Structure and hydraulics of natural woody debris jams

Authors


Abstract

[1] Numerous studies exist on the hydraulics of woody debris jams and the mechanisms driving their geomorphic influence. While most hydraulic studies treat jams as single, solid objects, jams are clearly not individual cylindrical logs but rather an accumulation of pieces ranging in size from leaves and twigs to entire trunks. Here we treat debris jams as complex and porous accumulations of heterogeneous material to understand the relative importance of the different size fractions comprising a jam. We systematically dismantled three deflector debris jams in four stages, removing a total of 17,783 individual wood pieces, to experimentally manipulate jam porosity. We measured the surrounding velocity, shear stress, and drag force (FD). The assumption of nonporosity can result in a 10−20% overestimation of FD. Back-calculated values of the combined drag coefficient and frontal area term (CDAF)calc represented the drag characteristics of natural debris jams, whereas separating frontal area (AF(emp)) and drag coefficient (CD) contributions in natural jams is misleading. Values of (CDAF)calc for each jam at each stage of removal captured the effects of size and composition of the jam. Wood piece size in debris jams dictates the surface area to volume relationship. This association in turn determines the rate at which FD and (CDAF)calc change with the addition of material. Only low-porosity jams produce the geomorphic and hydraulic characteristics commonly associated with deflector jams. Our results on natural debris jams also illustrate the importance of employing variable wood size fractions when using woody debris jams for river restoration.

1. Introduction

[2] Woody debris jams substantially influence river geomorphology and ecology, and rivers with debris jams are often distinct from those without [Bilby and Likens, 1980; Keller and Swanson, 1979; Montgomery et al., 1996]. The natural formation of wood into debris jams, defined as the buildup of woody material of variable sizes and quantities into a distinctive unit, has been imitated by the river restoration industry [Abbe et al., 2003; Bernhardt et al., 2005]. Debris jams are often used in restoration channel design to alter channel hydraulics and morphology for specific goals such as bank protection or habitat formation. Successful restoration projects require an understanding of the relationship between the structure of debris jams, resultant hydraulic processes, and eventual geomorphic forms. This relationship is based on the hydraulics of jams beginning with an alteration of local flow.

[3] Shifting flow patterns due to a single log, multiple logs or an entire debris jam alter the spatial distribution of shear stress [Manga and Kirchner, 2000; Daniels and Rhoads, 2004b]. At the reach scale, woody debris repartitions boundary shear stress, resulting in overall finer bed material [Buffington and Montgomery, 1999a]. Woody debris also decreases the spacing between pools [Gurnell and Sweet, 1998], increases pool area [Lisle, 1995], and increases channel width [Smith et al., 1993]. At the patch scale, the addition of woody debris alters the spatial distribution of shear stress, creating patches of scour and deposition and sediment sorting [Smith et al., 1993]. Scour around woody debris is caused by flow convergence where diverted flow intersects the main unaffected flow, forming coarse-bedded pools at the outer tip of the wood piece or jam [Cherry and Beschta, 1989]. Flow separation upstream of woody debris causes backwater effects, and blockage of flow causes reduced velocities downstream, resulting in reduced shear stress on the bed and deposition of fines [Wallerstein et al., 1997].

[4] The specific effects of a debris jam first depend on the magnitude of flow. Despite our current understanding that little geomorphic change occurs during base flows [Wolman and Miller, 1960], past research is limited to data collected at such low flows. The closest study to flood discharges was Daniels and Rhoads [2004a], who studied the three-dimensional flow structures around large woody debris for two flow stages. However, the higher discharge was below bankfull, and so we still lack empirical documentation of how debris jams affect flow at bankfull or flood stages.

[5] Second, debris jams alter channel morphology and hydraulics on the basis of the distinctive geometry of the wood piece or pieces [Lisle, 1986]. For many years, the majority of studies of the localized hydraulics of wood in rivers have been limited to those of single solid objects (for a review, see Gippel [1995]). Recently, our understanding of woody debris' local hydraulics have expanded to include the three-dimensional flow fields around entire jams within a meander bend [Daniels and Rhoads, 2004a] and the flow fields within and around engineered log structures [Shields et al., 2001]. Yet, the single log model still dominates the literature, providing the predictive relationships between woody debris geometry and flow. This single log model, defined as the representation of woody debris by cylinders, has been used in multiple flume-based studies to derive the predictive relationship between flow and the geometry and orientation of a single wood piece [Young, 1991; Gippel et al., 1992; Wilcox et al., 2006].

[6] However, woody debris often accumulates into jams, which can be combinations of wood pieces from the nearby bank and those transported fluvially from upstream sources [Abbe and Montgomery, 2003]. Beginning with a single key member, these “combination jams” evolve into intricate matrices of wood pieces of widely variable sizes. Therefore, to define the relationship between jam structure and composition and hydraulic changes associated with debris jams, it is necessary to treat them as complex and dynamic accumulations of material ranging in size from leaves and twigs to entire tree trunks. Here, jam composition is defined by the number and size of wood pieces, the volume and surface area of these pieces, and the open space between wood pieces, also referred to as the jam porosity.

1.1. Hydraulics of Debris Jams

[7] The relationship between jam structure and hydraulic function is quantified through drag force (FD), which is the difference in pressure the water exerts on the jam from upstream to downstream [Abbe and Montgomery, 1996]. Empirically, FD is given by

equation image

where U is the density of water, AF(emp) is the empirical (i.e., measurable) submerged frontal area of the obstruction normal to flow, UAp is the approach velocity measured as the mean free-stream velocity, and CD is the drag coefficient of the obstruction. When woody debris is modeled as a cylinder, AF(emp) is simply the diameter of the object multiplied by its length. The approach velocity is independent of the object, and may be manipulated in a controlled setting or measured in a natural one. Therefore the contribution of CD to FD can be directly quantified [Young, 1991].

[8] Natural debris jams are poorly described as cylinders; instead they are irregular, porous, and three dimensional. Therefore AF(emp) loses its meaning in that the frontal area of a jam is a poor representation of the geometry of the jam relative to the flow. Similarly, CD previously has been solved for a two-dimensional nonporous object (e.g., cylindrical rods) [Gippel et al., 1992], which likely misrepresents the drag properties of a debris jam. Isolating the influences of AF(emp) and CD cannot easily be done because they are interrelated, thus separating the terms may misrepresent their contributions to the drag force.

[9] Instead, we use the combined term (CDAF)calc in order to bridge the hydraulic and the structural realities of debris jams. This term describes the drag form of a jam, defined as the shape and size of a jam as it dictates its drag force. This combined term has the potential to account for the entire depth of accumulated material including total roughness within the jam (i.e., surface area of woody pieces) and the open space, or porosity.

1.2. Goals and Structure

[10] The goal of this study is to define the relationship between jam composition and the hydraulics of debris jams, document the hydraulic drag on natural debris jams at a high channel-forming flow, and illustrate the utility of the combined term (CDAF)calc in analyzing the hydraulics of natural debris jams. We work at the scale of a single jam in order to expand upon the single-log model used in previous studies. By using naturally formed and manipulated debris jams, we investigated the local hydraulics associated with these complex structures. We systematically altered debris jam porosity via targeted removal of specified size classes of wood, isolating the effects of differing structure and composition.

[11] In this paper, we first examine the composition of three natural jams in a high-gradient mountain river in the Northeast United States, as such detailed quantitative descriptions of jams are rare and broad qualitative measurements of jams was thought to be misleading for our intended hydraulic analysis. Second, we report local velocity and shear stress distributions, focusing on their adjustments as jam structure and composition change. Third, we predict the potential shifts in areas of scour and deposition. Fourth, we quantify at various stages of jam manipulation the drag force (FD), the drag coefficient (CD) and the combined term (CDAF)calc. Fifth, we explore the influence of total volume, surface area and porosity on FD and (CDAF)calc, thus providing a quantitative and predictive link between natural debris jam characteristics and drag force. Finally, we compare our results to those reported for single-log models, and the potential error in using such models when analyzing natural debris jams.

2. Study Site

[12] The Indian River is located within the Hudson River watershed in the Adirondack Mountains in New York State (Figure 1). The geology is dominated by metamorphic rock recently (∼10,000 yrs B.P.) scoured and modified by the Laurentian glacier during the Wisconsin glaciation. Deposition of glacial till and glacial erratics define the current landscape and constrain current geomorphic adjustments. Regional hydrology is characterized by snowmelt peak flows in April or May, low flow in July and August and large frontal systems associated with tropical depressions in the autumn.

Figure 1.

Study sites along the Indian River, Adirondack Mountains, New York. Cross sections represent the relative elevations (in meters) of the bed close to the jam where the y axis is the elevation above the channel thalweg and the x axis is the distance from the left, top bank.

[13] The Indian River is a laterally confined, high-gradient (1% slope), cobble and boulder bed tributary of the Hudson River. Abanakee Dam, located 4.5 km upstream of the confluence with the Hudson, alters the hydrology of the Indian and Hudson Rivers. Water is released from Abanakee Dam 4 days per week from April to October, primarily for a burgeoning whitewater rafting industry. Releases increase flow by nearly an order of magnitude (from 5 m3s−1 to 40 m3s−1), transitioning from very low base flow to near and above bankfull in several minutes. These releases reproduce consistent bankfull events that remain constant at around 40 m3s−1 for approximately 90 min during each release.

[14] Steep hillslopes and bedrock controlled valley walls minimize the extent of floodplain surfaces. Extensive logging in the late 19th century and two major fires in the early part of the 20th century converted the majority of the forests in the Adirondacks from spruce fir to Northern hardwoods. The hillslopes along the Indian River are primarily populated by yellow birch, sugar maple and ash. Safety concerns from the rafting industry result in the clearing of the main channel of most large trees and trunks.

[15] We studied three naturally occurring bank deflector jams (in the stricter sense, Abbe and Montgomery [2003]) that comprised in situ key members whose surface serves to collect large quantities of fluvially transported wood pieces (Figure 1). The key members of all three jams were anchored on the bank and stabilized naturally by large boulders. Jam 1 was located 1.4 km downstream of the dam in a steep (slope = 1.5%) and narrow (width = 40 m) reach with coarse bed material (D50 = 150−280 mm). Jams 2 and 3 were located 2.3 and 1.8 km downstream of the dam in a wide (width = 60 m), lower gradient reach (slope = 0.5%) with finer bed material (D50 = 100−150 mm). The jams of the Indian River are potentially affected by three anthropogenic modifications. First, according to the local rafting guides, large logs in the Indian River are frequently removed or cut. However, the initial key members of the studied jams had no sign of human intervention, and all accumulated material appeared to be the result of natural processes. Second, the study reaches may receive less fluvially transported wood than other undammed sections due to the cutoff of supply upstream by the dams. Third, wood in the river is frequently remobilized during releases.

3. Methods

3.1. Experimental Design

[16] We sought to study natural woody debris jams at high and low flows at various stages of removal and treatment. For each jam, the key member was defined as that piece or pieces which initiated the formation of the debris jam [Abbe and Montgomery, 1996]. We established a classification for all accumulated woody debris with small woody debris (SWD, diameter < 1 cm), medium woody debris (MWD, 1 cm < diameter <10 cm), and large woody debris (LWD diameter >10 cm). We used this classification to organize stages of removal and data analysis.

[17] Four stages were defined by size of material on the jam, representing differing degrees of porosity, volume of material, and frontal area (Figure 2). For stage A (the wrapped jam), we covered the entire surface of the natural jam with a plastic tarp, taking care to not change the overall dimensions of the jam. This simulated the nonporous condition assumed in previous modeling studies. Stage B (the natural jam) was porous with all accumulated material including all sizes of woody debris as well as soil and leaf litter. For stage C (partial jam), all SWD, soil, leaf litter and other pieces deemed unstable were removed. This resulted in a framework of the key member, MWD and LWD, and represents a highly porous (possibly recently formed) debris jam. For stage D (key member), the remaining woody debris was removed, leaving only the key member(s). While systematically removing the wood, soil and leaf litter between stages B, C, and D, we measured the length and diameter of each wood piece and accounted for the volume of soil and leaf litter using a graduated bucket. We calculated volume (V) and surface area (Asurf) for each piece of wood removed by treating each individual piece as a cylinder, assigning it a single length and diameter. Key member dimensions were measured in greater detail to account for tapering and branching. We related the total surface area, Asurf, to the total volume, V, of wood pieces within a jam for each stage of removal using the power function

equation image

Volume was used as the predictor variable because it is generally reported as an indicator of wood loading in a reach [Erskine and Webb, 2003; Gregory et al., 1993; Fox et al., 2003].

Figure 2.

Stage of removal defined for this study. The stages represent differing degrees of porosity and volume of material. Limitations due to the size and composition of jams 1 and 3 allowed for only three of the four stages.

[18] As a central goal of this study was to relate jam composition and structure to its hydraulic properties, we investigated the predictive capability of the variables measured for each jam at each stage. A multivariate linear regression model without an intercept [Littell et al., 1996] was run for the independent variables that we used to define jam composition (AF(emp), Asurf and V) in order to predict the drag force (FD) and drag form (CDAF)calc for each jam and at each stage of removal. While porosity was an important variable, it could not be empirically quantified. We used a fixed-effects model [Littell et al., 1996] to account for the within- and between-jam effects.

[19] We spatially separated the hydraulic effects of the LWD jams into three patches: upstream, adjacent, and downstream (Figures 3 and 4) . We defined each patch by the high-discharge flow fields and local morphologic features at each jam. In its natural state, part of the upstream patch was backwatered and had a strong lateral velocity component due to flow deflection and slightly finer bed material than unaffected areas. The upstream patch also included nonaffected flow and bed material. The adjacent patch encompassed the flow convergence zone and was characterized by increased lateral velocity components, elevated velocities, scour holes, and coarser bed material. The downstream patch, in the hydraulic shadow of the jams, had greatly reduced velocities, shallow water depths, and finer bed material.

Figure 3.

Schematic of a control volume based on a jam's influence on the local flow in order to solve the momentum equation. The upstream boundary (surface 1) is the total inflow and is used to account for the momentum lost to the bed and jam in the longitudinal direction (x) and the lateral direction (y). The left-hand side represents the bank and acts as a no-flow boundary. We divided the spatial effects of a jam based on the flow patterns, bed material, and morphology. The jams studied generally had some deposition upstream due to reduced velocities as well as downstream. Scour occurred in the adjacent area because of flow convergence and downstream because of flow acceleration under the jam.

Figure 4.

Example of locations of sampling platforms and velocity measurements. Each location represents a four-point velocity profile sampled during low- and high-flow conditions. Sampling locations remained the same at each jam from one stage to the next.

[20] To characterize each jam site, we surveyed cross sections and the local slope and measured the key member angle relative to the main flow (Figure 1). We quantified grain size distribution in each of the four patches via pebble counts [after Wolman, 1954]. The dimensions of each jam were measured for each stage of removal, including the surface normal to flow (total length and height) and its volumetric dimensions (depth of material behind surface normal to flow). Since debris jams are complex structures, these measurements were taken in discrete sections of similar dimensions. Depth of flow with respect to debris jam height was noted for each stage. All jams at all stages of removal were either overtopped or nearly overtopped during the high-flow releases. Therefore AF(emp), V and Asurf represent the entire jam for the high-flow cases. We calculated the blockage ratio for each jam as the area of the surface of the jam normal to flow over the cross-sectional area of the channel [Gippel et al., 1996]. Since drag has also been found to be a function of the Froude number of the approaching flow [Wallerstein et al., 2002], we calculated this value for each jam at each stage of removal as well.

[21] Two 1.2 by 2.4 m free-standing platforms were installed at each site making data collection at high flows possible without obstructing flow. We placed one platform upstream of the jam for access to the upstream patch. The other platform was placed at the adjacent patch. The downstream patch was wadeable (barely) during high flows. For each site, we established 22−25 fixed locations at which velocity profiles (4 vertical points per spatial location) were measured during high and low flow at each stage of removal (Figure 4). We sampled the same locations each time by using painted rocks on the streambed for reference points. Longitudinal (Ux) and lateral (Uy) velocity components were measured by three operators at each location divided into the three patches (upstream, adjacent and downstream) using 2 Sontek FlowTracker 2D acoustic Doppler velocimeters (ADV) (adjacent and downstream patches) or a Marsh McBirney Flow Mate electromagnetic current meter (ECM) (upstream patch). Since the ECM was one dimensional, the operator held the probe in the x and y directions. Differences in measuring devices were tested prior to the experiment and determined to be acceptable for the range of flows and velocities studied. Further, we used the ECM at the upstream sampling platform because flow was primarily longitudinal and thus error associated with using this approach would be minimized.

[22] Velocity profiles were obtained for each location by measuring the velocity at 80, 60, and 40 percent depth. Additionally, we measured near-bed velocities by lowering the probe down as far as possible without interference from the bed (signal-to-noise ratio (SNR) greater than 10 and a boundary adjustment value of good or better). Each point in the profile was averaged over a 60 s sampling period. Only 54% of the velocity profiles fit a logarithmic curve (p < 0.01). For consistency, we determined depth-averaged velocities for each profile by integrating the velocity point measurements over the entire depth [Byrd et al., 2000] rather than assuming a log profile [e.g., Wilcock, 1996].

[23] The resultant depth-averaged velocities were used to calculate near-bed local shear stress (τbed) using the law of the wall [Wilcock, 1996]. While the depth-averaged method of solving for shear stress assumes a logarithmic profile, Wilcock [1996] found that the alternate method of using a single near-bed measurement is less precise. We did not use the single near-bed measurement because of large bed material and associated inaccuracies with near-bed measurements. Thus, to calculate τbed, we first solved for the shear velocity (u*)

equation image

where U is the depth-averaged velocity from the integrated, measured profile, κ is von Karman's constant (0.40), h is the total water depth and z0 is the bed roughness length (estimated as z0 = 0.1D90). The local shear stress is then equal to

equation image

[24] Depth-averaged velocities and shear stress values were grouped by patch (upstream, adjacent and downstream) and averaged for each stage of removal at both high and low flow in order to quantitatively compare the spatial trends. For each stage of removal, the flow fields and shear stress distributions were interpolated by kriging using the software package Surfer (version 8.0).

[25] Using the grain size distributions for each patch at each jam, we applied the Shields equation to calculate the critical shear stress (τc50, assuming τc* = 0.06) needed to mobilize the D50 [Buffington and Montgomery, 1999a]. Excess shear stress (i.e., τbedc50 > 1) was mapped to evaluate the localized areas of high shear stress and the potential for scour. Values greater than 1 indicated potential particle entrainment.

3.2. Determining Drag Force Through Momentum Extraction

[26] We quantified the drag force on each jam at high flow for each stage of removal using the momentum principle: the sum of the external forces on a system is equal to the change of momentum through that system [Roberson and Crowe, 1997]. Applying this principle to a control volume (Figure 4), the difference in momentum between the inflow and outflow surfaces is equal to the force exerted on the surface of the control volume. In the case of a debris jam, the external forces changing the fluid momentum are forces exerted by the debris jam and the shear stress exerted by the bed and banks. An additional force, the resistance generated by waves and turbulence at locations of sharp velocity reductions [Leopold et al., 1964], known as the spill resistance, was present to varying degrees during the high flows. Wilcox et al. [2006] studied LWD in step pool channels via flume experiments and tested the reliability of partitioning the resistance contributed individually by LWD, spill effects, and bed roughness. They showed that resistance partitioning tends to inflate the values assigned to the unmeasurable or leftover components. Thus, when direct measurements of components of resistance are not available, adopting a partitioning or additive approach can introduce an unknown amount of error into calculations. We chose to not include an estimated spill resistance value with unknown error since the remaining resistance values could be directly measured.

[27] A control volume (Figure 4) was defined on the basis of the spatial extent of a jam's influence on the local velocities. We isolated the effect of the longitudinal and lateral forces on the jam by solving for their components separately. Assuming steady flow, the momentum equation simplifies to

equation image

where F is the external forces, U is depth-averaged velocity, and A is the area vector that has the magnitude of the area and is directed normal to the control surface (inflow or outflow) in question, integrated over the control surface.

[28] The longitudinal and lateral forces must be solved for separately. Solving for the longitudinal (x) direction only, the sum of the external forces is given by

equation image

where F1 is the hydrostatic force across the inflow surface (surface 1), F2 is the hydrostatic force across the outflow surface (surface 2), and Fcv(x) is the force exerted on the control volume in the x direction, including forces acting along the boundaries of the jam, bed, and banks. The hydrostatic force is equal to the pressure exerted by the water integrated over the area of flow:

equation image

where p is the local magnitude of the pressure acting on dA. The remaining force exerted on the control volume (Fcv(x)), is a function of the force exerted on the debris jam (the drag force, FD(x)) and the resistance of the channel boundary (Fboundary(x)).

equation image

[29] In order to separate out the drag force on the debris jam, we applied the concept of shear stress partitioning [Buffington and Montgomery, 1999a], accounting for all the roughness elements in the control volume contributing to the total boundary shear stress. Since the extent of our control volume was very small in proportion to the entire channel (about 20% of the width), we neglected large-scale roughness factors such as sinuosity. Skin friction (grain resistance from the bed and banks) and bed form drag are relevant, and we quantified these by computing τbed using measured velocity profiles. Integrating τbed over the surface of the control volume in contact with the bed gives Fboundary(x):

equation image

where Abed is the area of the bed within the control volume. This approach assumes that the roughness exerted by the banks is negligible. Such an assumption is generally valid for relatively wide channels and channels with limited bank undulations or limited obstructions protruding from the banks [Buffington and Montgomery, 1999b].

[30] Returning to the momentum equation (equation (5)) we sum the forces:

equation image

noting that the first term on the right side is negative because of the definition of the control surface relative to the direction of flow.

[31] Finally, substituting equation (8) into equation (10) results in the following equation for FD(x):

equation image

[32] In order to solve for the lateral force on the jam (y direction), the definition of the terms changes slightly. We assumed that with no jam the flow would be completely in the x direction (i.e., UyUx) and y-directional flow across surface 3 (Figure 4) was entirely due to the force exerted by the jam. Assuming no pressure gradient existed between the bank and surface 3, then the momentum equation in the y direction is given by

equation image

where

equation image

Resulting in the following equation for (FD(y)):

equation image

[33] The drag forces computed using equations (11) and (14) may be combined to provide the total resultant drag force on the jam:

equation image

3.3. CD and (CDAF)calc

[34] Finally, we sought to use the field quantification of FD, AF(emp) and UAp to back calculate CD using equation (1). With equation (15), we calculated FD for high-flow conditions. Using measurements and surveys, we calculated AF(emp) for each debris jam at each stage of removal. The approach velocity, UAp, was measured in the field during each stage of removal and was defined as the far-field velocity not affected by the jam or the changes we made to the jam (values of UAp did not differ greatly at each site because of the consistent nature of the flow releases). Using the empirical values of FD, AF(emp), and UAp, we calculated CD for each jam at each stage of removal. We calculated CD to compare our values of porous, natural debris jams to published values from modeled, simplified, woody debris.

[35] Because CD and AF(emp) in previous cases assumes a solid, nonporous object which is not reflective of natural debris jams, we also used empirical values of only FD and UAp to back calculate (CDAF)calc (i.e., not isolating the individual contribution of CD or AF(emp) to FD). Therefore the calculation of (CDAF)calc does not contain any predetermined value of AF(emp), and the influence of the surface of the debris jam is intertwined with CD. We hypothesized that this term is more useful for implementing equation (1) in natural debris jam settings in which neither CD nor AF(emp) have clear physical meanings on their own.

4. Results

4.1. Jam Structure and Composition

4.1.1. Key Member(s) and Total Accumulation

[36] While the three jams studied ranged in their key member sizes, overall dimensions and total volume of material (Table 1), the ratio of key member volume to accumulated material volume was similar. The key member or members contributed, on average, 45 ± 4% of the total volume of the jams (Table 1). The remaining 55± 4% was due to soil, leaf litter, twigs, sticks and other fluvial transported pieces of woody debris.

Table 1. Jam Composition Divided by Each Stage of Removala
 Jam 1Jam 2Jam 3
NaturalPartialKey MemberWrappedNaturalPartialKey MemberWrappedNaturalKey Member
  • a

    Numbers in parentheses represent the percentage of the total volume, surface area, or number of pieces for that jam at that stage. SWD means pieces less than or equal to 0.1 cm diameter, MWD means pieces with diameter between 0.1 and 10 cm, and LWD means pieces with diameter equal to or greater than 10 cm. NA means not applicable.

Volume, m3
Leaf litter0.72 (6)00NA0.18 (6)00NA0.05 (9)0
Soil0.08 (1)00NA0.19 (6)0.04 (2)0NA00
SWD0.73 (6)00NA0.15 (5)0.05 (2)0NA0.02 (3)0
MWD2.58 (19)1.81 (19)0NA0.46 (16)0.38 (16)0NA0.09 (16)0
LWD3.43 (27)2.64 (27)0NA0.55 (18)0.54 (21)0NA0.16 (27)0
Key member(s)5.16 (41)5.16 (54)5.16 (100)NA1.44 (49)1.44 (59)1.44 (100)NA0.26 (45)0.26 (100)
Total volume12.709.615.1611.622.972.451.442.120.580.26
 
Surface Area, m2
SWD292.9 (51)00NA67.0 (56)28.5 (38)0NA7.9 (35)0
MWD166.2 (29)103.2 (50)0NA32.2 (27)25.7 (34)0NA5.5 (24)0
LWD70.3 (12)59.4 (29)0NA14.7 (12)14.4 (20)0NA3.5 (16)0
Key member(s)44.0 (8)44.0 (21)44.0 (100)NA6.2 (5)6.2 (8)6.2 (100)NA5.6 (25)5.6 (100)
Total surface area573.4207.444.038.6120.274.86.24.822.55.6
 
Number of Pieces
SWD13681 (95)32 (8)0NA2560 (93)755 (84)0NA527 (91)0
MWD687 (5)303 (77)0NA190 (6)129 (14)0NA43 (7)0
LWD69 (0)54 (14)0NA15 (1)14 (2)0NA6 (1)0
Key member(s)2 (0)2 (1)2 (100)NA2 (0)2 (0)2 (100)NA1 (1)1 (100)
Total number144393912NA27679002NA5771

[37] Jam 1 had two key members with lengths of 14.0 m and 18.4 m and diameters of 50 cm and 30 cm, respectively. Jam 2 had one key member derived from the adjacent bank and another deposited from upstream. The locally derived key member had a length extending from the bank out into flow of 7.4 m and a diameter of 25 cm, and the secondary key member was 4.0 m long with a diameter of 48 cm. Jam 3 had one key member with a length of 7.4 m and diameter of 17 cm.

4.1.2. Volume and Surface Area of Accumulated Pieces

[38] The range in total jam volume resulted in a large variation in the number of pieces on each jam, changing by an order of magnitude between jams (Table 1), from 102 on jam 3, to 103 for jam 2, and 104 on jam 1. Relative proportions of woody debris piece size classes (e.g., SWD, MWD, LWD) were similar across all three jams in terms of total composition of the debris jams, total number of pieces, total surface area and total volume.

[39] Small woody debris (SWD) accounted for the majority of the wood pieces found on all three jams (91−95% of all pieces) contributing a greater proportion of the total surface area (35−56%) than total volume (3−6%). Similarly, MWD contributed comparable proportions of both volume (16−19%) and surface area (24−29%). While MWD still accounts for more of the total surface area than total volume, this relationship is much less than that seen for SWD. LWD added more volume than surface area. All three jams had a relatively small number of large pieces (69, 15 and 6 respectively), yet these pieces added a large proportion of volume (27%, 18% and 27% respectively).

[40] The surface area to volume relationships were distinct and characterized by the size classes of wood pieces on the jams (SWD, MWD, LWD, and key members). These size classes were used to define the stages of removal, where stage B had all classes, stage C was lacking the SWD, and stage D was only the key member(s). Therefore the stages of removal have distinct relationships between surface area and volume (Figure 5). Power law regression formulas for volume in terms of surface area (equation 2) explained the observed points well (R2 = 0.72 − 1.00). The rate at which Asurf varied with volume was greatest for stage B (b = 1.05), then stage C (b = 0.75) and least for stage D (b = 0.66). According to the predicted Asurf − V relationships, for a volume of approximately 1 m3, stages B and C had similar surface areas (38.3 and 39.3 m2 respectively) which was much greater than stage D (10.0 m2). At the maximum volume found in this study (12.5 m2), surface area on stage B increases an order of magnitude above stage D (556.4 m2 and 52.3 m2 respectively).

Figure 5.

Surface area and volume relationships as defined by stage of removal in the form Asurf = aVb. Stage A (wrapped jam) is not included because of the disconnect between the flow and the woody material in the jam.

4.1.3. Soil and Leaf Litter

[41] Compared to wood, soil and leaf litter played a very different role in the composition of jams. While wood pieces are rigid and irregularly shaped, soil and leaf litter can be compressed and molded into available spaces. At the studied jams, soil and/or leaf litter filled gaps within the matrices of wood. Jam 2 had the most significant amount of soil (0.19 m3) (Table 1). Jam 1 also had soil found in pockets of regularly dry sheltered areas totaling 0.08 m3. Jam 3 had no soil.

4.2. Velocity and Shear Stress Distributions

[42] Spatial patterns of velocity and shear stress were consistent among all three jams, with the greatest values at the adjacent patch and the lowest downstream. As woody debris was removed, velocities and shear stress increased at the downstream patches as more flow passed directly through the jams, decreasing velocities in the adjacent patch. Velocities and shear stress values in the upstream patch increased as backwater effects were dampened with the removal of material. Since we observed similar spatial patterns among the three jams, we only present the results for velocity and shear stress from jam 2 (section 4.2). Velocity and shear stress results for jams 1 and 3 can be found in Figures S1, S2, and S3 in the online supplemental material. Additionally, we focus on the high-flow results because differences between stages of removal are more pronounced and geomorphic changes are known to occur during these higher flows.

4.2.1. Upstream Velocity Fields

[43] Upstream velocity fields remained fairly constant between stages of removal and were mostly spatially homogenous (Figures 6, 7a, and 7d). Small changes in mean velocity occurred between stages of removal, the most significant of which was between stage D (92 cm s−1) and the other three stages (75, 76 and 81 cm s−1 respectively) (Figure 6). Similarly, the mean shear stress values of the first three stages did not differ (18, 19, 21 N m−2 respectively), yet were all statistically different (p < 0.01) than stage D (28 N m−2). The increase in shear stress for stage D did not exceed the critical shear for the D50 (Figure 7d).

Figure 6.

Velocity and shear stress distributions at jam 2 for all four stages of removal. Shear stress values (contours at 20 N m−2 intervals) are overlain by velocity vectors. Graphs represent planform view of debris jam and surrounding flow field, where the solid rectangle represents the location of the jam, the x axis is distance from the left bank (in meters), and the y axis is the downstream distance from the upstream flow boundary (in meters).

Figure 7.

Spatially divided mean velocity and shear stress values for each stage at both low- and high-flow conditions for jam 2: (a) velocity upstream, (b) velocity adjacent, (c) velocity downstream, (d) shear stress upstream, (e) shear stress adjacent, (f) shear stress downstream. Statistically significant differences (p < 0.10) between each stage are shown to the right of each graph. Dashed lines indicate the critical shear of the D50 and D16 for each affected area.

4.2.2. Adjacent Velocity Fields

[44] Throughout all stages of removal, velocities and shear stress values were higher in the adjacent patch than they were in the upstream and downstream patches. Shear stress values were relatively large (from 84 to 49 N m−2) compared to other areas. Upstream had a maximum mean shear stress value of 28 N m−2 and the maximum of downstream patch was 21 N m−2.

[45] During stage A (wrapped jam, no porosity), almost no flow passed through the jam, rather flow was routed around the jam and through the adjacent patch. With increased porosity of the jam (stages B − D), flow became more streamlined (reduced lateral velocity components; see Figure 6) as more flow passed directly through the jam. Mean velocities in the adjacent patch changed little between stages, ranging from 140 cm s−1 to 127 cm s−1 (Figure 7b), but shear stress values decreased significantly from 84 N m−2 at stage A to 49 N m−2 at stage B, and remained consistent for stages C and D (52 and 53 N m−2; see Figure 7e).

4.2.3. Downstream Velocity Fields

[46] Contrary to the abrupt changes in velocity and shear stress in the upstream and adjacent patches, values in the downstream patch increased gradually in response to the changing stages of removal, corresponding to the changing porosity. With no flow passing through the jam during stage A, there were extremely low velocities in the downstream patch. The only flow moving through the structure was close to the bank where the key member was perched on a boulder creating space between the jam and the bed. With increased porosity, velocities increased progressively from stages A to B (from 22 to 40 cm s−1), and from B to C (62 cm s−1). In stage D, velocities were nearly consistent from upstream (92 cm s−1) to downstream (78 cm s−1; see Figures 6 and 7). Similar to velocity, there were gradual changes in shear stress between stages of removal (4, 9, 14, 21 N m−2 for stages A, B, C, and D respectively).

4.2.4. Excess Shear Stress

[47] We measured shear stresses that were in excess of critical shear for D50. Excess shear stresses were concentrated and in localized cores determined by the relative dominance of different flow paths around or through the jam. The dominance of flow paths was a function of the porosity of the jam. As porosity of the jam decreased, the dominant flow path shifted from around the jam (adjacent patch) to through the jam (downstream patch).

[48] Two cores of excess shear stress were present in stage B, one at the adjacent patch localized around the flow convergence zone, and the other in the downstream patch where flow was accelerated under the jam (Figure 8). At low porosity (stage A), the high-stress area was fully concentrated at the flow convergence zone off the tip of the jam. As porosity increased and the dominant flow path shifted away from the tip and through the jam, the size of the core of excess shear adjacent to the jam decreased, and a new core of excess shear formed downstream. The size of this core increased as porosity of the jam increased.

Figure 8.

Excess shear for high-flow conditions at jam 2. Excess shear is defined as the measured near-bed shear stress divided by the critical shear for the median particle size. Any value greater than 1 indicates particle entrainment. Cores of excess shear exist and shift from one stage to the next.

4.3. Drag Force

4.3.1. Drag Force Equation: FD, AF(emp), CD, (CDAF)calc

[49] Observed values of total drag, FD ranged from 22.5 kN for stage B of jam 1 to 0.3 kN for stage D of jam 3 (Table 2). The longitudinal component of drag force (FD(x)) contributed the majority to FD and ranged from 22.5 kN for stage B of jam 1 to 0.3 kN for stage D of jam 3. The lateral component (FD(y)) only contributed a small amount to the total force ranging from 0.3 kN to 2.1 kN, and had no discernible pattern. The decreases in FD at each jam were consistent with stage of removal, as the highest FD was associated with the greatest amount of material and lowest porosity for each jam.

Table 2. Drag Force (FD) and Its Longitudinal (x) and Lateral (y) Components for Each Jam at Each Stage of Removala
Measured Approach
 FD(x), kNFD(y), kNFD, kNFrontal Area, m2Velocity, m/sCDAF, m2CDFr
  • a

    Frontal area (AF(emp)) and UAp were measured in the field and were used to solve for the drag coefficient (CD). The term representing drag form (CDAF)calc was back calculated from the drag equation.

Jam 1
Natural22.50.422.514.61.1037.32.60.33
Partial20.31.020.38.81.0536.94.20.31
Key member19.90.719.96.71.0039.85.90.28
 
Jam 2
Wrapped18.22.118.36.31.0036.65.80.35
Natural15.00.415.16.30.9831.34.90.34
Partial14.40.414.43.81.0028.87.60.35
Key member11.60.411.63.01.0122.77.50.35
 
Jam 3
Wrapped8.20.48.22.70.8224.39.00.30
Natural7.60.57.62.70.7823.78.80.29
Key member0.30.30.31.30.80.90.70.30

[50] Field-determined AF(emp) was a function of the total amount of material on the jam, decreasing between stages of removal. Also between jams, AF(emp) was larger at all stages for jam 1 (14.6−6.7 487 m2) than jam 2 (6.3−3.0 m2) and jam 2 was larger than jam 3 (2.7−1.3 m2).

[51] Values of CD back-calculated from FD, AF(emp), and UAp ranged from 0.7 to 9.0 (Table 2), and showed little systematic variation. In stage A, jam 3 had a CD of 9.0, decreasing to 8.8 for stage B, and to 0.7 for stage C. In contrast, drag coefficients for jams 1 and 2 did not decrease with stage of removal. Jam 1 had its highest CD in stage D (5.9) and lowest in stage A (2.6). In contrast, jam 2 had the highest CD during stage C (7.6) and lowest in stage B (4.9).

[52] Values of (CDAF)calcback-calculated from empirical values of FD and UAp decreased with stage of removal for jam 2 (36.6 to 22.7 m2) and jam 3 (24.3 to 0.9 m2). Within jam 1, (CDAF)calcshowed no systematic change with its highest value of 39.8 for stage D and the lowest value of 36.9 for stage C.

4.3.2. Relationship Between AF(emp), FD, (CDAF)calc and CD

[53] Drag force and (CDAF)calc were both related to AF(emp) between stages and jams (Table 2). A further consideration of the connection of AF(emp) with (CDAF)calcrevealed that the systematic removal of material from the jams and reduction of AF(emp) at the same jam resulted in a greater proportion of (CDAF)calcdue to CD (Figure 9). Therefore, while the magnitude of CD, back calculated from FD, did not exhibit any consistent trends between stages of removal, it did gain importance in determining drag.

Figure 9.

Proportion of (CDAF)calc made up by AFemp and CD. The comparison of the back-calculated drag form term (CDAF)calc, which serves to represent both the frontal area and drag coefficient as defined in the traditional drag force equation, shows that as AFemp decreases with the removal of material, CD gains importance in the determination of drag.

4.3.3. Influence of Jam Structure and Composition: AF(emp), V, Asurf, and Porosity

[54] The values of AF(emp),V, and Asurf were used as the metrics for quantitatively describing jam composition, including the range in the number and size of the wood pieces. Volume of material did not predict either FD or (CDAF)calcwhen using all three metrics of composition (run 1, Table 3). On the contrary, AF(emp) was a significant predictor for both FD and (CDAF)calc, when using all metrics of jam composition (run 1). Volume and AF(emp) were highly correlated within the first run of the model (R = 0.82), which included Asurf. Asurf lost its predictive power when V was taken out of the model for both FD and CDAF (run 3). Therefore, because AF(emp) and V were highly correlated, and AF(emp) was a better predictor for both FD and (CDAF)calc, the model including only AF(emp) and Asurf (run 4) was sufficient for describing drag force and drag form. This combination of variables describes both the size of the jam (AF(emp)) and the composition (Asurf). For run 4 the coefficient for Asurf is negative. This indicates that for a jam of a given size (AF(emp)) the drag force decreases as the number of pieces making up that jam increases.

Table 3. Results From the Multivariate Linear Regression Model Run on the Independent Variables V, Asurf, and AF(emp) for Predicting Both (CDAF)calc and FDa
Model RunIndependent Variables(CDAF)calcFD
CoefficientSignificanceCoefficientSignificance
  • a

    A fixed effects model was used to account for the between jam and between stage variations.

  • b

    Significant at p < 0.05.

1V−1.30.570.60.93
 Asurf−0.10.01b−47.90.003b
 AF(emp)8.10.002b3382.510.01b
2V−4.10.22−11540.41
 AF(emp)7.30.03b3034.80.04b
3V7.10.03b3584.90.02b
 Asurf−0.10.26−380.25
4Asurf−0.10.003b−47.40.005b
 AF(emp)7.3<0.001b3428.2<0.001b

[55] The difference in hydraulics between stage A (wrapped jam) and B (natural jam) for jams 2 and 3 can be attributed solely to porosity since frontal area did not change. Lacking porosity, FD was ∼20% greater in stage A than in stage B for jam 2, and ∼10% greater in stage A than in stage B for jam 3: the force on jam 2 during stage A was 18.2 kN, decreasing to 15.1 kN for stage B, and the force on jam 3 for stage A was 8.2 kN and decreased to 7.6 kN during stage B. Since frontal area remained the same between the two stages of removal, the drag coefficient accounted for the increase in FD, increasing by 17% for jam 2 and 3% for jam 3.

5. Discussion

5.1. Detailed Look at Jams in a Mountain River in the Northeast United States

5.1.1. Jam Composition

[56] The results from this study provide a detailed description of individual deflector debris jams in a high-gradient mountain river in the Northeast United States. While our findings are limited to a single reach on a regulated river, they fill in a significant gap in the literature. Previous research has established a solid understanding of the reach-scale controls on wood debris recruitment, retention, and accumulation, highlighting the importance of wood material, hydrology, and geomorphology (see review by Gurnell et al. [2002]). However, the current application of woody debris jams in river restoration (e.g., engineered log jams [Abbe et al., 2003]) is inconsistent with the scale and focus of previous studies. Data on the formation and structure of individual woody debris jams therefore is critical, yet lacking.

[57] Abbe and Montgomery [2003] provide a detailed physical inventory for single debris jams within a relatively pristine watershed in the Pacific Northwest, United States. Although we worked within a reach with different land use histories (heavily logged), hydrology (dam regulated), and geomorphology (constrained river), our jams were similar to those in a different physiographic region studied by Abbe and Montgomery [2003]. We found that the type of jam we studied, the combination bank-deflector jam, had similar physical characteristics and were found in a similar location within watersheds. Abbe and Montgomery [2003] distinguish between three types of debris found on these jams: key members, racked pieces, and loose pieces. The key members are locally derived in situ pieces which serve to anchor debris and begin the accumulation of the racked pieces (LWD and MWD). These racked pieces lodge against the key member and act as the framework in which the loose pieces fill in the open spaces (SWD). The jams in this study were constructed similarly. Most of the key members were locally derived and had frameworks of large and medium woody debris. While our study was limited to a single type of jam (a bankdeflector jam affecting less than 25% of the flow), the similarity between their study and ours may indicate a consistency in jam form across watershed characteristics.

5.1.2. Effect of Debris Jam on the Channel Bed

[58] This study has shown the localized effect of debris jams on the velocity and shear stress distributions. With changing jam composition, high shear stress cores exceeding τc50 shifted spatially. Those locations of excess shear stress indicate potential scouring of the bed and may result in future pool formation or enhancement.

[59] Previous studies of scour around deflector debris jams have suggested that pool formation occurs in the convergence zone off the tip of the structure [Cherry and Beschta, 1989; Buffington et al., 2002]. We observed cores of high excess shear stress primarily at low-porosity stages of removal (stages A and B, Figure 7). Therefore only after a jam has accumulated enough material to substantially decrease its porosity will flow be sufficiently concentrated at the tip, thereby creating the predicted scour hole. For increasing porosities, flow will be accelerated through one or multiple holes under and/or through the debris jam, resulting in localized areas of scour immediately downstream of the jam. While previous studies guide us in general patterns of upstream and adjacent erosion and downstream deposition [Abbe and Montgomery, 1996; Buffington and Montgomery, 1999a], our study suggests that the complex pattern of erosion and deposition associated with natural jams is first dependent on the degree of porosity, which can often be associated with the age of the jam. A jam with near-zero porosity will then default to a somewhat random spatial distribution of erosion and deposition around the debris jam based on the random nature of wood piece accumulation.

5.2. Drag on Natural Debris Jams

5.2.1. Drag Coefficient and the Drag Equation

[60] Solving for the drag force on a debris jam through the momentum extracted through a control volume may be compared to those studies which measure drag directly. A small number of flume studies have reported the drag force on woody debris, and our calculated values of FD are reasonable in comparison. Wallerstein et al. [2001] quantified values of 5 to 23 kN on a single cylinder and Shields et al. [2004] computed a drag force of 14 kN on a man-made large wood formation with frontal area similar to jam 1. The values of FD on our debris jams (0.3−22.5 kN) are well within the range of those measured in a more controlled setting. However, our values are likely overestimates of the actual drag force on the debris jams. The difficulty of working in a natural environment potentially created unquantified forces, including the loss of momentum due to the spill resistance [Wilcox et al., 2006] and the drag when wood creates surface waves [Wallerstein et al., 2002].

[61] The majority of studies of hydraulic forces on wood jams have focused heavily on CD because the other variables in equation (1) are easily quantified for simple cylinders whereas CD cannot be measured directly [Alonso et al., 2005]. Studies of woody debris elements show trends in CD values with changing log submergence, log slenderness [Wallerstein et al., 2002], blockage, orientation, distance from the bed [Gippel et al., 1996; Hygelund and Manga, 2003], and Reynolds and Froude numbers [Wallerstein et al., 2001]. All of these studies have only used the single-log model resulting in a range of CD from 0.4 to 4.5 in the flume [Gippel et al., 1996] and 1.0 − 3.3 [Hygelund and Manga, 2003] in the field. In contrast, our study has quantified the drag on natural debris jams in the field resulting in CD values from 0.7−9.0. CD did not change systematically with the amount of material and porosity (Table 3). While CD declined rapidly once the accumulated material was removed from jam 3 (from 9.0 to 0.7), it increased for jam 2 (4.9 to 7.5). Thus, given the consistent flow characteristics (Fr ranged between 0.29 and 0.35) we did not observe a clear relationship between jam porosity and CD.

[62] These differences observed at our site can most likely be attributed to some of the changes investigated in previously published studies. For example, removing material may alter the location of the jam in the water column. In the flume studies of Gippel et al. [1996], it was found that CD decreased as the relative depth of the “log” increased. Focusing on the change in CD for jam 1 we observe this trend. From the natural (stage B) to partial jam (stage C), CD dropped (Table 3). For this jam, the majority of the 3.09 m3 of material removed was below the key member. By creating space at the base of the jam, not only did the frontal area decrease but the relative depth also decreased. When the remaining material was removed, the frontal area decreased more rapidly than the relative depth, resulting in an increase in the drag coefficient. As shown above, the drag coefficient will change in a predictable pattern for single logs. However, CD for natural jams is much less predictable because its sensitivity to changing geometry and location in the water column becomes masked by changes in overall jam composition, including AF(emp). Assumed, single characteristic values of CD for natural debris jams are likely misleading.

[63] A complication in studying natural debris jams is separating CD from AF(emp). In reality, natural debris jams have poorly defined geometry, particularly frontal areas, because of their irregular shape and porosity, which also contributes to being poorly described by a single value of CD for all types of jams. Using (CDAF)calc acknowledges that drag coefficient and frontal area have a combined effect on the drag force, but their individual contributions are less important to know and likely to vary considerably between jams.

[64] Surprisingly, our results showed that (CDAF)calc for natural debris jams can be explained by the simple measure of AF(emp) (equation (1)) (Table 3). Field-determined AF(emp)was simply the height of the jam multiplied by its width measured in segments along the key member(s) in order to most closely represent a rectangular surface. It is important to note that AF(emp) did not explain all the variation in (CDAF)calc or FD as would be the case for a simple log model. Using both AF(emp) and Asurf provides greater information on the size (via AF(emp)) and material composition (via Asurf) of the debris jam, and using both terms explains more variability in (CDAF)calc and FD than either alone (Table 3). In addition, V was highly correlated with AF(emp), suggesting that they are somewhat interchangeable as indicators of jam size for this type of jam. However, Asurf only strengthened the model for predicting both FD and (CDAF) calc when combined with AF(emp).This may be due to the high correlation between Asurf and V. Therefore both the frontal area, as well as the surface area of woody debris within a jam, are important in quantifying the effect of variable wood sizes on the drag on an entire debris jam.

5.2.2. Surface Area and Porosity

[65] The importance of increased surface area has been widely studied in respect to its increased ecological function [Harmon et al., 1986]. Hydraulically, the role of wood surface area has been explored in terms of added roughness on a reach scale, lumping the effects of wood as single pieces and within jams into overall added resistance. The increased channel roughness affects flood peak time [Gregory et al., 1985] and water level height [Gippel et al., 1996]. The roughness associated with surface area of single logs has also been investigated experimentally. Adding branches to logs, both Hygelund and Manga [2003] and Gippel et al. [1996] reported a decline in CD, but for different reasons. While Hygelund and Manga found no increase in the drag force and therefore a decrease in the drag coefficient, Gippel reported an increase in the drag force concurrent with a larger increase in the frontal area.

[66] Greater surface area due to more pieces of woody debris within a jam increases the total roughness, or the total surface of wood interacting with the flow. Such added roughness has the potential to account for the increased resistance. Studies on wind through trees have shown that the greater the volumetric porosity (the proportion of the total dimensions filled by leaves and branches), the greater the drag [Grant and Nickling, 1998]. Open space, or porosity, is an indicator of the fraction of the total surface area that is interacting with the flow. The degree to which a structure is porous can affect the local hydraulics by increasing or decreasing roughness and hydraulic resistance.

[67] Debris jams are porous because of the presence of irregularly shaped wood pieces which range in size. Conceptually, medium pieces can “fill the holes” created by the large ones and small pieces do the same for medium pieces. While some of the void space can be filled by soil and leaf litter, not all of it is. This discrepancy represents a basic assumption many make when modeling debris jams for real-world applications: that enough sediment and fine organic matter has essentially filled all of the holes making the jam nonporous [Shields and Gippel, 1995]. Our results indicate that debris jams are highly porous structures, even with large volumes of soil and organic matter in them. Our experimental design allowed quantification of the difference in FD between a natural jam and one whose porosity is close to zero. The differences observed represent potential errors that studies introduce when modeling a natural jam as a single, nonporous object. By assuming solid debris jams, previous studies have over estimated CD, and thus overestimated the drag force exerted on natural debris jams.

[68] Natural jams also differ from the single-log model frequently used in both flume studies and stability analyses for debris jam structures [D'Aoust and Millar, 2000]. Representation of a jam as a key member underestimates the potential drag force on the object as the accumulation of additional wood pieces increases the drag force. Our results have shown that this increase in drag due to the accumulation of wood pieces may vary greatly, ranging from around 30% on jam 2 to an order of magnitude change on jam 3 (Table 2).

5.2.3. Theoretical Predictions Between Debris Jam Types

[69] The rate at which surface area increases relative to volume depends on jam composition. Since jams are complex structures which amass a wide range of pieces, the relationship between Asurf and V is a continuum. On the basis of the classification of wood piece size and the composition of the jams we studied, we have grouped jams into three types: (1) natural jams with complete accumulations including the key member, SWD, MWD and LWD; (2) partial jams lacking SWD with a framework of accumulated MWD and LWD; and (3) key member jams with only one or multiple large trunks.

[70] Differences in jam type can be defined by a relationship between (CDAF)calcand the volume and surface area combination, and we present these relationships as a contour plot (i.e., (CDAF)calc contours, Figure 10). While (CDAF)calc will vary with discharge, values presented are for a bankfull event during which the flow either just overtops the jam or almost overtops it. Therefore, while (CDAF)calc is independent of velocity in this situation, it does represent a limited range of flow conditions. In many cases, these are the flow conditions in which most restoration projects are interested because of the capability of the high flows to cause geomorphic change and instability of the structure.

Figure 10.

Contour plot of (CDAF)calc based on the AsurfV relationship for each stage (see Figure 4). Volume was used as the predictor of both Asurf and (CDAF)calc, making it easy to predict differences in drag on the basis of the nature of wood pieces for a given volume of them. Open circles represent AsurfV characteristics of three data points from single-log models. Solid circles are from Wallerstein et al.'s [2001] flume study, and the dashed circle is from Hygelund and Manga's [2003] field-based study. Wallerstein et al.'s [2001] drag values are 7–10 times less than we predict on the basis of our data. Hygelund and Manga's [2003] range of drag values for their AsurfV “logs” are nearly 2 times larger than what we predict on the basis of our contour plot.

[71] The (CDAF)calc contour plot has three distinct regions, roughly corresponding to the three jam types. In the range of Asurf − V corresponding to key member jams found in this study, small increases in surface area result in large changes in drag. As pieces are added to a key member, surface area increases quickly relative to volume, leading to an increase in (CDAF)calc. As wood pieces, in particular SWD, are added to a partial jam, surface areas become extremely large. However, the increase in (CDAF)calc from partial to natural jam will be relatively minor in comparison to the increase in (CDAF)calc from key member to partial jam. In terms of volume, these generalized relationships dictate the sensitivity of a particular jam type to the addition of material. In sum, these calculations suggest that relatively small additions of material to key members will cause large increases in drag, but that drag on natural jams is much less responsive to changes in material.

6. Implications of Findings

[72] Our results show that jam composition has an important influence on the creation of habitat. The size and number of wood pieces influence the porosity, which we found to have a control on the velocities upstream, adjacent and downstream of the jam, the shear stress distribution, and the drag force. Velocity adjustments provide areas of low flows in close proximity to high flows, thereby resulting in hydraulic complexity. High shear stress cores forming scour holes and areas of deposition provide geomorphic complexity. The interaction of both the hydraulic and geomorphic complexity enhances the habitat value [Zalewski et al., 2003; Lepori et al., 2005].

[73] Second, our results have demonstrated that the existing science on the hydraulics of woody debris may be at times misleading. Detailed flume-based studies on the local hydraulics of woody debris in rivers have provided the tools for engineering log jams and predicting their success [D'Aoust and Millar, 2000]. The majority of these studies use the single-log model, as general equations have been empirically derived and CD values solved for [Young, 1991]. Yet our results show that the relationship between single logs and complete debris jams is complex and nonlinear. The very nature in which jams form, as accumulations through time, means that their character continually changes with season and variations in discharge [Lienkaemper and Swanson, 1986] as well as through space [Kraft and Warren, 2003] and with channel geomorphology [Piegay and Gurnell, 1997]. Inputs of wood into rivers and standing biomass have been shown to vary widely depending on factors such as location in watershed [Bilby and Ward, 1989], land use, and disturbance history [Gregory and Davis, 1992]. Most studies show these trends through a reach, which tends to average out the individual jam characteristics. The translation of these watershed- and reach-scale variables into the character of single debris jams is still not well understood. We have shown that the number and size of pieces, as well as the arrangement of these pieces which is largely determined by broader geomorphic and land use conditions, can significantly affect the hydraulics and geomorphology of individual debris jams.

[74] This study has begun to address how the complex nature of jams can be broken down into estimable and/or measurable variables, looking at factors for the size of the jam (AF(emp) or V) and the character of the pieces (Asurf). By characterizing debris jams in this way, we may further our understanding of their formation and how their construction and composition affects the local hydraulics, geomorphic form and ecologic value. Varying wood populations, delivery mechanisms, frequency of delivery, hydrology and bed material may significantly affect the relationships between jam composition and hydraulics. This may be especially true in environments where woody debris forms fundamentally different jams.

7.

Notation
a

coefficient in surface area−volume relationship

A

area of control volume boundary, m2

Abed

area of bed within control volume, m2

AF(emp)

empirical debris jam frontal area, m2

Asurf

surface area of wood pieces, m2

b

exponent in surface area−volume relationship

CD

drag coefficient

(CDAF)calc

drag character, m2

F

external forces, N

F1,2,3

hydrostatic force across control volume surface, N

Fboundary(x,y)

force components exerted on bed and banks, N

FD

drag force, N

FD(x,y)

drag force components, N

Fcv(x,y)

force components exerted on the control volume, N

h

water depth, m

k

von Karman's constant

p

pressure, N m−2

ρ

density of water, kg m−3

τbed

local shear stress, N m−2

τbed(x,y)

local shear stress components, N m−2

τc*

Shields value, N m−2

τc50

critical shear stress for the D50, N m−2

u*

shear velocity, m s−1

U

depth-averaged velocity, m s−1

U1,2,3

depth-averaged velocity across control volume surface, m s−1

UAP

far-field approach velocity, m s−1

Ux

longitudinal velocity component, m s−1

Uy

lateral velocity component, m s−1

V

volume of wood pieces, m3

z0

bed roughness length, mm

Acknowledgments

[75] This study was funded by NSF grant DEB-04150365 with additional support to Doyle from a UNC Junior Faculty Award. We are appreciative of the field assistance given by Arjun Dongre, Chelsea Lane-Miller, and Ted Treska. Randy Fuller played a fundamental role, offering helpful insight and logistical support throughout the project, and Peter Wilcock provided valuable input on several issues of data collection and analysis. Extensive field work required to collect the data in this study would not have been possible without the help of Wild Water Outdoor Center, Rick Fenton at the NYDEC, and the staff of Cornell University's Little Moose Field Station. Finally, we would like to thank Jason Julian for an early review and Doug Shields for a thorough review and numerous discussions of this paper.

Ancillary

Advertisement