|
|
 |
|||||||||||||||||
|
|||||||||||||||||
 |
|||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
,1
,1
1 Department of Physical Geography, Division of Environmental and Life Sciences, Macquarie University, North Ryde, NSW, 2109, Australia
2 School of Geography and Environmental Science, University of Auckland, P.O. Box 92019, Auckland, New Zealand
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
Key Words: total stream power • longitudinal profile • floodplain formation • bedrock channels • upper Hunter catchment • Australia
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
Montgomery et al. (1996) and Sklar and Diet-rich (1998) characterized the transition from bedrock channel reaches (i.e., reaches that effectively flush alluvial materials) to reaches that store alluvial sediments in terms of catchment area–slope relationships. In the Sklar and Dietrich (1998) model, channel slopes >0.2 m/m are characterized by channel incision via debris-flow processes. Slopes of 0.001 m/m represent the transition zone between coarse-bed rivers and sand-bed rivers where channel slope is sufficiently low that the bedrock valley floor is covered with sediment and active bedrock erosion is unlikely. The diagonal line between these bounds represents the threshold between alluvial- and bedrock-dominated channels as defined by Montgomery et al. (1996). Below this transition, sediment supply exceeds sediment transport capacity, leading to deposition and the formation of alluvial sediment stores (e.g., bars) on the channel bed. Ultimately, these relationships determine the shape and evolution of the longitudinal profile, and the associated distribution of contemporary processes at differing positions along river courses (related primarily to local slope and discharge; i.e., stream power conditions).
These works highlight, among many things, the importance of the availability of erosion tools and sufficiently high stream power conditions as determinants of long-term bedrock incision rates in upstream areas, relative to downstream areas in which there are sufficient sediment stores on the valley floor to effectively shield the channel bed from erosion. Viewed in this way, the nature of geomorphic processes in sediment source, transfer, and accumulation zones, along with river character and behavior, may be directly related to the distribution of total stream power along river courses (cf. Bull, 1979; Graf, 1987; Church, 1992; Lecce, 1997; Knighton, 1999; Fonstad, 2003; Reinfelds et al., 2004; Jain et al., 2006).
Floodplain initiation is an important geomorphic process that defines landscape patterns in terms of erosion and depositional dominant zones. This work undertakes a systematic, cross-catchment appraisal of floodplain initiation processes along longitudinal profiles using a generic set of methods. Stream power is used to test the reliability of discriminating functions that differentiate where floodplains form. From this, we can explore the use of stream power as a unifying concept with which to account for the spatial distribution of depositional processes along river courses.
Floodplains and Valley Settings
A major transition along most river courses occurs where sediment becomes stored out of the channel in the form of floodplains. This transition essentially reflects the separation of channel processes from floodplain formation processes. At these locations, the residence times of sediment in storage increases as longer term sinks (i.e., floodplains) are formed. Floodplain forms can be differentiated for confined, partly confined, and laterally unconfined settings (Brierley and Fryirs, 2005) (Fig. 1). Initially, in a confined valley, coarse boulder features that are seldom reworked are found at valley margins and form small, occasional depositional pockets in the valley. These are transitional, eventually, to lower energy conditions under which finer grained sediments are deposited out of the channel. At this transition, discontinuous floodplain pockets are observed in partly confined valleys. Typically, at some distance downstream, flood-plains become continuous along both sides of the channel within laterally unconfined valleys. Viewing valley settings in this manner provides an appropriate basis to analyze the distribution and spatial transitions in floodplain forms and processes. Despite the well-established nature of these relationships, few studies have examined where floodplains begin to form along river courses, controls on this location, and relationships to the shape of longitudinal profiles.
|
Floodplains and the Stream Power
The spatial distribution of floodplains can be related to available energy in the system, which is typically expressed as total stream power (
). Total stream power is defined as the rate of liberation of kinetic energy from potential energy (Bagnold, 1966). The utilization of this energy depends upon the stream power per unit bed area, which is expressed as unit stream power (
= /w, where w = channel width). The available energy in a channel is dependent upon discharge and bed slope and expressed as:
|
| (1) |
is unit weight of water (9800 N/m3), Q is discharge (m3/s), and s is energy slope (m/m), which is considered equivalent to bed slope. Enhanced understanding of the catchment-scale distribution of total stream power has provided new opportunities to study stream power thresholds along longitudinal profiles. Various authors have highlighted the occurrence of a mid-catchment peak in total stream power along longitudinal profiles (Lecce, 1997; Knighton, 1999; Church, 2002; Fonstad, 2003). Some have suggested that this occurs along third- to fifth-order streams, others say it occurs at catchment areas between 10 and 100 km2 (Lecce, 1997; Fonstad, 2003). These relationships tend to occur in parts of landscapes where floodplains begin to form (Reinfelds et al., 2004). More recent work (Jain et al., 2006) has highlighted the occurrence of two peaks in the stream power plot along longitudinal profiles. The first peak occurs in steep headwater regions, and the second peak further down-catchment. The manifestation of these catchment-scale stream power patterns in terms of channel forms and processes requires further analysis. Macnab et al. (2006) showed that the longitudinal distribution of within-channel sediment storage is closely related to stream gradient and stream power, but local factors such as topographic variations, variability in valley width, and tributary inputs also influence the pattern.
We explain the initiation of floodplain formation using the upper Hunter catchment, New South Wales, as a case study (Fig. 2). The initiation of floodplains is analyzed using a catchment area–slope plot, the pattern and shape of longitudinal profiles, and the downstream distribution of stream power along longitudinal profiles. The relationship between total stream power and the distribution of floodplain pockets is characterized for the transition between confined, partly confined, and laterally unconfined valley settings.
|
|
STUDY AREA |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
4200 km2 (Fig. 2). The catchment area is in a tectonically passive geological setting. The Hunter-Mooki fault, which defines the boundary between the Sydney Basin and New England Fold Belt, divides the catchment into two distinct geological zones. The eastern side generally comprises Carboniferous and Devonian metasediments, and a Tertiary basalt cap at higher elevations. The western side consists of near-horizontal Permian and Triassic sedimentary rocks (primarily shales, sandstones, conglomerates, and coal), with relatively small areas of dissected Tertiary basalt (Spencer et al., 2004). The metasediments and basalts are characterized by weathering-prone high-temperature ferromagnesian minerals that can be considered less resistant rocks than the near-horizontal sedimentary rocks. Only two rivers in the upper Hunter catchment, the Pages and Hunter Rivers, cross the Hunter-Mooki fault. The upper Pages River crosses the fault about halfway between the towns of Murrurundi and Gundy. The Hunter River crosses the fault between the town of Aberdeen and the Pages River confluence. All streams contain gravel bedload material. The stream power profiles for these rivers were analyzed in Jain et al. (2006).
Landscape Units and Valley Settings
The four landscape units of the upper Hunter catchment roughly align with the geological characteristics of the catchment (Fig. 2). The Remnant Plateau landscape unit encompasses the Tertiary basalt plateau. It has a relatively high elevation (often >1000 m above sea level) and is deeply dissected. The Plateau Slopes landscape unit, which occurs at the margin of plateau remnants and high-elevation ridges, comprises steep slopes and confined valleys. The Rugged and Hilly landscape unit in the central part of the catchment is characterized by partly confined valleys. The Undulating Plains landscape unit, which occurs primarily to the west of the Hunter-Mooki fault, comprises low slopes and laterally unconfined valleys.
Most of the confined–partly confined transitions occur in the Rugged and Hilly terrain except in some western rivers, the Pages River, Dart Brook, and Middle Brook, where the transition occurs on the Plateau Slopes. The partly confined–laterally unconfined transition only occurs in the Undulating Plains landscape unit. Laterally unconfined settings are only found in the Undulating Plains landscape unit.
Hydrological Characteristics
Average annual rainfall across the upper catchment ranges from 1400 mm/yr in the Barrington Tops to 550 mm/yr west of Murrurundi. The largest floods since European settlement in the 1820s occurred in 1870 and 1955. The discharge for a 1 in 2 yr flow at Muswellbrook is 379 m3s–1, while bankfull discharge at Muswell-brook gauge is 2025 m3s–1 (roughly equivalent to a 20 yr return period flood). There is no marked difference in the discharge characteristics of eastern and western draining streams (Jain et al., 2006). The stream power values in this paper are based on the 2 yr return period flow. Catchment area (A)–discharge (Q) relationships, compiled using the Pineena database of gauging stations in the upper Hunter catchment (described in Jain et al., 2006), yielded the following relationship for the 1 in 2 yr event:
|
| (2) |
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
Absolute values of slope, catchment area, and stream power were calculated for the confined, partly confined, and laterally unconfined valley settings. Two forms of extraction were undertaken. The first involved averaging slope over the full length of the valley setting. Catchment area was measured at the downstream end of each valley setting. The second form of value extraction involved using a running average (1 km length) to generate slope and catchment area data at 25–30 m intervals within each valley setting (i.e., each DEM grid cell along the longitudinal profile). These data were used to test whether a catchment-scale trend in slope–catchment area–stream power values can be used to detect thresholds between valley settings (i.e., where floodplains begin to form and become continuous).
Analysis of Longitudinal Profile Patterns
To supplement the absolute values data set, the form of the longitudinal profile and the form of the stream power plot along these longitudinal profiles were also analyzed. Removing local variation in longitudinal profile form using a smoothing method, longitudinal profiles of streams in the upper Hunter catchment are best represented by second-order exponential curves (Jain et al., 2006; Table 1; GSA Data Repository Fig. DR11). Curve fitting improves with an increase in the order of the exponential curve. However, this effect is marginal after a particular stage. Selection of the second-order exponential curve fitting approach was based on the criterion that the difference between the normalized curve fitting error (curve fitting error/relative relief) of two successive orders was no more than 0.1 m (Jain et al., 2006). The second-order exponential curve is characterized by high decay rate and low decay rate components. These components were plotted separately over the longitudinal profile. This process divides the longitudinal profile into three parts: (1) a high decay rate curve, (2) a low decay rate curve, and (3) a transition zone. Floodplain initiation was analyzed with respect to this classification of longitudinal profile form.
|
|
| (3) |
is the unit weight of water, a, b and e are coefficients, β1 and β2 represent profile concavity (decay constant) of upstream (fast) and downstream (slow) parts of the longitudinal profile curve, L is channel length measured from channel head, H0 is the elevation of the channel initiation point, which normally represents the highest point in the catchment, and k1 and k2 are the percent of elevation covered by the upstream (fast) and downstream (slow) longitudinal profile curves, respectively. The summation of coefficients k1 and k2 will approach unity. Values of k1 and k2 were generated automatically through a spreadsheet optimization tool to minimize the sum of squared differences between the long profile and the fitted curve.
The first and second derivatives of these stream power profiles can be expressed as:
|
| (4) |
|
| (5) |
Peaks in the theoretical profile are defined as having zero value for the first derivative and negative values for the second derivative, while the trough is characterized by zero value for the first derivative and positive values for the second derivative. When the first peak is not well developed its "shoulder" effect causes an inflection point on the curve. This position is defined where the second derivative of stream power changes its sign.
Before considering the relationships between the total stream power peaks and trough on floodplain formation processes, the relationship between longitudinal profile form and total stream power requires further explanation. Using regression analysis, equation 3 has been derived to examine this relationship. In this analysis, the increasing rate of discharge with respect to channel length was kept constant through fixing the value of coefficients a and b, in the discharge function a · Lb. Among the longitudinal profile parameters, the decay rate of the downstream curve is almost the same as profile concavity (β2). Hence, the variation in total stream power is a function of variation in the β1 and k1/k2 values. This means that the headwater section of a longitudinal profile is driving the total stream power distribution pattern. The parameter β1 represents the decay rate of the headwater curve. The parameters k1 and k2 represent the proportion of the total vertical distance of the longitudinal profile composed of high and low exponential curves, respectively. As these two values are related (k1 + k2 
1), the higher value of one coefficient is associated with a lower value of the other. Hence, variation in one coefficient will affect the location of both peaks in the total stream power distribution.
The six plots shown in Figure 3 indicate the stream power distribution for different k1 values. Plots in each set represent the control exerted by longitudinal profile concavity (β1) on the stream power distribution. We observed the following. (1) The k1 and k2 values control the amplitude of each peak. At lower k1 values (k1= 0.1), the fast curve only provides the shoulder effect on the stream power profile that generates an inflection point. With increasing k1 values, the shoulder develops into a well-defined peak and the inflection point becomes a distinct trough. This characteristic of the longitudinal profile will result in the formation of a distinct upstream peak that has a high value, and a lowering of the mid-catchment peak. (2) β1 in the longitudinal profile equation controls the position and narrowness of the first peak in the stream power distribution. Higher values of β1 result in the formation of a narrow peak that is shifted upstream. This upstream shifting and distinctiveness of the first peak reduces the overlap between the two peaks and generates a distinctly bimodal distribution in the stream power plot. As a result, the trough region in the stream power distribution will shift upstream due to the higher value of β1.
|
Air Photograph and Field Analysis
Identification of the floodplain initiation zone was undertaken using 1:25,000 air photographs. These locations were verified in the field. The initiation zone is defined by the occurrence of distinct floodplain pockets that comprise inter-bedded silts and sands with occasional gravel lenses. These floodplain pockets tend to be several hundred meters long and several tens of meters wide. The sedimentology of these floodplains contrasts markedly with the coarse-grained gravel bedload found on the channel bed and in bars at these sites. Small, occasional pockets of floodplain that comprise boulder berm, colluvial, or fan deposits in more head-water, confined settings were not considered to reflect the initiation of floodplain formation processes along the longitudinal profile.
Figure 4 shows four representative examples of the transition in valley width and valley morphology between confined and partly confined valleys. Two streams on the western side of the Hunter-Mooki fault are depicted (Dart Brook and Kingdon Ponds) along with two streams on the eastern side (Hunter River and Rouchel Brook). Across the upper Hunter catchment, valleys widen by >75% at the transition between confined and partly confined valleys. These transitions may be quite abrupt at tributary confluences or at significant changes in slope. In other cases, transitions are more gradual and extend over several kilometers of river course. Abrupt transitions from a confined valley setting to a partly confined valley setting occur over 100–500 m while more gradual transitions occur over 1–3 km of river course. In general, the confined valleys are v-shaped with steep valley sides. The partly confined valleys are either flat bottomed and symmetrical with relatively gentle valley sidewalls, or are distinctly asymmetrical with one steep valley sidewall. Partly confined valleys often contain terraces against the valley margin. Terraces comprise coarse-grained, indurated gravels that differ markedly from the fine-grained silt and sand deposits that compose floodplains.
|
|
RESULTS: CONTROLS ON FLOODPLAIN INITIATION |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
|
Figure 5 shows that the broad trends identified by Sklar and Dietrich (1998) are also observed in the upper Hunter catchment. Figure 5A shows that a clear segregation occurs between the slope–catchment area characteristics of each valley setting. All the confined rivers in the upper Hunter catchment are within the bedrock-fluvial region of the Sklar and Dietrich (1998) model (Fig. 5A). In this zone, sediment transport capacity exceeds sediment supply, resulting in limited sediment storage. In the upper Hunter catchment, floodplain pockets occur above the bedrock-alluvial threshold, suggesting that floodplains tend to form on steeper slopes for a given catchment area relative to the Sklar and Dietrich (1998) threshold of instream sedimentation. It is clear that this transition zone in catchments is critical for both instream and floodplain deposition processes.
In the upper Hunter catchment, discontinuous floodplains tend to form on slopes ranging between 0.003 and 0.015 m/m. On slopes >0.015 m/m, floodplains are absent or occasional. On slopes <0.003 m/m, continuous alluvial floodplains are formed. The laterally unconfined rivers in the upper Hunter catchment are low on the Sklar and Dietrich (1998) plot, typically around the 0.003 m/m slope threshold. When the analysis is repeated using the running average data, a similar trend emerges (Fig. 5B). However, while the initial set of thresholds can be identified, there is considerable overlap in the values at the contact between confined–partly confined and partly confined–laterally unconfined valley settings. Hence, these threshold values should be considered as a continuum, with transitions in conditions rather than discrete or absolute threshold values.
The range of catchment areas under which floodplains occur is highly variable and no distinct trend emerges. However, it is interesting to note that the catchment-scale averaged values show that generally no floodplains form in catchments with an area <70 km2. These results suggest that slope, rather than catchment area, has reasonable predictive capability for identifying where discontinuous floodplain pockets occur and where they become continuous along river courses.
Longitudinal Profile Form
Longitudinal profiles of the upper Hunter catchment are characterized by variable proportions of confined, partly confined, and laterally unconfined valleys (Table DR1; see footnote 1). Two types of longitudinal profile occur. The first type of profile is found in the west of the catchment on the less resistant geologies. These profiles have a gentle, lower-elevation (1200–760 m) headwater zone. The second type of profile is found in the east of the catchment on the more resistant geologies. These profiles have a steep, higher-elevation (1500–1300 m) head-water zone.
On the eastern side of the Hunter-Mooki fault, longitudinal profiles are characterized by long confined valleys in their headwater reaches (averaging 16 km in length) (Fig. 2). Hence, the partly confined valleys that occur immediately downstream of these reaches are farther away from their headwater source. In contrast, on the western side of the Hunter-Mooki fault, the longitudinal profiles contain short confined valleys in their headwaters (averaging 3 km in length) (Fig. 2). In these cases, the partly confined valleys are closer to the headwater source. Hence, in relative terms, rivers draining from the east of the Hunter-Mooki fault have a significant proportion of their longitudinal profiles characterized by confined valleys (42%), compared to rivers on the western side of the fault, where 7% of the profile contains a confined valley.
The lengths of the partly confined valleys that occur downstream of the confined reaches, both east and west of the Hunter-Mooki fault, average 34 and 39 km, respectively. Hence a very similar proportion of the longitudinal profiles (between 52% and 43%) is partly confined valleys.
The laterally unconfined valley setting occurs primarily on the western side of the Hunter-Mooki fault. For streams that contain this valley setting in their downstream reaches, laterally unconfined rivers average 22 km in length. Only along the lower Pages River and Hunter River do short stretches of the laterally unconfined valley setting occur east of the Hunter-Mooki fault.
Long profiles in the upper Hunter catchment are marked by second-order exponential curves that indicate that upstream and downstream parts of the profiles are characterized by different decay constants (Jain et al., 2006). The decay constant of an exponential curve represents the concavity of the curve and its control on slope. A curve with a high decay constant will be characterized by steeper slope and vice versa.
The upstream parts of all longitudinal profiles in the upper Hunter catchment are characterized by relatively high decay rates (0.11–0.46) (first part of the exponential equation) compared to downstream parts that are marked by low decay rates (0.01–0.02) (second part of the exponential equation) (see Table 1). Where these decay rate curves meet the longitudinal profile defines the upstream and downstream limits of a transition zone between the headwaters and lowland sections of a longitudinal profile (Fig. 6)
. Where the high decay rate curve (representative of the headwater zone) meets the longitudinal profile defines the upstream boundary of the transition zone. The position where the longitudinal profile meets the low decay rate curve (the latter being representative of the downstream section of the profile) defines the downstream boundary of the transition zone. The meeting point of the exponential curve and the longitudinal profile is defined by the point where the difference between the two curves becomes <1% of the long profile value. Using this technique, head-water, transition, and downstream zones on each longitudinal profile are identified.
|
|
; Table DR1). This zone occurs at elevation ranges extending from 1119 m to 395 m for the upstream boundary and 700 m to 192 m for the downstream boundary. In most instances, floodplain pockets begin to form within this transition zone (exceptions are discussed in the following) (Fig. 6). Hence, the relative change in decay rate between the headwater and the lowland sections of the longitudinal profile defines where partly confined valleys begin to occur and where floodplains begin to form. This indicates that meaningful differentiation can be made between steep, confined upland sections of longitudinal profiles, characterized by a high decay rate (in which alluvial sedimentation may commence but floodplains are not evident), and lower slope reaches in downstream areas that are characterized by a lower decay rate (in which partly confined valleys with discontinuous floodplains begin to form).
When the location of the first distinct flood-plain pocket is placed on these plots, it is clear that the location of floodplain initiation does not occur at a consistent position within the transition zone along all streams (Fig. 6). Along Pages River and Dart Brook, floodplains start to form upstream of the transition zone, and along Stewarts Brook and Davis Creek, floodplains do not form in the transition zone. Longitudinal profile morphology and the variable decay rates in upstream and downstream sections can only be used to broadly define a zone within which floodplains begin to form.
To define more accurately where floodplains are formed in relation to this transition zone, the absolute values and patterns of total stream power are used to more accurately explain where floodplain formation is initiated.
Total Stream Power
Absolute Values
Confined, partly confined, and laterally unconfined rivers in the upper Hunter catchment operate within a broad range of total stream power conditions, with significant overlap among the valley settings (Fig. 7; Table DR2). In many cases, the average total stream power generated in the partly confined valley is roughly equivalent to that of the confined valley along the same river course. However, when the ranges along each stream are examined, the confined valley setting tends to generate total stream powers at the higher end of the spectrum relative to the partly confined and laterally unconfined valley settings. For example, along the Hunter River, total stream power in the confined valley ranges from 699 to 15299 W/m; in the partly confined valley the range is narrower at 1443–10459 W/m; and in laterally unconfined valley settings the range is 20–13237 W/m. Overall, however, there is no well-defined total stream power range for rivers in each valley setting.
|
Catchment-Scale Patterns of Total Stream Power
The pattern of total stream power along river courses in the upper Hunter catchment has been derived using the smoothing and theoretical methods of Jain et al. (2006). High-resolution analysis using the smoothing method reveals a highly variable pattern, reflecting significant local slope variability (Fig. 8; Jain et al., 2006). Theoretically derived plots, represented by equation 3, provide a coarse but well-defined downstream pattern of stream power that is characterized by two stream power peaks, in the headwater region and in mid-catchment (Jain et al., 2006). The peaks and troughs in the profile were defined by first- and second-order derivatives of stream power (equations 4 and 5). The specific form and location of these total stream power peaks vary markedly along longitudinal profiles in the upper Hunter catchment (Fig. 8), but are directly related to the location of flood-plain formation.
|
|
. This is noted along 8 of 13 streams studied, namely Hunter River, Pages Creek, Moonan Brook, Brush Hill Creek, Branch Creek, Rouchel Brook, Isis River, and Middle Brook. Along these streams, energy at the first peak (2500–8000 W/m) is sufficient to erode and transport sediments that are subsequently deposited when total stream power declines and transport capacity decreases in the trough. However, along three other streams, Pages River, Dart Brook, and Kingdon Ponds, floodplain deposition begins on the rising limb of the first peak. The lithology of these catchments is characterized by less resistant Tertiary olivine basalt, which has the potential to generate high sediment load. The olivine and calcic plagioclase, the major constituent of this basaltic rock, has very low resistance to weathering because of its crystallization at higher temperatures and weak bonding between silicate molecules and Fe, Mg ions (Goldich, 1938; Bloom, 1978). Hence, this olivine basalt rock is highly susceptible to weathering and potentially results in high sediment contributions. The value of the first stream power peak along these streams is low (2000 W/m) and the peak has lower amplitude. This suggests that flows with lower stream power values around the first (upstream) peak are unable to transport high volumes of sediment through the upstream reaches. In these cases, floodplains begin to form further upstream. The valleys in this part of the upper Hunter catchment are low-relief, wide, open settings in which weathering of less resistant rocks has resulted in relatively high sediment supply. The low-relief, low-energy setting results in floodplains forming before the transition zone along these rivers. The characteristics of the longitudinal profile directly affect fluvial processes, controlling the location and magnitude of peaks and troughs in the stream power distribution. Along profiles with high k1 values, a higher magnitude first peak occurs and floodplains form downstream of this peak in the trough zone. However, the location of the trough zone is controlled by β1. Higher values of β1 shift the trough position upstream and vice versa. Lower k1 values are reflected in the lower magnitude of the first peak, and in these cases, floodplains may start to form upstream of this peak. Profile concavity (β1) in the upstream part of the longitudinal profile has a direct control on the position of the trough zone in the stream power distribution pattern, thereby exerting a direct control on the location of floodplain initiation in the catchment. Higher values of β1 will result in floodplain formation in the upstream part of the catchment and vice versa.
Davis Creek and Stewarts Brook are anomalies to these patterns. Along these streams, flood-plains do not form in the transition zone. Along Stewarts Brook, floodplains form downstream of the trough, while in Davis Creek, well-defined floodplain pockets are absent. In both these cases, other controls such as limited upstream sediment supply or limited accommodation space in which floodplain pockets can form must be invoked. However, the high-resolution stream power plot shows a distinct peak in stream power at this location (Fig. 8).
Six rivers are characterized by the development of continuous floodplains in the laterally unconfined valley setting (Fig. 8). In general, the partly confined–laterally unconfined transition is either in the trough area or after a distinct local scale stream power peak as depicted in the high-resolution plots. This suggests that local-scale variability in total stream power, not detected by the theoretical model, can also be used to explain where floodplains are formed.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
In any conceptual model, reliability is either based on quantitative predictors (thresholds) or on systematic forms and patterns. Thresholds of stream power (either total or unit) have been used for some time to discriminate fluvial processes along a continuum. In their examination of floodplain forms and processes, Nanson and Croke (1992) classified high-energy (>300 W/m2), medium-energy (10–300 W/m2), and low-energy (<10 W/m2) floodplains using unit stream power based on bankfull discharge and sediment mix as discriminating parameters. In a similar context, Magilligan (1992) (building on earlier work by Miller [1990] and O'Connor et al. [1986]) identified a 300 W/m2 "threshold" for floodplain stripping processes. Unit stream power thresholds for instream sedimentation have also been defined for gravel-bed rivers in England (50 W/m2; Ferguson, 1981) and in New Zealand (30 W/m2; Carson, 1984), whereas Macnab et al. (2006) showed that instream sediment storage along river courses ceased when total stream power exceeded 15,000 W/m in an urban stream in Sydney. Flores et al. (2006), working in a bedrock-confined reach, defined unit stream power ranges for pool-riffle, plane-bed, step-pool, and cascade systems as 290–490 W/m2, 540–880 W/m2, 880–1760 W/m2, and 2150–2750 W/m2, respectively.
While valuable in providing a sense of where transitions in fluvial process may occur, it is dangerous to use these values outside the areas in which they were derived. Montgomery (2001) suggested that such analyses should be conducted on a region-by-region basis to account for variability in other controls such as the magnitude and frequency of discharge and landscape setting. The results from this study on floodplain initiation in the upper Hunter catchment demonstrate that factors such as longitudinal profile form and downstream patterns of total stream power are more powerful predictors of where transitions occur along river courses than absolute values of controlling parameters. This suggests that catchment-scale analysis of systematic trends and patterns along longitudinal profiles provides a more reliable predictor of where floodplains form than threshold analysis.
Longitudinal Profile Evolution and Valley Morphology as Controls
Longitudinal profiles evolve through two dominant processes: stream power–dependent erosion and step-wise lowering caused by knickpoint migration (Seidl et al., 1994). Second-order exponential curves used to derive longitudinal profiles may be used to analyze the former of these two processes. The analysis presented here shows that the location of peaks and troughs in total stream power, and hence the utilization of energy within the system, varies at different locations along the longitudinal profile. In general, bedrock incision rates increase with increasing stream power (Foley, 1980; Seidl et al., 1994, Stock and Montgomery, 1999; Snyder et al., 2000; Tomkin et al., 2003, Whipple, 2004); a high-amplitude first stream power peak will reflect enhanced bedrock incision. In these upstream parts of the longitudinal profile, bedrock incision rates exceed valley sidewall erosion rates (Nott et al., 1996; Seidl, et al., 1994, 1996; Young and Wray, 1999). Energy is expended by eroding bedrock in these supply-limited channels, resulting in v-shaped, narrow valleys with steep sidewalls. These valleys store little in the way of alluvial sediments, effectively flushing available material downstream. Downstream of the first stream power peak, the balance between valley incision and sidewall erosion shifts such that rates of side-wall erosion exceed incision, resulting in valley widening (Nott et al., 1996; Seidl et al., 1996). This produces a wider partly confined valley with graded sidewalls and larger accommodation space in which sediments can accumulate. At this transition, transport-limiting conditions promote instream sedimentation and floodplains begin to form (cf. Montgomery et al., 1996; Fryirs, 2002). Unless sediments are reworked, these materials protect (armor) the bedrock valley floor against incision and erosion. Over time, longitudinal profile concavity increases due to the concentration of bedrock incision processes in upstream parts of the profile and deposition in downstream reaches. Where floodplains begin to form represents the downstream boundary of bedrock-confined channels in a catchment. Thus, in the reach that spans the confined and partly confined transition, the linear relationship between sediment load and catchment area that is commonly used in bedrock incision models (Sklar and Dietrich, 1998; Whipple and Tucker, 2002) should be considered with care.
In the upper Hunter catchment, these relationships can be demonstrated using the two types of profile that occur on either side of the Hunter-Mooki fault. The profiles that occur on the more resistant geologies in the east of the catchment have a high-amplitude, narrow upstream total stream power peak, and in general a subdued mid-catchment second peak. The profiles that occur on the less resistant geologies in the west of the catchment have a low-amplitude upstream total stream power peak that grades into the second, mid-catchment peak. These profiles occur in two geological provenances, suggesting that there is an interplay between sediment supply and transport capacity along different sections of the longitudinal profiles. Along longitudinal profiles with high-amplitude upstream peaks, energy is sufficient to erode the resistant bedrock. As a result, floodplains are formed in the trough zone where transport capacity declines. Along longitudinal profiles with lower amplitude upstream peaks, less resistant geology controls sediment supply. Hence capacity limits are reached before the upstream stream power peak and sediments are deposited in floodplains. Farther downstream along these longitudinal profiles, floodplain formation is maintained as transport-limiting conditions dominate.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
More powerful predictors of where flood-plains begin to form are the downstream stream power trend and associated form of the longitudinal profile. Analysis of the longitudinal profile form shows that a transition zone can be identified where the decay rates of the profile change. Floodplains begin to form in this transition zone.
Representation of longitudinal profiles as second-order exponential curves produces a bimodal stream power pattern that can be used to identify causative factors on floodplain formation. The locations of the first and second peaks in the stream power plot, and their associated troughs, depict where floodplains form. This highlights limitations of single peak mid-catchment distribution patterns based on first-order exponential longitudinal profile derivation (Lecce, 1997; Knighton, 1999; Fonstad, 2003). While these patterns provide a primary level of understanding of where floodplains begin to form along river courses, other factors such as basin geology and valley morphology affect this relationship. Ultimately, sediment storage along valley floors reflects, and in turn exerts an influence upon, the rate of river incision and the evolution of longitudinal profiles (Seidl and Dietrich, 1992; Howard, 1994; Tucker and Whipple, 2002). The analyses performed here could be usefully extended to consider the influence of sediment supply and caliber in dictating where floodplains begin to form, relative to these peak and trough patterns.
Systematic quantifiable relationships provide powerful tools that aid prediction of river character and behavior. This study enhances our understanding of variation in landscape processes and form in terms of available driving forces along longitudinal profiles, providing a readily transferable methodological framework with which to assess the boundary condition controls under which different river processes operate. While the results may be specific to the Hunter catchment, the methods used to examine controls on floodplain formation can be applied in any landscape. The zone of floodplain initiation along longitudinal profiles defines a key transition point in river character and behavior, the specific location of which may vary in differing landscape settings. Better understanding of floodplain initiation processes may also have a range of applications relating to aquatic ecosystems, such as wetland forms and processes on floodplains and implications for biophysical fluxes such as groundwater and nutrient links (Kondolf and Wolman, 1993; Church, 2002).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Current address: Department of Earth and Environment Science, University of Texas, San Antonio, Texas 78249, USA; vikrant.jain{at}utsa.edu 
GSA Data Repository item 2007215, Tables DR1 and DR2 and Figure DR1, is available at http://www.geosociety.org/pubs/ft2007.htm or by request to editing{at}geosociety.org.
|
REFERENCES CITED |
|---|
|
TOPABSTRACTINTRODUCTIONSTUDY AREAMETHODSRESULTS: CONTROLS ON FLOODPLAIN...DISCUSSIONCONCLUSIONSREFERENCES CITED |
|---|
Bagnold, R.A., 1966, An approach to the sediment transport problem from general physics: U.S. Geological Survey Professional Paper 422-I, p. I1-I37.
Bloom, A.L., 1978, Geomorphology: A systematic analysis of late Cenozoic landforms: Englewood Cliffs, New Jersey, Prentice-Hall. 510 p.
Brardinoni, F., and Hassan, M.A., 2006, Glacial erosion, evolution of river long profiles, and the organization of process domains in mountain drainage basins of coastal British Columbia: Journal of Geophysical Research, v. 111 p. F01013 doi: 10.1029/2005JF000358.[CrossRef]
Brierley, G.J., and Fryirs, K., 2005, Geomorphology and river management: Applications of the river styles framework: Oxford, UK, Blackwell Publishing. 398 p.
Bull, W.B., 1979, Thresholds of critical power in streams: Geological Society of America Bulletin, v. 90 p. 453-464 doi: 10.1130/0016-7606(1979)90<453: TOCPIS>2.0.CO;2.[CrossRef]
Carson, M.A., 1984, The meandering-braided river threshold: A reappraisal: Journal of Hydrology, v. 73 p. 315-334 doi: 10.1016/0022-1694(84)90006-4.[CrossRef][ISI][GeoRef]
Church, M., 1992, Channel morphology and typology: in Calow, P., and Petts, G.E., eds., The rivers handbook: Oxford, UK, Blackwell Scientific Publications, p. 26-143.
Church, M., 2002, Geomorphic thresholds in riverine landscapes: Freshwater Biology, v. 47 p. 541-557 doi: 10.1046/j.1365-2427.2002.00919.x.[CrossRef][ISI]
Coates, D.R., and Vitek, J.D., eds1980, Thresholds in geomorphology: London, Allen and Unwin. 498 p.
Ferguson, R.I., 1981, Channel forms and channel changes: in Lewin, J., ed., British rivers: London, Allen and Unwin, p. 90-125.
Flores, A.N., Bledsoe, B.P., Cuhaciyan, C.O., and Wohl, E.E., 2006, Channel-reach morphology dependence on energy, scale, and hydroclimatic processes with implications for prediction using geospatial data: Water Resources Research, v. 42. W06412 doi: 10.1029/2005 WR004226.[CrossRef]
Foley, M.G., 1980, Bedrock incision by streams—Summary: Geological Society of America Bulletin, v. 91 Part I, p. 577-578 doi: 10.1130/0016-7606(1980)91<577: BIBSS>2.0.CO;2.[CrossRef]
Fonstad, M.A., 2003, Spatial variation in the power of mountain streams in the Sangre de Cristo Mountains, New Mexico: Geomorphology, v. 55 p. 75-96 doi: 10.1016/S0169-555X(03)00133-8.[CrossRef][ISI][GeoRef]
Fryirs, K., 2002, Antecedent controls on river character, behaviour and evolution at the base of the escarpment in Bega catchment, South Coast, NSW, Australia: Zeitschrift für Geomorphologie, v. 46 p. 475-504.[ISI][GeoRef]
Goldich, S.S., 1938, A study in rock weathering: Journal of Geology, v. 46 p. 17-58.[GeoRef]
Graf, W.L., 1987, Late Holocene sediment storage in canyons of the Colorado Plateau: Geological Society of America Bulletin, v. 99 p. 261-271 doi: 10.1130/0016-7606(1987)99<261:LHSSIC>2.0.CO;2.[Abstract][CrossRef][ISI][GeoRef]
Howard, A.D., 1994, A detachment-limited model of drainage basin evolution: Water Resources Research, v. 30 p. 2261-2285 doi: 10.1029/94WR00757.[CrossRef][ISI][GeoRef]
Ijjasz-Vasquez, E.J., and Bras, R.L., 1995, Scaling regimes of local slope versus contributing area in digital elevation models: Geomorphology, v. 12 p. 299-311 doi: 10.1016/0169-555X(95)00012-T.[CrossRef][ISI][GeoRef]
Jain, V., Preston, N., Fryirs, K., and Brierley, G.J., 2006, Comparative assessment of three approaches for deriving stream power plots along long profiles in the upper Hunter River catchment, New South Wales, Australia: Geomorphology, v. 74 p. 297-317 doi: 10.1016/j.geomorph.2005.08.012.[CrossRef][ISI][GeoRef]
Knighton, D.A., 1999, Downstream variation in stream power: Geomorphology, v.
