penetrating radar



FIG. 18.Comparison of the architecture of the Hawkesbury Sandstone at Kurnell Peninsula with the dimensions of channels and bars of the modern Brahmaputra River. At top, a typical reach of the Brahmaputra (Jamuna) River of northern Bangladesh is shown next to a summary of the Kurnell Peninsula outcrop prole reduced to the same horizontal scale. Numbers 1 and 2 in circles indicate the rst- and second-order channels and within-channel bar complexes of the river, as classied by Bristow (1987). Note the scale of the interpreted crevasse channel in the Kurnell Peninsula prole. At bottom, the Kurnell Peninsula prole is shown at the same scale as a longitudinal cross-section through a modern bar in the Brahmaputra River, as reconstructed from GPR data.

data (GPR) was offered by Best et al. (2003), and the discussion provided here draws on this study and on others that have been carried out on the surface bed- and bar-forms. The following considerations provide the basis for an estimate of sand-body dimensions. In its simplest condition the evolution of a braided channel can be considered as the development of opposite-facing low-sinuosity meanders migrating away from a central (midchannel) bar (Bridge 1993b). The work of Ashworth et al. (2000) explicitly ruled out this mode of evolution in the case of the bar they studied, although they made a comparison with the small bar in the Calamus River, Nebraska, analyzed by Bridge et al. (1998), which the latter demonstrated to have grown by a comparable pattern of lateral and downstream accretion from an upstream nucleus. Where bar migration is symmetrical, as proposed by Bridge (1993b), channel scour would be expected to sweep out an erosional channel form approximating the width of two channels plus

the intervening bar. Assuming two channels of second-order Brahmaputra scale, each 2 km wide, and a mid-channel bar also 2 km wide, if both channels were lled prior to abandonment this theoretically could generate a second-order sand body bounded by a fth-order surface on the order of 6 km wide. With an average depth of 12 m such a sand body would have a width-to-depth ratio of 500. If a channel remained in approximately the same position, undergoing aggradation until it switched elsewhere as a result of some geomorphic event, the resulting sand body would be even deeper and wider. However, this scenario is quite speculative. Several groups of researchers have demonstrated patterns of active anabranch migration and bar growth and erosion in the Brahmaputra/Jamuna River (Thorne et al. 1993; Ashworth et al. 2000) which indicate that sand bodies of the full theoretical width estimated here may never develop. Sand bodies bounded by surfaces of fth-order rank are likely to be substantially less



FIG. 19.Comparison of sandbody scales between the modern Brahmaputra River and the Hawkesbury Sandstone at the Kurnell Peninsula. The Brahmaputra architecture reconstruction is speculative, based on data provided by Bristow (1987), Thorne et al. (1993), Ashworth et al. (2000) and Best et al. (2003).

than 6 km wide. The nal preserved architecture of sand bodies of the type described by Ashworth et al. (2000) would depend on the balance between (1) lateral growth of the bar under conditions of anabranch migration, and (2A) erosional incision brought about by events of avulsive anabranch switching or (2B) migration and lateral erosion of an anabranch from another location within the channel belt. Final preserved sand body widths are presumably somewhere between the hypothetical maximum of 6 km and the width of individual barsa minimum of 1 km. The bar documented by Best et al. (2003) is approximately 3 km long and 1 km wide, and it contains a complex pattern of internal third-order upstream, downstream, and lateral-accretion surfaces (a simplied rendering of this architecture, adapted from Best et al. 2003, is included in Fig. 18). Even less information is available regarding the lengths of preserved sand bodies in the modern Brahmaputra oodplain. Channel reaches migrate and avulse laterally, and given the moderate sinuosity of the modern river, with channel reaches locally oriented at a high angle to the average downstream direction, this means that at least some of the erosional effects of channel migration and avulsion result in the truncation of sand bodies along the downstream trend. Intuitively we might expect sand-body lengths to be greater than widths, but there is no reason to predict that the downstream lengths of second-order sand bodies in the modern Brahmaputra oodplain would have dimensions signicantly different than the widths of the same sand bodies. Given the dynamic nature of this river, the 3 km 1 km bar mapped by Best et al. (2003) will probably not survive intact into the geological record but will undergo both further depositional extension and subsequent fragmentation by channel migration and avulsion. Turning to the Hawkesbury Sandstone: despite the length of the two studied proles, data regarding sand body dimensions is very sparse. The exposures of the sand bodies are at varying orientations with respect to paleoow, and in most cases this orientation is not known. The outcrop lengths of the sand bodies therefore represent lengths, widths, or oblique cross sections (relative to paleoow) through each channel and bar complex. These arguments about the scale of the modern Brahmaputra sand bodies and the signicance of outcrop lengths of the Hawkesbury Sandstone sand bodies permit only an order-of-magnitude comparison between them. A comparison with the exposure lengths of the sand bodies in the Hawkesbury Sandstone suggests that the Hawkesbury Sandstone units are comparable, but perhaps smaller (Figs. 18, 19). The largest sand body for which the width is reasonably certain is channel element D, at 2.7 km wide, in

the Kurnell prole. The extremely limited paleocurrent data available from the crossbedding in this body suggests that the Kurnell Prole is oriented approximately perpendicular to the ow direction within this channel, so that the 2.7 km gure is an approximate measure of sand body width. This is in the middle of the range of possible widths of a typical second-order Brahmaputra sand body (Fig. 19). Other Hawkesbury Sandstone bodies appear to be somewhat smaller, although most are incomplete (Table 3). Fifth-order channel elements in the Hawkesbury Sandstone are typically 1820 m thick, which is considerably thicker than the average channel depth of 12 m for the Brahmaputra, but most of the Hawkesbury Sandstone channel elements consist of two or three superimposed macroforms, each 510 m thick, which again suggests a scale for constructional channel depth comparable but perhaps slightly smaller than that of the Brahmaputra. Bristow (1987) noted that macroforms in the Brahmaputra River typically range from between one-half and just less than total bankfull depth. Scour depths in the Hawkesbury Sandstone, as indicated by the height of HO elements, range up to about 20 m, less than half the maximum 50 m scour depths recorded in the Brahmaputra. However, some of the HO elements may be top-truncated, and some may represent scour at anabranch conuences rather than the junction between major tributaries. Note that the interpreted crevasse channel in the Kurnell Peninsula prole (channel element G in Fig. 13) is comparable in scale to the smallest of the channels shown in the map of the modern Brahmaputra River (Fig. 18). The conclusion is that earlier interpretations of the Hawkesbury Sandstone as the deposit of a very large river system are correct. The major parameters that can be estimated for the Hawkesbury Sandstone rivers channel width, constructional channel depth, and scour depthshow that maximum values are within the mid range of those of the modern Brahmaputra/Jamuna, as estimated from modern channel and bar patterns in northern Bangladesh, suggesting that the Hawkesbury river system was large but measurably smaller than the modern Brahmaputra. However, even the availability of the exceptionally long Kurnell Peninsula prole is adequate only as a basis for tentative interpretations of scale, given the large size of the ancient rivers that deposited the Hawkesbury Sandstone. The 75100 km preserved width of the Hawkesbury Sandstone depositional system indicates that the river or rivers responsible for its deposition were free to comb across a wide, at, alluvial plain. The marked changes in paleocurrent direction recorded in successive channel-ll deposits in the Curracurrong and Kurnell sections attest to the lateral mobility of the rivers and the very low depositional gradient in this part of the Sydney Basin.




Farther north, marine incursions have been recorded within the Hawkesbury Sandstone (e.g., Herbert 1997), but they are not known to extend into the central part of the basin. Judging by the occurrence of compound dune (CD) elements, it seems likely that the Hawkesbury rivers were in part tidally inuenced within our project area, but we have not yet been able to discern any stratigraphically consistent patterns of occurrence of this element in the study area that would permit an interpretation in terms of changing accommodation and transgressionregression of the kind that Shanley and McCabe (1994) used in their sequence model for non-marine deposits. In conclusion, this preliminary architectural study of the Hawkesbury Sandstone has demonstrated that in the project area south of Sydney, channel lls and their component bars were large but measurably smaller in width and depth than those in the modern Brahmaputra River. Further study is required to explore the regional consistency of these estimates.

One of the outcomes of this study has been a realization of the limitations of the databases upon which modern-to-ancient comparisons are based in the uvial realm. While individual lithofacies and vertical proles through bar and channel deposits may readily be comparedand have been done so repeatedly since the development of the facies-model concept in the 1960ssuch comparisons are limited in their value by the nearly universal nature of such deposits. Bedform character reects the common physics of turbulent ow (Ashley 1990), and bar development is increasingly becoming recognized as a largely scale-independent process. However, the database on large-scale uvial architecture, especially sandbody width and length, remains extremely small, and more studies of this type (e.g., see Robinson and McCabe 1997) need to be carried out. While the Hawkesbury Sandstone remains a focus of interest because of its outstanding outcrop character as the deposit of a large river, the nature of its facies and architectural assemblage raises questions about how typical this unit is as the deposit of a large braided uvial system. Sandstone with primary current lineation (lithofacies Sh) is virtually absent, possibly reecting a particular combination of high stream power and coarse grain size, whereas the hollow element (element HO), a product of scour at channel conuences, might have been expected to be much more abundant in a river system as vigorous as the one that deposited the Hawkesbury Sandstone is interpreted to have been. We can only ask why, while noting that the recognition of scour-ll deposits (the HO architectural element) is a relatively recent addition to our standard suite of uvial units. In addition to a need for further bar- and channel-scale architectural analysis it would be instructive to investigate the details of the internal sequence stratigraphy of the unit, to determine whether the occurrence of tidal inuences reveals any systematic regional or vertical (time-related) variations in accommodation during Hawkesbury sedimentation. Such regional studies are also required to evaluate the signicance of the negrained channel ll and thick ripple-cross-stratied units exposed at The Cobblers. These are the only locations in the project area where such facies are exposed, and they may also provide important information on accommodation changes if they can be situated within a regional stratigraphic context.

This research was supported by the Natural Science and Engineering Research Council, Canada (ADM) and the Sustainable Earth Research Centre at the University of Wollongong (BGJ). The comments and suggestions on various versions of this paper by Jim Best, Charlie Bristow, Brian Willis, Colin North, Peter Friend, and Phil Ashworth have led to many useful renements and, we hope, improvements in the nal product. Our thanks to all of these individuals. Thanks are due to Charlene Miall for assistance in the eld.

ALLEN, J.R.L., 1980, Sand waves: a model of origin and internal structure: Sedimentary Geology, v. 26, p. 281328. ALLEN, J.R.L., 1983, Studies in uviatile sedimentation: bars, bar complexes and sandstone sheets (low-sinuosity braided streams) in the Brownstones (L. Devonian), Welsh Borders: Sedimentary Geology, v. 33, p. 237293. ALLEN, J.R.L., 1984, Sedimentary Structures; Their Character and Physical Basis: Amsterdam, Elsevier, Developments in Sedimentology, v. 30, 663 p. ASHLEY, G.M., 1990, Classication of large-scale subaqueous bedforms: a new look at an old problem: Journal of Sedimentary Petrology, v. 60, p. 160172. ASHLEY, G.M., AND DUNCAN, I.J., 1977, Discussionthe Hawkesbury Sandstone: a critical review of proposed environmental models: Geological Society of Australia, Journal, v. 24, p. 117120. ASHWORTH, P.J., BEST, J.L., RODEN, J.E., BRISTOW, C.S., AND KLAASSEN, G.J., 2000, Morphological evolution and dynamics of a large, sand braid-bar, Jamuna River, Bangladesh: Sedimentology, v. 47, p. 533555. BEST, J.L., 1987, Flow dynamics at river channel conuences: implications for sediment transport and bed morphology, in Ethridge, F.G., and Flores R.M., eds., Recent and Ancient Nonmarine Depositional Environments: SEPM, Special Publication 31, 2735. BEST, J.L., AND ASHWORTH, P.J., 1997, Scour in large braided rivers and the recognition of sequence stratigraphic boundaries: Nature, v. 387, p. 275277. BEST, J.L., ASHWORTH, P.J., BRISTOW, C.S., AND RODEN, J., 2003, Three-dimensional sedimentary architecture of a large, mid-channel sand braid bar, Jamuna River, Bangladesh: Journal of Sedimentary Research, v. 73, p. 516530. BRIDGE, J.S., 1993a, Description and interpretation of uvial deposits: a critical perspective: Sedimentology, v. 40, p. 801810. BRIDGE, J.S., 1993b, The interaction between channel geometry, water ow, sediment transport and deposition in braided rivers, in Best, J.L., and Bristow, C.S., eds., Braided Rivers: Geological Society of London, Special Publication 75, p. 1371. BRIDGE, J.S., COLLIER, R., AND ALEXANDER, J., 1998, Large-scale structure of Calamus River deposits (Nebraska, USA) revealed using ground penetrating radar: Sedimentology, v. 45, p. 977986. BRISTOW, C.S., 1987, Brahmaputra River: Channel migration and deposition, in Ethridge, F.G., Flores, R.M., and Harvey, M.D., eds., Recent Developments in Fluvial Sedimentology: SEPM, Special Publication 39, p. 6374. BRISTOW, C.S., 1993, Sedimentary structures exposed in bar tops in the Brahmaputra River, Bangladesh, in Best, J.L., and Bristow, C.S., eds., Braided Rivers: Geological Society of London, Special Publication 75, p. 277289. BRISTOW, C.S., BEST, J.L., AND ROY, A.G., 1993, Morphology and facies models of channel conuences, in Marzo, M., and Puigdefa bregas, C., eds., Alluvial Sedimentation: International Association of Sedimentologists, Special Publication 17, p. 91100. CANT, D.J., 1976, Braided stream sedimentation in the South Saskatchewan River [Unpublished Ph.D. thesis]: McMaster University, Hamilton, 125 p. COLEMAN, J.M., 1969, Brahmaputra River: channel processes and sedimentation: Sedimentary Geology, v. 3, p. 129239. CONAGHAN, P.J., AND JONES, J.G., 1975, The Hawkesbury Sandstone and the Brahmaputra: a depositional model for continental sheet sandstones: Geological Society of Australia, Journal, v. 22, p. 275283. CONOLLY, J.R., 1969, Models for Triassic deposition in the Sydney Basin, in Brown, D.A., ed., Proceedings of Specialists Meeting Held at Canberra, 2531 May 1968: Geological Society of Australia, Special Publication 2, p. 209223. CONOLLY, J.R., AND FERM, J.C., 1971, PermoTriassic sedimentation patterns, Sydney Basin, Australia: American Association of Petroleum Geologists, Bulletin, v. 55, p. 20182032. COWAN, E.J., 1991, The large-scale architecture of the uvial Westwater Canyon Member, Morrison Formation (Jurassic), San Juan Basin, New Mexico, in Miall, A.D., and Tyler, N., eds., The Three-Dimensional Facies Architecture of Terrigenous Clastic Sediments, and its Implications for Hydrocarbon Discovery and Recovery: SEPM, Concepts in Sedimentology and Paleontology, no. 3, p. 8093. COWAN, E.J., 1993, Longitudinal uvial drainage patterns within a foreland basin-ll: Permo Triassic Sydney Basin, Australia: Sedimentary Geology, v. 85, p. 557577. HERBERT, C., 1997, Sequence stratigraphic analysis of Early and Middle Triassic alluvial and estuarine facies in the Sydney Basin, Australia: Australian Journal of Earth Sciences, v. 44, p. 125143. JACKSON, R.G., II, 1975, Hierarchical attributes and a unifying model of bed forms composed of cohesionless material and produced by shearing ow: Geological Society of America, Bulletin, v. 86, p. 15231533. JONES, B.G., AND RUST, B.R., 1983, Massive sandstone facies in the Hawkesbury Sandstone, a Triassic uvial deposit near Sydney, Australia: Journal of Sedimentary Petrology, v. 53, p. 12491260. JONES, L.S., AND SCHUMM, S.A., 1999, Causes of avulsion: an overview, in Smith, N.D., and Rogers, J., eds., Alluvial Sedimentology VI: International Association of Sedimentologists, Special Publication 28, p. 171178. MCLELLAND, S.J., ASHWORTH, P.J., BEST, J.L., RODEN, J., AND KLAASSEN, G.J., 1999, Flow structure and transport of sand-grade suspended sediment around an evolving braid bar, Jamuna River, Bangladesh, in Smith, N.D., and Rogers, J., eds., Alluvial Sedimentology VI: International Association of Sedimentologists, Special Publication 28, p. 4357. MIALL, A.D., 1985, Architectural-element analysis: A new method of facies analysis applied to uvial deposits: Earth-Science Reviews, v. 22, p. 261308. MIALL, A.D., 1988, Reservoir heterogeneities in uvial sandstones: lessons from outcrop studies: American Association of Petroleum Geologists, Bulletin, v. 72, p. 682697.


MIALL, A.D., 1996, The Geology of Fluvial Deposits: Sedimentary Facies, Basin Analysis and Petroleum Geology: Heidelberg, Springer-Verlag Inc., 582 p. MOSLEY, M.P., 1976, An experimental study of channel conuences: Journal of Geology, v. 84, p. 535562. REINFELDS, I., AND NANSON, G., 1993, Formation of braided river oodplains, Waimakariri River, New Zealand: Sedimentology, v. 40, p. 11131127. ROBINSON, J.W., AND MCCABE, P.J., 1997, Sandstone-body and shale-body dimensions in a braided uvial system: Salt Wash Sandstone Member (Morrison Formation), Gareld County, Utah: American Association of Petroleum Geologists, Bulletin, v. 81, p. 12671291. RUST, B.R., AND JONES, B.G., 1987, The Hawkesbury Sandstone south of Sydney, Australia: Triassic analogue for the deposit of a large braided river: Journal of Sedimentary Petrology, v. 57, p. 222233.


SHANLEY, K.W., AND MCCABE, P.J., Perspectives on the sequence stratigraphy of continuous strata: American Association of Petroleum Geologists, Bulletin, v. 78, p. 544568. THORNE, C.R., RUSSELL, A.P.G., AND ALAM, M.K., 1993, Planform pattern and channel evolution of the Brahmaputra River, Bangladesh, in Best, J.L., and Bristow, C.S., eds., Braided Rivers: Geological Society of London, Special Publication 75, p. 257276. WEBER, K.J., AND VAN GEUNS, L.C., 1990, Framework for constructing clastic reservoir simulation models: Journal of Petroleum Technology, v. 42, p. 12481253, 12961297. WILLIS, B.J., 1989, Palaeochannel reconstruction from point bar deposits: a three-dimensional perspective: Sedimentology, v. 36, p. 757766. Received 17 June 2002; accepted 15 November 2002.