The sedimentology and alluvial architecture of ... - Wiley Online Library

Sedimentology (2006) 53, 413–434

doi: 10.1111/j.1365-3091.2005.00769.x

The sedimentology and alluvial architecture of the sandy braided South Saskatchewan River, Canada G . H. SAMBROO K SMITH*, P. J. ASHWORTH , J. L. BEST, J. WOODWARD§ and C. J. SIM PSON– *School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK (E-mail: [email protected]) Division of Geography, School of Environment, University of Brighton, Lewes Road, Sussex BN2 4GJ, UK Earth and Biosphere Institute, School of Earth and Environment, University of Leeds, Woodhouse Lane, Leeds, West Yorkshire LS2 9JT, UK §Division of Geography, School of Applied Science, Northumbria University, Ellison Building, Newcastle upon Tyne NE1 8ST, UK –Department of Geography, Simon Fraser University, 8888 University Drive, Burnaby, BC, Canada V5A 1S6 ABSTRACT

Ground penetrating radar (GPR) surveys of unit and compound braid bars in the sandy South Saskatchewan River, Canada, are used to test the influential facies model for sandy braided alluvium presented by Cant & Walker (1978). Four main radar facies are identified: (1) high-angle (up to angle-of-repose) inclined reflections, interpreted as having formed at the margins of migrating bars; (2) discontinuous undular and/or trough-shaped reflections, interpreted as cross-strata associated with the migration of sinuous-crested dunes; (3) lowangle (< 6) reflections, interpreted as formed by low-amplitude dunes or unit bars as they migrate onto bar surfaces; and (4) reflections of variable dip bounded by a concave reflection, interpreted as being formed by the filling of channel scours, cross-bar channels or depressions on the bar surface. The predominant vertical arrangement of facies is discontinuous trough-shaped reflections at the channel base overlain by discontinuous undular reflections, overlain by low-angle reflections that dominate the deposits near the bar surface. High-angle inclined reflections are only found near the surface of unit bars, and are of relatively small-scale (< 0Æ5 m), but can be found at a greater range of depths within compound bars. The GPR data show that a high spatial variability exists in the distribution of facies between different compound bars, with facies variability within a single bar being as pronounced as that between bars. Compound bars evolve as an amalgamation of unit bars and other compound bars, and comprise a facies distribution that is representative of the main bar types in the South Saskatchewan River. The GPR data are compared with the original model of Cant & Walker (1978) and reveal a much greater variability in the scale, proportion and distribution of facies than that presented by Cant & Walker (1978). Most notably, high-angle inclined strata are over-represented in the model of Cant and Walker, with many bars being dominated by the deposits of low- and high-amplitude dunes. It is suggested that further GPR studies from a range of braided river types are required to properly quantify the full range of deposits. Only by moving away from traditional, highly generalized facies models can a greater understanding of braided river deposits and their controls be established. Keywords Alluvial facies, ground penetrating radar, sandy braided river, sedimentary architecture, South Saskatchewan River. 2006 The Authors. Journal compilation 2006 International Association of Sedimentologists

413

414

G. H. Sambrook Smith et al.

INTRODUCTION Many authors have proposed facies models for sandy braided rivers from study of both contemporary and ancient rivers (Collinson, 1970; Smith, 1971; Cant & Walker, 1978; Miall, 1978; Blodgett & Stanley, 1980; Ethridge & Flores, 1981; Allen, 1983; Crowley, 1983; Bridge et al., 1986, 1998; Bristow, 1987, 1993; Bridge, 1993; Ashworth et al., 2000; Bridge & Tye, 2000; Best et al., 2003; Skelly et al., 2003), with one of the most influential facies models being that of Cant & Walker (1978). This model (derived from the same reach as is reported in this paper) was based on repeat aerial photographs, surface mapping, box coring, description of cutbanks and shallow trenches, echo sounding and measurements of bedforms within the channels of the South Saskatchewan River, Canada, together with examination of the Devonian Battery Point Forma-

C

tion (see also Cant, 1976, 1978; Cant & Walker, 1976). Cant & Walker (1978) proposed that the South Saskatchewan channel-belt could be characterized by three different facies profiles (Fig. 1): ‘sand flat’ (profile A), ‘channel’ (profile C) and a combination of the two termed ‘mixed influence’ (profile B). ‘Sand flats’ are large expanses of sand, that may be 50 m to 2 km in length when exposed at low flow stages, occupy up to 80% of the channel-belt width and typically have minor channels crossing them (Fig. 1). Cant & Walker (1978) envisaged that such sand flats originate from deposition on the tops of ‘cross-channel bars’. These were judged to form by accretion at the downstream margin of slipface-bounded bars, which are usually found in zones of flow expansion (Fig. 1), are commonly the width of the primary channel and may be up to hundreds of metres long. The key differences between facies profiles A and C proposed by Cant & Walker

B

A

Fig. 1. Three-dimensional block diagram and vertical facies profiles redrawn from Cant & Walker (1978), showing their interpretation of the key morphological and sedimentological features of the sandy braided South Saskatchewan River. Note that no scale was included on the 3-D diagram in the original paper of Cant & Walker (1978). 2006 The Authors. Journal compilation 2006 International Association of Sedimentologists, Sedimentology, 53, 413–434

Sedimentology of the South Saskatchewan River (1978) are the abundance of deposition attributed to bars (represented as large-scale planar crossstratification) and dunes (represented as mediumscale, trough cross-stratification; see Fig. 1). The model of Cant & Walker (1978) for deposition in the South Saskatchewan River has been highly influential and adopted by a number of authors as a suitable analogue for ancient sandy braided river deposits (Miall, 1978; Allen, 1983; Miall & Jones, 2003). However, although the model of Cant & Walker (1978) provides detail down to 5 m below the bar surface, most of the data used in its derivation were collected within only the top 0Æ5 m, with no detailed quantification of the deeper sub-surface. Data between this shallow depth and down to the thalweg were predominantly based on extrapolation of observations of surface processes, sections taken in shallow trenches, box-cores and examination of cutbank exposures. Additionally, it is also unclear whether the facies models proposed by Cant & Walker (1978) can be applied to all types of bar morphology (e.g. symmetrical or asymmetrical, attached or isolated), or in a reach with bars that may have very different depositional histories (e.g. unit or compound bars). A critical reexamination of the model of Cant & Walker is thus long overdue. It is now possible to provide much higher resolution records of sub-surface alluvial stratigraphy in sandbed rivers by using ground penetrating radar (GPR) (Bridge et al., 1998; Bristow et al., 1999; Ferguson & Brierley, 1999; Fielding et al., 1999; Lehmann & Green, 1999; Vandenberghe & van Overmeeren, 1999; van Dam & Schlager, 2000; Best et al., 2003; Skelly et al., 2003; Woodward et al., 2003; Neal, 2004). The present paper reports on the use of GPR to describe and quantify the deposits of several bars within the South Saskatchewan River. Aerial photographs taken before and after the GPR survey are used to identify surface morphological changes that are then related to the subsurface sedimentary architecture, thus allowing linkage of channel change with depositional structure. Specific objectives of this paper are to: (1) describe the principal morphological features of a 10 km long reach of the South Saskatchewan River; (2) classify the dominant depositional sedimentary facies and alluvial architecture for a range of bar types with different depositional histories; (3) investigate the link between the depositional architecture and the channel and bar dynamics; and (4) critically analyse the Cant & Walker (1978) facies model

415

for the South Saskatchewan River with these new data.

STUDY SITE The South Saskatchewan River originates in the Rocky Mountains, Alberta, Canada, and flows eastwards into Lake Diefenbaker, the downstream end of which is 25 km upstream of the study reach at Outlook (Fig. 2). The river is incised by up to 30 m into Cretaceous shales and sandstones and Quaternary deposits. At Outlook, the channel belt is approximately 0Æ6 km wide, has an average bed slope of 0Æ0003 (Cant, 1976) and a braided planform. The braiding index (the number of active main channels per cross-stream transect at low-flow stage) within the 10 km long study reach is 2Æ2. Fifty-two grain-size samples taken from bar surfaces and cutbanks in the study reach show the D50 grain-size ranges from 0Æ22 to 0Æ44 mm, with a mean of 0Æ30 mm, identical to that reported by Cant (1976). Clay is rarely found within the sediments (< 1% by weight). Gravel is also rare but can be found in channel thalwegs, with cobbles and boulders near some cutbanks where the river has eroded into Quaternary sediments. Scrub grass vegetation, willow bushes and small trees stabilize bars within the channel and on the floodplain. The main (or first-order) channels are 2–5 m deep and 50–150 m wide. Cant & Walker (1978) used echo-sounding to show that the deepest channels are dominated by dunes that may be up to 1Æ5 m high during floods, but are more commonly 0Æ3–0Æ5 m high. Ripples are ubiquitous in shallow areas and on bartops, whilst aeolian reworking of bartop surfaces commonly creates both aeolian ripples and barchanoid dunes. Winter conditions result in bar surfaces and smaller channels being covered in ice. However, flow continues beneath the ice cover and the major channels remain ice-free. Cant (1976) noted that the main impact of ice is to immobilize the compound bars for the winter months. In 1967, the South Saskatchewan River was impounded by the Gardiner Dam, creating Lake Diefenbaker (Fig. 2), which subsequently caused incision of 0Æ5 m up to 5 km downstream of the dam. A series of permanent benchmarks and cross-sections were established by Environment Canada downstream from the dam and have been surveyed pre-dam (1964) and up to 16 times since then. The town of Outlook is located between benchmarks 17Æ4 and 21Æ0, which are 28Æ0 and 33Æ8 km downstream of the dam respectively.

2006 The Authors. Journal compilation 2006 International Association of Sedimentologists, Sedimentology, 53, 413–434

416



Sedimentology of the South Saskatchewan River Galay et al. (1985) reported that bed degradation within 8Æ0 km of the dam had slowed significantly by 1979, and that by this time the river had largely adjusted to the introduction of the dam. This conclusion is corroborated by Helfrick (1993) who reported that degradation was only significant within 5Æ0 km of the dam and that beyond this distance any changes in minimum and mean bed elevation could be accounted for by migrating sand bars. The most recent resurveys in the summer of 2002 by Phillips (2003) conclude that the study reach reported herein has not experienced any statistically significant change in mean bed elevation since building of the dam (Fig. 3). The Gardiner dam has reduced some of the very largest flood events, with mean annual peak discharge pre- and post-dam being 1536 and 595 m3 sec)1 respectively. The mean annual discharge pre- and post-dam is 280 and 203 m3 sec)1 respectively, and most bars become overtopped at 230 m3 sec)1. Thus, although the very highest discharges do not occur anymore, the flow regime still causes bedload transport over the entire braidplain during floods and many of the channels are active for large parts of the year. Thus, the more moderate and frequently occurring flow events continue to shape the channel, and ensure the South Saskatchewan River near Outlook remains an active braided system and that the results reported herein can thus be applied to other braided rivers.

DATA COLLECTION

Ground penetrating radar Approximately 3Æ5 km of GPR profiles were collected from recently deposited unit and compound bars (Fig. 2; see definitions of bars later) in June 2000, and Table 1 provides a summary of the GPR methodology employed (a detailed description is given in Woodward et al., 2003). Data were collected with common offset (CO) surveys using a PulseEKKO 100 radar system (Sensors & Software Inc., Mississauga, Canada). Antennae (200 MHz) were fixed 0Æ75 m apart on a purpose-

417

built plastic sledge and moved perpendicular to the profile in step-mode with a step spacing of 0Æ10 m and trace stack of 64. Twelve common midpoint (CMP) surveys allowed calculation of the mean radar wave velocity as 0Æ051 ± 0Æ006 m nsec)1. GPR profiles were processed using Gradix 1.10 software (Interpex, Golden, CO, USA) and included time-zero correction, dewow and bandpass filtering, background removal, application of gains, elevation statics and depth conversion. Migration algorithms did not optimize the radar signal and, therefore, were not applied. The sedimentological interpretation of the radar facies is based on ground-truth control from a series of GPR lines over an exposed bar cutface (Fig. 4; Woodward et al., 2003) and shallow trenching of profiles immediately after data collection (Fig. 5). The cut-face surveys showed that reflections in the GPR profile are caused by both clay drapes/layers, varying in thickness from < 0Æ02 to 0Æ10 m, as well as changes in sand grain-size, for example, from 0Æ30 to 0Æ23 mm. This subtle variability in grain-size is caused by grain-size sorting on individual cross-strata, low flow deposition, reworking by channel erosion and the presence of gravel lags. As an example, Fig. 5 shows inclined reflections in the GPR survey, up to the angle-of-repose, that are produced by grain-size differences in cosets produced at a bar margin, which can be seen both in planform on the surface and in a shallow trench cross-section.

Other survey techniques The topography of each bar and all GPR lines was surveyed using a total station, and surveys of the GPR lines were used for topographic correction of the GPR profiles. To aid interpretation of the GPR surveys, the evolution of key morphological features along the channel reach was analysed using rectified 1:10,000 scale aerial photographs flown along the river course before (April 2000) and after (October 2000) the main GPR data collection period. Information on longer-term evolution was obtained by comparing the aerial photographs from 2000 with

Fig. 2. Location map of the 10 km long study reach on the South Saskatchewan River near Outlook, Saskatchewan. This reach is the same as that studied by Cant & Walker (1978). Also shown is the location of all 200 MHz ground penetrating radar profiles undertaken on the bars studied. All survey lines are given a three digit code; the first letter refers to the specific bar (A–E), the second letter indicates whether the line was taken down-bar (X) or across-bar (Y). The number discriminates between the different X and Y lines on each bar. Bar edges shown here were defined by the water edge at the time of survey. Dashed line at edge of Bar E shows the approximate edges of the unit Bar Ea. 2006 The Authors. Journal compilation 2006 International Association of Sedimentologists, Sedimentology, 53, 413–434

418


Fig. 3. Mean bed elevation change (in m) on the South Saskatchewan River at ranges 17Æ4 (28.0 km downstream of dam) and 21Æ0 (33Æ8 km downstream of dam) from 1964 to 2002. The vertical dashed line is 1967, the year of dam completion. Horizontal lines mark the approximate mean bed elevation in 1967. Plots adapted from Phillips (2003).

Table 1. Summary of the methodology used in the collection, processing and ground-truth control of the South Saskatchewan River GPR data. Data collection GPR unit Antennae frequency (MHz) Antennae separation (m) Station spacing (m) Mode of data collection Stacks Processing Dewow (MHz) Drift removal Set time-zero Bandpass filter (MHz)

Background removal Depth conversion velocity (m nsec)1) Elevation statics AGC Gain (nsec) Ground-truthing Cut-face interpretation

PulseEKKO 100 200 0Æ75 0Æ10 Stop-and-collect 64 8Æ71 Yes Yes Trapezoidal with gates at 50–100–220–440 200 trace window 0Æ05 Yes 25 Yes

images from 1971 published in Cant & Walker (1978).

BAR TYPES AND DYNAMICS A bar is defined herein as a bedform whose length is proportional to channel width and whose height is comparable with the mean depth of the formative flow (ASCE, 1966). Figure 6 illustrates the key morphological features of bars that are common in the South Saskatchewan River: two

principal bar types are recognized here as either ‘unit’ or ‘compound’ bars. Unit bars are defined (after Smith, 1978) as having a shape that remains relatively unmodified during migration, and being simple forms that are not amalgamated/ superimposed upon other bar forms. Smith (1978) identified four main shapes of unit bar, although in the South Saskatchewan River the overwhelming majority of unit bars have a lobate planform with a slipface (up to the angle-of-repose) at their downstream margin. This style of unit bar would be termed a ‘transverse bar’ in the scheme of Smith (1978). Based on the work of Bridge (2003), compound bars are defined herein as forms that comprise more than one unit bar and evolve through several erosion and deposition events. Hence, compound bars possess a more complicated history than unit bars that is reflected in their wider range of planform shapes. Other bedforms, most typically ripples and dunes, are also superimposed on unit and compound bars. Cant & Walker (1978) also refer to ‘sand waves’ that are common in the South Saskatchewan River and are characterized by a high wavelength:height ratio. However, Allen (1982), Ashley (1990) and Bridge (2003) argue that such features are dunes and this terminology is adopted herein.

Unit bars Unit bars (labelled UB in Fig. 6) typically have a lobate planform, with their highest point at the downstream end of the bar that terminates in an avalanche face. Cant & Walker (1978) classified this bar type as a cross-channel bar. Unit bars have little topographic relief above the water


Sedimentology of the South Saskatchewan River

419

A

B

C Fig. 4. Ground-truth control of ground penetrating radar (GPR) using: (A) photographic montage of the cut-face, (B) the GPR profile for the cut-face and (C) the interpretation of the GPR profile. SHR1: interface between organic-rich soil horizon and underlying sediment. DR1: 0Æ02–0Æ10 m wide clay drape. DR2: clay drape < 0Æ02 m thick. DR3: erosion surface between depositional units, defined by an irregular interface between the units above and below the contact. DR4: an erosional interface defined by a marked change in grain-size (0Æ30–0Æ23 mm) and depositional characteristics (low-angle to angle-of-repose cross-strata). SHR2: thin (< 0Æ01 m) gravel lag. SHR3: thick (> 0Æ10 m) layer of clay at the base of the cut-face. Figure reproduced from Woodward et al. (2003).

surface, and can possess heights equivalent to the channel depth (i.e. up to 1Æ5–3 m). The area of the smallest unit bars is 100 m · 100 m, increasing in size to that of some of the smaller compound bars described below. For example, the maximum length of unit bars was 300 m during the study period. Unit bars experience significant change over time periods of 1 year, with the highest rates of downstream migration (20–130 m between April and October 2000) being recorded for small unit bars (e.g. £ 150 m long) in the largest channels. Unit bars in the South Saskatchewan River represent an early stage in the development of compound bars, as discussed below and shown in the braid bar model of Bridge (1993).

Compound bars Although displaying a wide range in planform shape, a distinctive type of compound bar

(labelled CBl in Fig. 6) is that with downstream elongated limbs or horns (Cant & Walker, 1978; Ashworth et al., 2000). These compound bars appear to be generated from two to four unit bars, in which one unit bar forms the central core that then promotes accretion from additional unit bars on either side, thus forming the characteristic limbs. Compound bars can develop an asymmetric morphology (labelled CBa in Fig. 6), with one bartail limb becoming longer than the other, that is associated with developing flow asymmetry within the distributaries (Ashworth et al., 2000). These newly formed compound bars also have little topographic relief above the water surface, and average bar dimensions are 180 m downstream and 120 m across-stream. The nature and rate of change for small compound bars are similar to that recorded for unit bars, but as compound bars evolve and become larger, their origin from the


420


A

B

C

formative discrete unit bars becomes less clear; they may no longer retain the downstream limbs in their morphology (labelled CB in Fig. 6) and their rate of migration slows. Compound bars grow when either other bars migrate and stall on their head/margins (see Fig. 7) or the channel separating a compound bar from an adjacent bar becomes abandoned and fills, resulting in creation of a single, much larger bar. Thus, the largest compound bars are the result of numerous episodes of erosion and deposition, have lengths and widths up to 800 m · 400 m, and may form in the widest areas of the river where the bankfull width is 550 m. Cross-bar channels (Bridge, 1993) are formed on these large compound bars (see Fig. 7), often during flood recession, and may extend along the entire length of the bar. However, the depth of these cross-bar channels is restricted and usually < 0Æ5 m, with channel widths of 10 m. These dimensions are thus significantly smaller than those of the primary channels that may have a depth and width of 5 and 150 m respectively.

Fig. 5. (A) Section of 200 MHz ground penetrating radar (GPR) profile from an asymmetric compound bar, Bar C (see Fig. 2 for location of the profile). Note the high-angle inclined reflections (facies 1) at the top left of the profile and enclosed within the white-lined envelope. The vertical thickness of this set is approximately equal to the flow depth adjacent to the bar margin where this profile was taken. Note that this bar can be seen in Fig. 6 labelled as Bar C on the right-hand side of the channel. (B) Photograph of the bar margin over which this GPR survey was taken, showing the surface expression of the high-angle inclined strata. Trench is approximately 5 m long. (C) Trench along the GPR profile showing the tops of the high-angle inclined strata imaged by the GPR. Trowel handle is approximately 0Æ1 m long, numbers on GPR profiles denote radar facies.

Compound bars have a similar, but more complex, morphology to the ‘sand flats’ described by Cant & Walker (1978). Aerial photographs taken in 1971 and 2000 show that over the decadal time scale, even the largest compound bars can go through a sequence of initiation, growth and destruction. Over an annual period, change is less pronounced, with the maximum erosion rates between April and October 2000 being 50 m on one barhead and maximum deposition being 20 m of lateral accretion on the margin of one compound bar. Vertical accretion was also recorded over the study period, as bar surface depressions (see below) present in April were filled by October. However, aerial photographs also show that some compound bars show no discernable change over time periods of at least 30 years. These bars are exclusively those that are dominated by a heavily vegetated surface (labelled CBv in Fig. 6), possess a relatively high elevation (1– 2 m above mean water level), and possibly have a substantial thickness (> 0Æ5 m) of fines (clay and silt). Such bars are usually found at the margins of



421

Fig. 6. Aerial photograph of a section of the study reach taken in April 2000 illustrating the key morphological features of the South Saskatchewan River. CB, compound bars; CBv, vegetated compound bars; CBl, compound bars with limbs; CBa, asymmetric compound bars; UB, unit bars. Note Bar C on the right-hand side (east) of the braid belt is that shown in Fig. 5.

the floodplain but can also be located midchannel. Study of the April 2000 aerial photographs, which were taken at low flow, reveals that there are approximately equal numbers of unit (67) and compound bars (69) in the study reach, with 57% of the compound bars possessing downstream elongated limbs and 9% being stable and vegetated.

SUB-SURFACE ALLUVIAL ARCHITECTURE

Classification and interpretation of radar facies Ground penetrating radar surveys were conducted over the full range of bar types outlined above (Fig. 2), and the main radar facies detailed below are summarized in Table 2.

Radar facies 1: high-angle inclined reflections Facies 1 consists of inclined reflections with an angle of dip from > 6 up to the angle-of-repose (Figs 5, 7–9). These reflections are most commonly truncated at their top and bottom by subhorizontal reflections, and can extend in sets for

100 m with a maximum thickness of 1Æ5 m. Facies 1 can be found throughout a compound bar, but is more prevalent in the lower sections of a vertical profile. An example of this facies (Fig. 8), from the left-hand side of a compound bar, suggests that the formative flow was thus across the bar limb and into the adjacent deep channel. The strong reflections at the top and bottom of this radar facies are interpreted as representing episodes of erosion before and after the accretion respectively. Less commonly, this facies is found with no truncation at either top or bottom, and in these cases the maximum thickness is 2 m, similar to the channel depth, and the lateral continuity is reduced to 20 m. An example of this (Fig. 5) from an active bar margin directly adjacent to a deep channel, shows inclined strata up to the angle-of-repose that were formed by bar migration. As outlined above, dunes in the South Saskatchewan reach a maximum height of 1Æ5 m and may also be expected to produce angle-of-respose inclined strata. However, as preserved cross-set thickness is only approximately one-third of the original dune height (Leclair et al., 1997), inclined cross-strata up to the angle-of-repose and which are greater than 0Æ5 m in thickness are interpreted herein as being the product of bar migration. Facies 1 thus


422


A

B

C

Fig. 7. (B) Aerial photograph taken in April 2000 of a large compound bar, Bar E (see Fig. 2 for the location of the profiles). Note the presence of numerous cross-bar channels (labelled CBC) and accretion of a unit bar and smaller compound bar onto the right-hand side of the bar (Bar Ea) and upstream bar-head (Bar Eb) respectively. Also shown are 200 MHz ground penetrating radar (GPR) profiles from Bar Ea and Bar Eb (see Fig. 2 for the location of the profiles). The white-dotted lines show the approximate location of Bars Ea and Eb, and the solid lines give the approximate location of the GPR survey lines. (C) Bar Ea is a unit bar similar to Bar B and shows a predominance of cross-strata formed by dune migration. (A) This contrasts with Bar Eb, a compound bar similar to Bar A, which has pronounced sets of high-angle inclined reflections (facies 1) within its profile. Numbers on GPR profiles denote radar facies.

forms when flow over the bartop causes bar migration into the channel thalweg, with associated slipface accretion at the bar margin (Fig. 5). The orientation of the inclined strata shown in facies 1 is linked principally with flow across the bartop, which is often laterally or obliquely across the bar, rather than with that of the principal downstream channel flow direction. Addition-

ally, if the bar margin is curved in planform, then the resulting strata will also be curved in planform (see surface expression of inclined strata, Fig. 5B). The angle of dip of the reflections can be variable over only a few metres (Fig. 9), a feature noted in other studies (see Lunt et al., 2004), and is caused by variability in the orientation of the bar relative to the GPR profile, which is most



423

Table 2. Characteristics of the four radar facies identified in the South Saskatchewan River. Radar facies

Reflection pattern

Sedimentological interpretation

Examples from other modern and ancient sandy braided rivers Blodgett & Stanley (1980), Crowley (1983), Best et al. (2003), Miall & Jones (2003), Skelly et al. (2003) Sarkar & Basumallick (1968), Conaghan & Jones (1975), Crowley (1983), Bridge (1993), Bridge et al. (1998), Bridge & Tye (2000), Best et al. (2003), Miall & Jones (2003) Allen (1983), Bristow (1987), Bridge et al. (1998)

1

High-angle inclined reflections

Large-scale inclined strata formed by migration of bar margin

2

Discontinuous undular or trough-shaped reflections

Medium- and small-scale cross-stratification formed by sinuous crested dunes

3

Low-angle reflections

4

Reflections of variable dip enclosed by a concave reflection

Strata formed by migration of low-amplitude dunes or unit bars Channel fills

Miall & Jones (2003), Skelly et al. (2003)

Examples of other studies of modern and ancient sandy braided rivers that have recorded facies with similar characteristics are listed.

pronounced for bars with a curved planform. Alternatively, changes in the dip of the reflections may be caused by a change in the slope of the bar margin, perhaps related to variability of sediment supply, as it migrates during high flow. Such an interpretation can be made for radar facies 1 shown in Fig. 9, where the angle of dip at first increases, and then decreases to the right before being replaced by discontinuous reflections (see radar facies 2), which is interpreted here as being due to of deposition by superimposed ripples/ dunes as flow stage dropped (see below).

Radar facies 2: discontinuous undular or trough-shaped reflections Individual reflections of facies 2 range from being trough-shaped, < 3 m across and 0Æ5 m high, to more horizontal or sub-horizontal and < 5 m long (Figs 5 and 9). Sets of reflections may extend laterally for tens of metres and be up to 3 m thick. The degree of concavity of the troughshaped reflectors of facies 2 is variable, and they may grade vertically and laterally into more horizontal or sub-horizontal undular reflections, with clearly defined troughs generally occurring only in the bottom part of the profiles (i.e. 2Æ5– 5 m below the bar surface). As discussed above, based on the premise that only approximately one-third of the dune form is preserved (Leclair et al., 1997), these trough-shaped reflections are interpreted as representing trough cross-stratification associated with the largest sinuous-crested dunes (i.e. 1Æ5 m high) that can only form

in the deepest channel thalwegs. Inclined crossstrata associated with smaller dunes that are below the radar resolution used herein will also be present, and will be significantly smaller than those associated with bar migration (see radar facies 1). The variability in the degree of concavity of the troughs may be attributed to differences in the orientation of the radar profile with respect to the preserved bedform. Thus, where the profile was taken perpendicular to the direction of dune migration, then the trough shape is most pronounced, but as the profile becomes more oblique to the migration direction, then the reflections become more undular. In the top part of the GPR profiles (i.e. 0–2Æ5 m below the bar surface), the absence of clear troughshaped reflections is interpreted as being due to the decrease in size of the dunes. Although the radar reflections may be horizontal or sub-horizontal, this facies represents composite sets of planar and trough cross-strata, and their bounding surfaces, which formed as a result of the migration of sinuous-crested dunes. Bartop trenches also show the presence of ripples, but the associated cross-lamination cannot be resolved by the GPR and hence only the bounding surfaces between sets are imaged.

Radar facies 3: low-angle reflections Facies 3 consists of reflections with an angle of dip < 6, and that can be arranged in sets that dip either upstream, downstream, laterally or are horizontal. Sets are usually extensive, up to a


424


A

Flow

B

April 2000

C

October 2000

Fig. 8. Ground penetrating radar profiles (200 MHz) from tail end of compound Bar A (see Fig. 2 for location of the profiles). The principal feature is the presence of high-angle inclined strata (facies 1) present in the top section of all profiles. This facies is present in both the downstream and cross-stream profiles and indicates oblique flow across the bar surface, a conclusion corroborated by the two aerial photographs (B, C) that show the bar migrated to the left as well as downstream. The black outline on the October 2000 image shows the approximate location of the bar in April 2000, and the lines give the approximate location of the ground penetrating radar (GPR) surveys. Numbers on GPR profiles denote radar facies.

maximum of 80 m in a downstream direction and 2Æ5 m in thickness (Fig. 10), with individual reflections being traceable over 10s of metres. This facies is found at all depths within the bar, but becomes more prevalent towards the surface. Observation of the exposed bar surfaces suggests that this facies forms from (1) deposition by dunes with a large wavelength:height ratio or (2) unit bars that migrate from the channel onto the bar surface, an example of which (Fig. 9D) possesses a crest with a clear low-relief slipface. The low-angle radar reflections hence represent the base of the dune or unit bar as it migrates across the bar surface, thus producing a continuous surface that can be traced both laterally and upstream/downstream. These deposits may also

contain cross-strata similar to the smaller dunes and ripples as discussed above, but again these cannot be resolved by the GPR.

Radar facies 4: reflections of variable dip enclosed by a concave reflection The key feature of facies 4 is a concave basal reflection that truncates the underlying reflections (Fig. 11), and is overlain by a series of steeply dipping inclined reflections, although low-angle and undular reflections may also be present (Fig. 11). This facies may extend downstream and laterally for 20 m, be up to 1Æ5 m in thickness and is found throughout the bar depth. Facies 4 is interpreted as having formed by the filling of channel scours and cross-bar channels.



425

A C

D

B

E

Fig. 9. (A) Section of 200 MHz ground penetrating radar (GPR) profile from unit Bar B, (see Fig. 2 for the location of the profile) with (B) interpretive sketch. Reflections below 3 m are beneath the current active channel base from the section of the river where this bar was located. The complex pattern of reflections from 3 to 5 m in the profile may relate to channel or confluence scours or the migration of unit bars. Some of the reflections probably also relate to larger dunes given the more trough-shaped nature of the reflections. Above 3 m, these trough-shaped reflections are overlain by undular reflections, probably reflecting the decreasing dune size with flow depth. These trough-shaped reflections are then replaced by low-angle reflections (facies 3), interpreted as accretion by low-amplitude dunes higher up the profile. At the very top of the profile, some high-angle inclined reflections are present (facies 1) that are < 0Æ5 m in vertical thickness. These relate to the slipface at the leading edge of the bar and are much thinner than those associated with bar margin slipface accretion shown in Figs 5 and 8. Note also the variability in the dip of these reflections, possibly related to a change in the slope of the bar margin during migration. (C)–(E) Photographs showing the surface morphological expression of facies 1, 3 and 2 respectively. Spade in (C) is approximately 1Æ2 m long, bar width in (D) and (E) is approximately 100 m. Numbers on GPR profiles denote radar facies.

The angle-of-repose reflections may form from unit bars migrating within the channels, whereas low-angle or undular reflections are interpreted to represent deposition by dunes. The distinctive concave reflection that represents the channel base is most readily seen in radar profiles taken perpendicular to the channel flow direction, with channel fills viewed in profiles taken parallel to the flow direction being more difficult to identify. In some cases, facies 4 appears to be laterally constrained, cannot be traced for any distance

between survey lines (e.g. < 20 m), occurs only in the top 50% of the profile and may lack an erosional lower surface. These characteristics are interpreted as representing a ‘bartop hollow’ (Ashworth et al., 1994), which are discrete, circular/ovoid depressions in the bartops of the South Saskatchewan River, can be up to 1Æ5 m deep and extend 10–30 m both down and across flow. These bartop hollows form largely by convergence of two inwardly accreting bar tail limbs that create a depression between them that is


426


A

Flow

B

April 2000

Cut

C

October 2000

Fill

Fig. 10. Ground penetrating radar (GPR) profiles (200 MHz) from compound Bar D (see Fig. 2 for the location of the profiles). The black outline on the October 2000 image shows the approximate location of the study bar in April 2000, and the lines denote the approximate location of the GPR survey lines. Note the very low angle and continuous nature of the reflections that represent low-amplitude dunes or unit bars accreting onto the bar surface (facies 3). Some steepening of the reflections is evident on the extreme left of profile DY1 (facies 1), and represents limited bar migration towards the channel at this point in the profile. Numbers on GPR profiles denote radar facies.

subsequently filled. Facies 4 represents a similar style of deposition to the solitary sets of crossstrata described by Lunt et al. (2004) that were ascribed to the filling of scours around ice-blocks or vegetated mudclasts.

Quantification of the occurrence of facies within different types of braid bar So as to establish the frequency of occurrence of the different facies throughout the 3Æ5 km of GPR survey lines, the following methodology was used. Individual survey lines for each bar were first interpreted based on the facies types described above. A vertical section was then

sampled every 10 m along each line, thus allowing the proportion of each facies within each vertical section to be determined for each survey line. By combining data from all the vertical sections, the overall facies proportions for each bar were produced (Table 3). Additionally, so as to quantify the vertical variability of facies within each bar, every vertical section was also divided into 0Æ5 m intervals measured upwards from a clear basal erosion surface present within the GPR surveys. Topographic surveys of the major channels indicate that this surface is at a depth similar to that of the modern river thalweg. The erosion surface in the GPR thus represents the channel base above



427

Fig. 11. Ground penetrating radar (GPR) profiles (200 MHz) from the same compound bar (Bar D) as that shown in Fig. 10 (see Fig. 2 for the location of the profiles), together with an interpretive sketch. The sequential aerial photographs in Fig. 10 show that the main change that has occurred on this bar is the filling of a bartop channel (facies 4) that formed towards the left-hand side of the bar tail area (labelled ‘cut’ and ‘fill’). The erosion surface and filling of this channel is seen preserved in both the downstream and cross-stream GPR profiles in Fig. 11. Numbers on GPR profiles denote radar facies. Table 3. Summary of the facies proportions in all bars discussed in the text.

Bar A B C D E Ea Eb Average (A–D, Ea and Eb)

No. of vertical profiles

Facies 1 (%)

Facies 2 (%)

Facies 3 (%)

Facies 4 (%)

103 9 59 26 99 23 35 255

23 6 7 1 10 9 32 13

35 62 59 47 44 61 38 50

39 32 31 46 39 27 30 34

3 0 3 6 7 3 0 3

Bars B and Ea are classified as unit bars, whilst Bars A, C, D, E and Eb are classified as compound bars. Also shown are the average percentage occurrence figures for all the combined smaller compound and unit bars, and which is compared with Bar E, the largest compound bar studied.

which bars initially develop and so identifies the likely maximum bar depth. The proportion of each facies within each 0Æ5 m vertical interval was then determined and summed, thus

enabling the vertical distribution of facies to be quantified. The total number of vertical sections sampled and facies proportions present for each bar are given in Table 3.


428


Unit bars Data from Bar B (Fig. 2; see also Fig. 9 and Table 3) show the highest proportion (62%) of radar facies 2 (discontinuous undular or troughshaped reflections) of all the GPR surveys. The lack of trough-shaped reflections higher in the profile is interpreted as representing the decrease in dune size with decreasing flow depth (i.e. the trough cross-beds of small dunes cannot be resolved by the GPR). Accretion towards the top of the profile may contain some small sets of highangle inclined reflections (facies 1, 6%). This vertical succession is also displayed in unit Bar Ea (Figs 2, 7 and 12A; Table 3), that was accreting onto a compound bar (Bar E) when the GPR survey was undertaken. Bars B and Ea show similarities in the low proportion of high-angle inclined reflections in the upper parts of the

profiles (facies 1, 9%) and the higher amounts of discontinuous undular or trough-shaped reflections (facies 2, 61%), with trough-shaped reflections only found at depth. In contrast to unit Bar B, Bar Ea possesses some reflections of variable dip that are enclosed by a concave reflection (facies 4, 3%).

Compound bars A vertical proportion plot (Fig. 12B), based on all survey lines from compound Bar A (Fig. 2), shows a consistent trend in the vertical change in facies. The lower 0Æ5 m of the bar, with its base defined by a distinct erosion surface at 2Æ5–3 m depth, is characterized by discontinuous undular or trough-shaped reflections (facies 2, 63%) and high-angle inclined reflections (facies 1, 27%). Towards the surface of the bar (top 0Æ5 m), facies 2

A

B

Fig. 12. Vertical proportion of facies derived from all ground penetrating radar profiles on (A) unit Bar Ea, and (B) compound Bar A. The principal difference between the two distributions is that the unit bar has a dominance of cross-strata formed by dune migration at depth (facies 2) with high-angle inclined strata (facies 1) only present higher up in the profile, whereas facies 1 is prevalent throughout the profile of the compound bar. For both figures, facies descriptions are as given in Table 2. 2006 The Authors. Journal compilation 2006 International Association of Sedimentologists, Sedimentology, 53, 413–434


429

Flow

Fig. 13. Facies interpretation of ground penetrating radar lines with corresponding vertical proportion curves for compound Bar A. In both figures facies descriptions are as given in Table 2.

is then progressively replaced by low-angle reflections (facies 3, 84%) as the dominant facies. The mean frequency of occurrence of facies for all GPR lines on Bar A (Table 3) is 39% (facies 3), 35% (facies 2), 23% (facies 1) and 3% (facies 4). However, these averages mask significant spatial variability within the bar, which was assessed by analysing each GPR line separately (AX1–AX6 and AY1–AY8). The downstream lines (AX; Fig. 13) display the greatest degree of variability, with the line closest to the channel thalweg, AX1, showing the greatest proportion of high-angle inclined reflections (facies 1, 39%). Away from the channel thalweg, the proportion of facies 1 drops to between 10% and 19% or is absent entirely (AX4 and AX6). Discontinuous undular or troughshaped reflections and low-angle reflections predominate on the right-hand side of the bar (looking downstream): for example, line AX6 is dominated by discontinuous undular or trough-shaped reflections (facies 2, 65%). The cross-stream lines (AY; Fig. 13) show much less variability than the downstream lines, although each line does possess some heterogeneity in facies type. For example, as noted above, high-angle inclined reflections are dominant on the left-hand side of the bar and thus facies 1 is present in the left-hand side of the cross-stream lines. Bar A migrated 150 m downstream between April and October 2000 (Fig. 8), a distance equivalent to the entire length of the bar, and most of this migration

occurred on the left-hand side of the bar as it prograded into the main thalweg (shown as the much darker area in the aerial photograph, see Fig. 8). The sets of high-angle, inclined reflections (facies 1) thus formed as sediment, which had been transported across the bar surface, avalanched down the slipface of the bar margin. Overbar flow is thus required to produce this type of deposition and these sets can be expected to be thickest where the bar margin progrades into a deep thalweg, as also shown in the Jamuna River where such bar-margin slipfaces were recorded as up to 8 m high (Best et al., 2003). As well as facies variability within a compound bar, facies variability is also apparent between bars as demonstrated by GPR data from a further three, relatively small, compound bars (Bars Eb, C and D).

Bar Eb. Two 100 m survey lines were taken over Bar Eb (Figs 2 and 7; Table 3) that was migrating within a deep channel and up onto the head of Bar E. Based on the results from Bar A, it may be expected that this migration would result in the presence of high-angle inclined reflections in the GPR profile, which is confirmed (Fig. 7) by 32% of the survey lines for Bar Eb being classified as facies 1. Bar Eb also clearly demonstrates that compound bars grow by an amalgamation of different bars, as evidenced by the stacked sets of inclined strata in the GPR profile (Fig. 7). The


430


sets of inclined strata lower in this profile (Fig. 7) are interpreted as representing other unit bars from which compound Bar Eb evolved.

Bar C. Flow was directed over the bartop and into the thalweg on the left-hand side of the bar (see Fig. 6): one bartail limb thus prograded downstream further than the other, with a steep slipface developing on one side of the resulting asymmetrical bar. Flow over this asymmetric bar, and subsequent deposition into the channel thalweg at the bar margin, again resulted in the presence of high-angle (angle-of-repose) inclined reflections (facies 1; Fig. 5) in the GPR survey, although the overall abundance (Table 3) of facies 1 within this bar was only 7%. Bar D. In contrast to Bars A, B and C that migrated downstream over the study period (April 2000– October 2000), Bar D (Figs 2 and 10; Table 3) displayed limited movement and possessed different facies characteristics despite the similarity in its planform (Fig. 10). High-angle inclined reflections were largely absent (facies 1, 1%) from the GPR profiles that were dominated by lowangle reflections (facies 3, 46%) and discontinuous undular or trough-shaped reflections (facies 2, 42%). These characteristics are in contrast to the more active bar migration and prevalence of high-angle inclined reflections seen in Fig. 8. The principal change in Bar D between April and October 2000 was infill of the channel at the downstream end of the bar (Fig. 10), as highlighted in the profiles shown in Fig. 11. As well as the relatively small compound bars detailed above, GPR profiles were also taken over a compound bar with a longer and more complicated history of erosion, deposition and reworking (Bar E; Figs 2 and 7; Table 3). Analysis of these data again reveals the major facies are highangle inclined reflections (facies 1, 10%), discontinuous undular or trough-shaped reflections (facies 2, 44%) and low-angle reflections (facies 3, 39%). The distribution of facies within Bar E also reveals a broad similarity to Bar A. Highangle inclined reflections are slightly less common (10%) when compared with the more simple compound Bar A (23%), whilst the proportions of discontinuous undular or trough-shaped reflections and low-angle reflections are similar (35% and 39%, respectively, for Bar A). Cross-strata interpreted as having formed by dune migration are found throughout the vertical profiles from Bar E and diminish in size and occurrence towards the top of the profiles. High-angle

inclined reflections are more prevalent in the bottom 65% of the profile, while low-angle reflections dominate in the upper sections of the profile. The amount of facies 4 (fill of channel scours and cross-bar channels) found within Bar E (7%) is slightly greater than that of the Bar A (3%), probably because Bar E has evolved by multiple erosional and depositional events and an amalgamation of different bar remnants, a process also documented in gravel-bed rivers by Bluck (1976).

DISCUSSION Comparison of the GPR data presented herein with the original work of Cant & Walker (1978) reveals seven important differences: 1 The channel model of Cant & Walker (1978; profile C in Fig. 1), dominated by cross-strata associated with dune migration, can apply equally to the deposits of braid bars. For example, unit bars are dominated by dune-scale cross-stratification, related to the fact that they evolve initially through a process of dune stacking. Likewise, compound bars can also display a similar vertical sequence of facies from thalweg to bartop as that given by Cant & Walker’s (1978) channel model (profile C in Fig. 1). Cross-strata produced by dune migration can thus be found throughout much of the sequence and decrease in size towards the bar surface, reflecting that dune height is related to flow depth. 2 The avalanche face at the downstream end of unit bars is of low height (0Æ5 m) and only preserves minor sets of high-angle inclined strata at the very top of the profile (Fig. 9). This is in contrast to Cant & Walker (1978) who suggested that the dominant facies of unit bars (their ‘crosschannel bars’) would be high-angle inclined strata, present throughout the full height of the bar form (Fig. 1, profile A). 3 The GPR data demonstrate that large, highangle inclined strata are most likely to form because of sedimentation at the steep margins of compound bars. The largest sets will form where a compound bar migrates into a deep channel, in which case the set may extend from bartop to channel base. Thus high-angle inclined strata should not be restricted to just 2–3 m below the surface of a compound bar, as suggested in the sand flat model of Cant & Walker (1978; see Fig. 1, profile A).


Sedimentology of the South Saskatchewan River 4 All three profiles given in the model of Cant & Walker (1978; Fig. 1) show high-angle inclined strata associated with unit bars (the cross-channel bars of Cant & Walker). However, given that it has been shown herein that unit bars do not preserve 1–2 m thick sets of high-angle inclined strata, this facies (facies 1 herein) is probably over-represented in the model of Cant & Walker (1978). 5 The present study has found no evidence for the substantial (> 0Æ1 m thick) deposition of fines that are preserved at depth (3 m below the bar surface) as shown by Cant & Walker (1978, see Fig. 1). Within a GPR profile, clay is readily recognizable by the high-amplitude reflections produced, usually with an associated loss of resolution beneath (Best et al., 2003). Although drapes of fine-grained material were apparent in bartail areas at the surface, no evidence of thicker deposits of fines was found at depth within the GPR profiles presented herein. 6 Cant & Walker (1978) state that the facies sequence depicting channel aggradation (labelled C in Fig. 1) in the South Saskatchewan River is a minimum of 5 m thick. The present study suggests that these sequences will be much more variable than this figure, as the profiles scale with the size of channel in which they formed. Thus some of the GPR profiles presented herein are only 3 m in thickness from channel base to bar surface. Accurate assessment of such variability is an important consideration when attempting to reconstruct the dimensions of ancient channel deposits (Bridge & Tye, 2000). 7 The facies distributions derived from the present GPR data demonstrate that there can be great variability both within bars and between bars, and that the three facies profiles presented by Cant & Walker (1978; Fig. 1) may not capture the intrinsic variability of sandy braided alluvium. Additionally, there still remains much uncertainty concerning the precise vertical and lateral variability in facies across the entire braid belt, given the logistical problems of collecting GPR data from channels. This area remains a key region for future research. The deposits of the South Saskatchewan River are broadly similar to those reported from other modern sandy braided rivers of different scale and braid intensities (Sambrook Smith et al., 2005). The main facies identified herein have been reported from numerous other studies of modern and ancient sandy braided rivers (Table 2). This broad similarity in the facies of sandy braided rivers (modern and ancient) is not

431

surprising, if it is assumed that generally the same physical processes occur in all sandy braided rivers, regardless of environment or scale (Skelly et al., 2003; Sambrook Smith et al., 2005). However, the present GPR study has revealed a variability in the frequency and sequence of facies that has previously been unreported. Past work, which has largely been qualitative in its approach, has tended to focus on broad similarities between different systems rather than investigating the differences that only become evident with a more quantitative approach. Although the present study possesses limitations in the areal coverage achieved, it provides the most detailed quantitative assessment of sandy braided river facies yet collected and highlights the clear need for further quantitative analysis of more braided rivers as new techniques and data become available. Comparison of the data from different bars within the South Saskatchewan River shows that bars with the same surface planform morphology may not necessarily have the same sub-surface alluvial architecture. For example, one compound bar (Bar D, Fig. 2 and Table 3) had almost a complete absence of high-angle inclined strata whereas this facies comprised approximately 30% of the deposits of other compound bars (e.g. Bar Eb, Fig. 2 and Table 3). These differences in facies proportions may be attributed to the exact nature of bar growth from the initial unit bar, which displays a less variable alluvial architecture. Compound bar growth is controlled by discharge regime, anabranch width:depth ratio, the abundance of vegetation and the local bar and channel topography, and therefore, flow direction (Bridge, 1993; Ferguson, 1993; McLelland et al., 1999). These factors will thus control the occurrence of sediment-laden flow over bar surfaces that is important for the development of highangle inclined strata (Sambrook Smith et al., 2005). As the growth of compound bars may be viewed as an amalgamation of smaller bars, the facies proportions of these large compound bars may reflect an ‘average’ of the other smaller bars discussed above. In order to test this assumption, if the facies proportions of the four smaller compound bars and two unit bars studied herein are averaged they yield: facies 1, 13%; facies 2, 50%; facies 3, 34% and facies 4, 3%. The same averaged data for the large compound bar (Bar E) yield figures of 10%, 44%, 39% and 7% for facies 1–4 respectively (Table 3). The similarity between these figures is not unexpected given that


432


compound bars grow as other bars accrete onto them. The slightly higher proportion of facies 4 in the large compound bar is also to be expected, reflecting the greater presence of cross-bar channels and the generally more complex reworking of the bar that may have occurred. This comparison also suggests that larger compound bars will provide a better characterization of the overall facies proportions within the river when compared with the smaller bars that have been shown to be more variable. However, for logistical reasons, it is generally the smaller compound and unit bars that attract study, again highlighting the need for full braidplain characterization of facies to achieve a representative facies model.

CONCLUSION This paper has used GPR to describe and quantify the deposits of unit and compound bars within the sandy braided South Saskatchewan River, enabling an assessment of the influential facies model proposed by Cant & Walker (1978). On the basis of this new evidence, six principal conclusions can be drawn: 1 Bars in the South Saskatchewan River are dominated by the following facies: (1) large-scale inclined strata formed by migration of bar margins; (2) medium- and small-scale cross-stratification formed by sinuous crested dunes and ripples; (3) low-angle strata formed by the migration of dunes or unit bars; and, (4) cross-bar channel fills. 2 Bars possess a characteristic vertical sequence of facies, with trough cross-stratification most prevalent above the basal erosive surface. Higher up in the profile, cross-strata associated with smaller dunes (and ripples revealed in shallow trenches) becomes more common with low-angle strata dominating the deposits near the bar surface. High-angle inclined strata are only found near the surface of unit bars, and are relatively small-scale, but can be found at a wider range of depths in compound bars. 3 Unit bars are dominated by cross-stratification formed by dune migration (60% of facies 2), which may merge laterally to form high-angle inclined strata. Cross-stratification associated with dunes decreases in size from thalweg to bartop, reflecting the decreasing formative flow depths. Unit bars possess only a minor amount of high-angle inclined strata (< 10% of facies 1), generally found at bar margins.

4 Flow across the top of a compound bar generates high-angle cross-stratification (facies 1) at the bar margin, whereas if a compound bar does not experience overbar flow, then low-angle planar stratification (facies 3) will predominate the resultant facies in the middle-upper parts of the profile. 5 The largest compound bars comprise an amalgamation of other smaller bars (both compound and unit), and their facies distribution is thus similar to the average for all other bars. 6 The features described herein for the South Saskatchewan River possess similarities to many other modern and ancient sandy braided rivers described in the literature. However, the GPR data demonstrate a high variability in facies both within and between bars, with this variability being greatest within compound bars. The present study suggests there is a need for more GPR studies from a range of other braided rivers to help elucidate the key relationships between formative conditions and alluvial architecture, assess if a single generic facies model can be applied to all braided rivers, or conversely, if a range of models is required. Future work should focus on establishing a direct link between the processes operating in braided rivers, the changing morphology through successive flood events and the resultant sub-surface alluvial architecture across the entire braidplain.

ACKNOWLEDGEMENTS This work was carried out with the support of NERC grant GR9/04273 to P. J. A., J. L. B. and G. H. S. S. The NERC Geophysical Equipment Pool kindly loaned a PulseEKKO 1000 GPR system to the project. We wish to thank David Ashley, for invaluable assistance in the field, Tavi Murray (University of Leeds) for supply of a PulseEkko100 GPR system, Derald Smith (University of Calgary) for the loan of a boat, outboard motors and coring equipment, Dirk De Boer (University of Saskatchewan) for help with local logistics and supply of water discharge data, and Bob and Sandy Stephenson for their vital logistical support in Outlook. J. L. B. is grateful for award of Leverhulme Trust Research Fellowship, which was partly conducted at the Ven Te Chow Hydrosystems Laboratory, University of Illinois at Urbana-Champaign, which aided completion of this paper. Ian Lunt is thanked for discussions that led to improvements in the clarity of the


Sedimentology of the South Saskatchewan River manuscript. The Graphics Unit at the University of Birmingham are thanked for producing the figures. Gary Brierley, and in particular John Bridge, provided a range of thorough, critical and perceptive review comments that greatly sharpened our interpretation of the GPR data.

REFERENCES Allen, J.R.L. (1982) Sedimentary Structures: Their Character and Physical Basis, Volume 1. Developments in Sedimentology, vol. 30. Elsevier Science Publishers, Amsterdam, 593 pp. Allen, J.R.L. (1983) Studies in fluviatile sedimentation: bars, bar-complexes and sandstone sheets (low-sinuosity braided streams) in the Brownstones (L. Devonian), Welsh Borderlands. Sed. Geol., 33, 237–293. American Society of Civil Engineers (1966) Nomenclature for bed forms in alluvial channels. Task force on bedforms in alluvial channels. J. Hydraul. Eng., 92, 51–64. Ashley, G.M. (1990) Classification of large-scale subaqueous bedforms: a new look at an old problem. J. Sed. Petrol., 60, 160–172. Ashworth, P.J., Best, J.L., Leddy, J.O. and Geehan, G.W. (1994) The physical modelling of braided rivers and deposition of fine-grained sediment. In: Process Models and Theoretical Geomorphology (Ed. M.J. Kirkby), pp. 115–139. Wiley, Chichester. 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, 47, 533–555. Best, J.L., Ashworth, P.J., Bristow, C.S. and Roden, J.E. (2003) Three-dimensional sedimentary architecture of a large, midchannel sand braid bar, Jamuna River, Bangladesh. J. Sed. Res., 73, 516–530. Blodgett, K.H. and Stanley, K.O. (1980) Stratification, bedforms and discharge relations of the braided Platte River system, Nebraska. J. Sed. Petrol., 50, 139–148. Bluck, B.J. (1976) Sedimentation in some Scottish rivers of low sinuosity. Trans. R. Soc. Edinburgh, 69, 425–456. Bridge, J.S. (1993) The interaction between channel geometry, water flow, sediment transport and deposition in braided rivers. In: Braided Rivers (Eds J.L. Best and C.S. Bristow). Geol. Soc. London Spec. Publ., 75, 13–71. Bridge, J.S. (2003) Rivers and Floodplains. Blackwell Publishing, Oxford, 491 pp. Bridge, J.S. and Tye, R.S. (2000) Interpreting the dimensions of ancient fluvial channel bars, channels, and channel belts from wireline logs and cores. AAPG Bull., 84, 1205–1228. Bridge, J.S., Smith, N.D., Trent, F., Gabel, S.L. and Bernstein, P. (1986) Sedimentology and morphology of a low-sinuosity river: Calamus River, Nebraska Sandhills. Sedimentology, 33, 851–870. Bridge, J.S., Collier, R. and Alexander, J. (1998) Large-scale structure of Calamus River deposits (Nebraska, USA) revealed using ground-penetrating radar. Sedimentology, 45, 977–986. Bristow, C.S. (1987) Brahmaputra River: channel migration and deposition. In: Recent Developments in Fluvial Sedimentology (Eds F.G. Ethridge, R.M. Flores and M.D. Harvey). Spec. Publ. Soc. Econ. Paleontol. Mineral., 39, 63–74.

433

Bristow, C.S. (1993) Sedimentary structures in bar tops in the Brahmaputra River, Bangladesh. In: Braided Rivers (Eds J.L. Best and C.S. Bristow). Geol. Soc. London Spec. Publ., 75, 277–289. Bristow, C.S., Skelly, R.L. and Ethridge, F.G. (1999) Crevasse splays from the rapidly aggrading, sand-bed, braided Niobrara River, Nebraska: effect of base-level rise. Sedimentology, 46, 1029–1049. Cant, D.J. (1976) Braided stream sedimentation in the South Saskatchewan River. Unpublished PhD thesis, McMaster University, ON, Canada, 248 pp. Cant, D.J. (1978) Bedforms and bar types in the South Saskatchewan River. J. Sed. Petrol., 48, 1321–1330. Cant, D.J. and Walker, R.G. (1976) Development of a braided fluvial facies model for the Devonian Battery Point Sandstone, Quebec. Can. J. Earth Sci., 13, 102–119. Cant, D.J. and Walker, R.G. (1978) Fluvial processes and facies sequences in the sandy braided South Saskatchewan River, Canada. Sedimentology, 25, 625–648. Collinson, J. (1970) Bedforms in the Tana River, Norway. Geogr. Ann., 52, 31–56. Conaghan, P.J. and Jones, J.G. (1975) The Hawksbury Sandstone and the Brahmaputra: a depositional model for continental sheet sandstones. J. Geol. Soc. Aust., 22, 275–283. Crowley, K.D. (1983) Large-scale bed configurations (macroforms), Platte River Basin, Colorado and Nebraska: primary structures and formative processes. Geol. Soc. Am. Bull., 94, 117–133. van Dam, R.L. and Schlager, W. (2000) Identifying causes of ground-penetrating radar reflections using time-domain reflectometry and sedimentological analyses. Sedimentology, 47, 435–450. Ethridge, F.G. and Flores, R.M. (1981) Recent and ancient nonmarine depositional environments: models for exploration. SEPM Spec. Publ., 31, 349. Ferguson, R.I. (1993) Understanding braiding processes in gravel-bed rivers: progress and unsolved problems. In: Braided Rivers (Eds J.L. Best and C.S. Bristow). Geol. Soc. London Spec. Publ., 75, 73–87. Ferguson, R.J. and Brierley, G.J. (1999) Levee morphology and sedimentology along the lower Tuross River, south-eastern Australia. Sedimentology, 46, 627–648. Fielding, C.R., Alexander, J. and McDonald, R. (1999) Sedimentary facies from ground-penetrating radar surveys of the modern, upper Burdekin River of north Queensland, Australia: consequences of extreme discharge fluctuations. In: Fluvial Sedimentology VI (Eds N.D. Smith and J. Rogers). Int. Assoc. Sedimentol. Spec. Publ., 28, 347–362. Galay, V.J., Pentland, R.S. and Halliday, R.A. (1985) Degradation of the South Saskatchewan River below Gardiner Dam. Can. J Civil Eng., 12, 849–862. Helfrick, D. (1993) Bed degradation of South Saskatchewan River below Lake Diefenbaker. Environment Canada Report, Regina, Saskatchewan, 29 pp. Leclair, S.F., Bridge, J.S. and Wang, F. (1997) Preservation of cross-strata due to migration of subaqueous dunes over aggrading and non-aggrading beds: comparison of experimental data with theory. Geosci. Can., 24, 55–66. Lehmann, F. and Green, A.G. (1999) Semi-automated georadar data acquisition in three dimensions. Geophysics, 64, 719– 731. Lunt, I.A., Bridge, J.S. and Tye, R.S. (2004) A quantitative, three-dimensional depositional model of gravely braided rivers. Sedimentology, 51, 377–414.


434


McLelland, S.J., Ashworth, P.J., Best, J.L., Roden, J.E. and Klaassen, G.J. (1999) Flow structure and transport of sandgrade suspended sediment around an evolving braid bar, Jamuna River, Bangladesh. In: Fluvial Sedimentology VI (Eds N.D. Smith and J. Rogers). Int. Assoc. Sedimentol. Spec. Publ., 28, 43–57. Miall, A.D. (1978) Lithofacies types and vertical profile models in braided river deposits: a summary. In: Fluvial Sedimentology (Ed. A.D. Miall). Can. Soc. Petrol. Geol. Mem., 5, 597–604. Miall, A.D. and Jones, B.G. (2003) Fluvial architecture of the Hawkesbury Sandstone (Triassic), near Sydney, Australia. J. Sed. Res., 73, 531–545. Neal, A. (2004) Ground-penetrating radar and its use in sedimentology: principles, problems and progress. Earth-Sci. Rev., 66, 261–330. Phillips, R.T.J. (2003) Downstream geomorphic impacts of reservoir construction on the sand-bed braided South Saskatchewan River, Saskatchewan. Unpublished B.Sc. (Hons) thesis, Simon Fraser University, BC, Canada, 75 pp. Sambrook Smith, G.H., Ashworth, P.J., Best, J.L., Woodward, J. and Simpson, C.J. (2005) The morphology and facies of sandy braided rivers: some considerations of spatial and temporal scale invariance. In: Fluvial Sedimentology VII (Eds M.D. Blum, S.B. Marriott and S.F. Leclair). Int. Assoc. Sedimentol. Spec. Publ., 35, 145–158.

Sarkar, S.K. and Basumallick, S. (1968) Morphology, structure and evolution of a channel island in the Barakar River, Barakar, West Bengal. J. Sed. Petrol., 38, 747–754. Skelly, R.L., Bristow, C.S. and Ethridge, F.G. (2003) Architecture of channel-belt deposits in an aggrading shallow sandbed braided river: the lower Niobrara River, northeast Nebraska. Sed. Geol., 158, 249–270. Smith, N.D. (1971) Transverse bars and braiding in the Lower Platte River, Nebraska. Geol. Soc. Am Bull., 82, 3407–3420. Smith, N.D. (1978) Some comments on terminology for bars in shallow rivers. In: Fluvial Sedimentology (Ed. A.D. Miall). Can. Soc. Petrol. Geol. Mem., 5, 85–88. Vandenberghe, J. and van Overmeeren, R.A. (1999) Ground penetrating radar images of selected fluvial deposits in the Netherlands. Sed. Geol., 128, 245–270. Woodward, J., Ashworth, P.J., Best, J.L., Sambrook Smith, G.H. and Simpson, C.J. (2003) The use and application of GPR in sandy fluvial environments: methodological considerations. In: Ground Penetrating Radar in Sediments (Eds C.S. Bristow and H.M. Jol). Geol. Soc. London Spec. Publ., 211, 127–142.

Manuscript received 29 April 2004; revision accepted 2 August 2005
