Effects of Debris on Bridge Pier Scour (2010)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

92.1 Review of Current Practice 2.1.1 Introduction Debris (or drift), for the purposes of this project, is defined as floating woody debris that is delivered to and transported along a stream or river. The effects of debris on bridges have been well documented (see for example Chang and Shen 1979, Diehl 1997, Parola et al. 2000). Debris prob- lems are most common on rivers and streams with active bank erosion and that drain wooded or forested areas or corridors. As discussed in Chapter 1, there is a pressing need for state DOTs and other bridge owners to have improved prediction methods for the geometry (size and shape) of typical debris accumulations; the conditions under which debris can be expected to be recruited, transported, and deposited on a bridge; and the resulting depth and extent of scour at bridge piers. Currently, only limited guidance is available on which to base critical public safety decisions during flooding on debris-prone rivers. Consequently, there is a need for accurate methods of quantifying the effects of debris on scour at bridge pier foundations for use by DOTs and other agencies in the design, operation, and maintenance of highway bridges. Both FHWA (Diehl 1997) and NCHRP (Parola et al. 2000) have published recent debris-related studies that have excellent reference source lists. These sources were screened and a pre- liminary working bibliography of the most relevant references was assembled. Additional sources were evaluated as well. For example, a laboratory and field study on single-pier debris accumulation was completed at Purdue University in 2001 and presented at the TRB 82nd Annual Meeting (Lyn et al. 2003a). The Purdue study was subsequently published through the University’s Civil Engineering Joint Transportation Research program (Lyn et al. 2003b). Of course, studies such as the ini- tial work by Laursen and Toch (1956)—which characterize the scour pattern at a pier for floating, submerged, and buried debris—were revisited as well. The international literature was also reviewed, but little information or data were forthcoming. For example, following the devastating damage to infrastructure by Hurricane Mitch in 1998 in Honduras, Nicaragua, and Guatemala, a major international relief and recovery effort was launched. Emphasis was on replacement of infrastructure, particularly bridges, as hundreds of bridges had been lost because of scour and debris accumulation. Although a number of universities in Central and South America conducted research on the debris prob- lem to develop improved design procedures for debris, the results of this work have not made the mainstream literature in the United States. Some studies have been conducted in Australia and New Zealand (see for example Dongol 1989, Young 1991, Shields and Gippel 1995, and Gippel et al. 1996). In Europe, very few studies on debris have been conducted because the lack of significant riparian forest corridors have made large woody debris rare. Also, debris accumulations are seen as a river management problem and are routinely cleared from river channels (Piégay and Gurnell 1997). Much of the recent literature pertains to large woody debris (LWD) or coarse woody debris (CWD) and is associated with riparian forest ecology and aquatic habitat concerns and the influence of LWD on flow patterns, stream mor- phology, and alluvial processes (Hickin 1984, Fetherston et al. 1995, Wallerstein et al. 1997, Abt et al. 1998, Dudley et al. 1998, Duncan 2000, Manga and Kirchner 2000, Ringgold et al. 2000, Haga et al. 2002, Rollerson and McGourlick 2002, Hygelund and Manga 2003, Kraft and Warren 2003, Montgomery et al. 2003, Simon et al. 2004). As a result, much of the literature on LWD in streams and rivers is associated with these types of studies and most studies are associated with work conducted in old-growth coniferous forests of the Pacific Northwest. Some studies, such as the one conducted by Robinson (2003) on the White River in Indiana, attempt to document existing debris production, transport, and storage conditions. Yet, other than the reviews by Gippel (1995), Diehl (1997), and Parola C H A P T E R 2 Findings

et al. (2000), there is still a paucity of literature on the processes associated with debris delivery to bridge sites. Three primary processes are associated with debris accu- mulation at bridges: (1) debris source loading, distribution, and recruitment, which is defined as drift generation by Diehl (1997); (2) debris transport; and (3) debris accumulation, deposition, and storage. Diehl (1997) provides a relatively concise review of the literature on these processes. The follow- ing is an expanded review covering the literature that was not included in or was subsequent to the reviews conducted by Chang and Shen (1979), Diehl (1997) and Parola et al. (2000). The References section provides a bibliography of current practice. 2.1.2 Debris Source Loading, Distribution, and Recruitment Montgomery and Piégay (2003) indicate that the geomor- phological effects of wood in streams and rivers are highly variable and are reflected in the differences in wood size, den- sity, and shape that are partly controlled by wood availability, depending on external factors controlling wood recruitment, and the character of the surrounding forest as well as on stream size, characteristics, and processes. However, the observations of Lyn et al. (2003b) and many other researchers would sug- gest that the recruitment, transport, and accumulation of debris appears to be a somewhat random process while the results of Manners et al. (2007) show that “the relationship between individual logs and complete debris jams is complex and nonlinear.” Woody debris recruitment pertains to the processes by which live trees and standing snags fall along a stream corri- dor or where existing forest deadfall and residues from logging and clearing activities are delivered to the stream channel for potential transport downstream to a bridge site. The avail- ability of wood for recruitment is, in part, dependent on the distribution, density, and health of forested hillslopes, flood plains, and riparian corridors along a stream. Most woody debris delivered to streams originates from that part of the riparian forest closest to the channel. For example, based on work conducted in the Pacific Northwest, Fetherston et al. (1995) noted that 70% to 90% of riparian input of LWD occurs within 100 ft (30 m) of the channel edge. Most studies to date confirm that LWD recruitment is driven by a fluvial process through channel meandering and attendant bank undercutting, mass failure, and input of trees. In upland, low order streams, landslides and debris flows are the principal mechanism by which trees and other organic debris are delivered to a stream channel. Although wind- throw in some areas can be a significant contributor of woody debris to streams and rivers, in medium to high order streams and rivers, which may or may not have flood plains, the pri- mary methods of delivery are associated with stream meander- ing and bank erosion/undercutting. Nakamura and Swanson (1994) concluded that channel width and sinuosity are the primary factors that influence distribution of storage sites for LWD. But Wallerstein et al. (1997) suggest that reach stabil- ity and channel sinuosity are probably better measures of debris volume and frequency of debris jams because these factors, to a large extent, determine the rate of debris input. Empirical and theoretical analyses of the probability of input of woody debris to a channel as a function of distance from the streambank have been developed (Murphy and Koski 1989, Robison and Beschta 1990, Van Sickle and Gregory 1990, Andrus and Lorenzen 1992, Downs and Simon 2001, Teply 2001). Robison and Beschta (1990) indicate that the possible surface area that could be impacted is defined by the area of a circle whose radius is equal to the effective tree height (He) (Figure 2.1), assuming that a tree has an equal chance of falling in any direction. The effective tree height of a conifer is consid- ered to be the total tree height minus about the last 5 ft (1.5 m) since the crown of the tree is not considered to qualify as coarse woody debris. According to Robison and Beschta (1990), the probability, P, of a tree’s falling and delivering coarse woody debris to the channel is proportional to the ratio of the arc distance, AD, along the stream to the total arc distance (circumference) of the circle: where D is the distance of the tree from the stream (Figure 2.2). Robison and Beschta (1990) suggest that the integration of P AD H D H e e = = ( ) ° − 2 180 1 π cos (2.1) 10 Figure 2.1. Potential tree fall area showing total tree height, Ht; effective tree height, He (used to determine CWD delivery from conifers); and total arc distance, 2π He. t H H2 H H e ee Source: after Robison and Beschta (1990)

this model with tree-growth and “fall-down” (risk-rating) models may be useful for identifying trees that will have a high probability of providing coarse woody debris to a stream in later years. Once a tree falls, its limbs and branches may break off, adding to the debris litter on the flood plain. In addition, the tree trunk may break into smaller pieces when hitting the ground or other large upright trees. The broken limbs and small branches, twigs, and leaves are available for recruitment as well. Often, when a tree falls, its rootwad (or root bole) is still attached, particularly if the tree is a product of bank ero- sion. Woody debris that is available for transport may be found in all positions and orientations on the flood plain, on top of and along the stream bank, in the channel, and on bars and islands. Gregory (1991) reports that the majority of woody debris along streams in the McKenzie River basin in Oregon is retained along channel margins and flood plains with less than 30% of the debris volume occurring within the active chan- nel. He also indicates that reaches with broad flood plains and complex channels make up less than 20% of the channel length, but contain more than 50% of the large woody debris. However, Andrus et al. (1988) found no correlation between riparian stand volume in sample plots and the volume of new debris in adjacent channel reaches. In a similar plot survey, Robison (1988) found that while the riparian stand, stream morphology, and debris in the channel were interrelated, there was too much variability to distinguish definite rela- tionships. Gippel et al. (1996) suggest that “the general lack of correlation between the distribution of debris in streams and the distribution of adjacent riparian trees is indicative of the importance of re-distribution of debris by flood events, the potentially long residence time of wood in streams, and the re-exposure of ancient wood buried in flood plain sediments as rivers alter their course.” Some studies have also been conducted to evaluate debris recruitment on incised or degrading streams in the southern United States (Downs and Simon 2001, Wallerstein et al. 1997, Wallerstein and Thorne 2004). In these studies, the authors suggest that debris recruitment can be tied directly to channel stability and sinuosity. Through an understanding of incised channel evolution, they suggest that the spatial density of debris jams in degrading rivers can be qualitatively predicted on the basis of location and migration of headcuts and degra- dational processes. Based on studies conducted on degrading streams in northern Mississippi, Wallerstein et al. (1997) and Wallerstein and Thorne (2004) note that debris input was found to be the result of the following key mechanisms (per- centages are rounded): • 37% due to outer bank erosion in channel bends • 36% due to bank mass-wasting in degrading reaches • 12% due to windthrow • 7% resulting from paleodebris (material introduced into the channel from old alluvial deposits containing preserved debris) • 5% initiated by large logs floated in from upstream • 4% from beaver dams Wallerstein and Thorne (2004) suggest that the possibility exists for developing a geomorphic tool to predict the distri- bution of debris inputs and jams on the basis of the Incised Channel Evolution Model (ICEM). The species composition of the riparian vegetation influ- ences the amount and distribution of LWD; species that achieve large size produce more stable, longer-lived debris than smaller species. For example, in the Pacific Northwest, chan- nels located in younger communities, which are often domi- nated by smaller hardwood species, have smaller average-size pieces of LWD compared to streams flowing through mature stands of conifer (Bilby and Ward 1991). Table 2.1 presents a list of the characteristics and distribution of LWD for various geographic and channel network locations as compiled by Lassettre and Harris (2001). 2.1.3 Debris Transport The mobilization and transport of woody debris is depen- dent on the physical characteristics of the piece as it relates to channel width; diameter of the piece as it relates to flow depth; orientation of the piece within the channel; and, to a lesser degree, channel slope. The size of LWD accumulations increases downstream while the frequency decreases (Swanson et al. 1982, Bilby and Ward 1989) because LWD on small, low order streams generally tends to be longer than the channel 11 Figure 2.2. Schematic illustrating the probability of coarse woody debris falling into a stream from a tree located (A) at the edge of the stream, (B) at a dis- tance less than the effective tree height, and (C) at a distance greater than the effective tree height. AD AD D D He He He D AD Stream (A) ≥50% Probability (where D=0) (B) 1-49% Probability (where 0<D<He) (C) 0% Probability (where D≥He) = Tree Location He = Distance of Tree from Stream = Arc Distance Along Stream Source: after Robison and Beschta 1990 = Effective Tree Height

12 Location Channel Network Position Results References Western Oregon 0.10 to 0.30 channel gradients LWD in small streams randomly distributed LWD easily transported in larger rivers leading to size sorting and accumulations in distinct jams Swanson et al. 1976 Western Oregon 0.02 to 0.50 channel gradients Debris loading highest in small, steep streams, decreasing downstream In 1st and 2nd order streams, LWD randomly located because streams too small to redistribute In 3rd to 5th order streams, flows large enough to redistribute debris from distinct accumulations that directly affect channel width In large rivers, LWD thrown on islands or on banks, having little influence on channel, except during high flows Swanson and Lienkaemper 1978 Indiana, North Carolina, Oregon 0.001 to 0.40 channel gradients Loading (kg/m2) decreased with increasing channel width, watershed area, stream order, and decreasing channel gradient Keller and Swanson 1979 Northwest California 0.01 to 0.40 channel gradients 1.0 km2 to 27 km2 drainage areas 2nd through 4th order streams Redwood debris usually dominates total loading Loading (m3/m2) decreased as drainage area and width increased Debris accumulations in lower reaches larger, more complex, and spaced further apart than in upper reaches Keller and Tally 1979, Tally 1980, Keller and MacDonald 1983, Keller et al. 1985 New England 1.5 m to 7 m wide channels Frequency of debris dams decreased from 1st order (20 to 40 dams per 100 m) to 2nd order (10 to 15 dams per 100 m) to 3rd order (1 to 6 dams per 100 m) streams Likens and Bilby 1982 Western Oregon 0.03 to 0.37 channel gradients 3.5 m to 24 m bankfull widths 0.1 km2 to 61 km2 drainage areas 1st through 5th order streams Amount of LWD generally decreased from small to large channels Lienkaemper and Swanson 1987 Western Washington 0.13 channel gradient (<7 m channel width) 0.08 channel gradient (7 m to 10 m channel width) 0.03 channel gradient (>10 m channel width) Mean diameter, length, and volume of LWD increased with increasing channel width Frequency of occurrence (number of pieces/m) decreased with increasing channel width Changes related to increased capacity of larger streams to move wood downstream Higher proportion of wood input remained in the stream channel as stream size decreased Bilby and Ward 1989 Southeast Alaska <0.03 channel gradients 0.7 km2 to 55 km2 drainage areas 1st through 4th order streams Abundance of LWD and volume per channel length (m3/m) increased with increasing stream size LWD loading (m3/m2) decreased with increasing bankfull width Robison and Beschta 1990 Western Washington 3 m to 24 m bankfull channel widths Average LWD volume increased with increasing stream size (bankfull width) LWD abundance (number of pieces/m) decreased with increasing bankfull width Bilby and Ward 1991 United Kingdom 110 km2 drainage area Density of LWD jams (number of dams per 100 m and number of dams per 500 m) decreased downstream from headwaters and with increasing channel width Abundance of partial spanning dams increased in downstream direction Debris loading (kg/m2) decreased in downstream direction Gregory et al. 1993 Table 2.1. Characteristics and distribution of LWD.

width and remains in place during most flow events whereas wood in larger, higher order streams is generally shorter than the channel width and is, therefore, more easily transported, leading to a reduction in LWD frequency due to flushing of smaller pieces and clumping of the remaining LWD pieces. Evidence suggests that pieces shorter than bankfull width and with a diameter less than bankfull depth are more likely to be transported downstream (Bilby 1984, Bilby and Ward 1989). Rootwads increase the stability of logs by increasing the poten- tial for snagging on instream obstructions and potentially increasing the log diameter to greater than the average bank- full depth (Sedell et al. 1988). Log position and orientation also influence the potential mobilization and transportability of a piece. The percentage of the piece anchored to the bank either by the rootwad or by burial, the proportion of the piece in the water, and the angle of orientation to flow contribute to the stability of in-channel wood (Bryant 1983). Once debris is available for recruitment, it generally remains in place until conditions are sufficient to mobilize and transport 13 Location Channel Network Position Results References Southeastern France 0.0012 to 0.0018 channel gradients 3700 km2 drainage area 6th order river LWD differentially deposited on banks depending on flow and forest conditions Piégay 1993 Western Oregon 0.022 average channel gradient 23 m average channel width 60 km2 drainage area 5th order channel Channel width and sinuosity were main factors controlling LWD supply and distribution Number and volume of LWD highest in wide sinuous reaches Nakamura and Swanson 1994 Southwest Alaska, Western Washington 0.002 to 0.085 channel gradients 2.5 m to 38 m channel widths Number of LWD pieces per m2 decreased with increasing channel width Logs in larger channels were more readily transported Inverse proportionality between pool spacing and LWD frequency Montgomery et al. 1995 North Central Colorado 0.005 to 0.065 channel gradients 4 m to 10 m bankfull width 2 km2 to 30 km2 drainage areas 1st through 3rd order streams LWD abundance (number of pieces/m) greater in smaller streams when sorted by drainage area and width Lower percentage of pieces spanned channel in larger streams (sorted by drainage area and width) than in smaller streams Percentage of LWD lying perpendicular to stream channel decreased with increasing drainage area Percentage of debris pieces in wetted channel decreased in larger streams Percentage of LWD above wetted channels increased in smaller streams LWD randomly distributed in streams <5.0 m width and clumped into jams in streams >5.0 m in width Richmond and Fausch 1995 Northwest Washington 0.002 to 0.05 channel gradients 5 m to 20 m channel widths 2 km2 to 120 km2 drainage areas Loading and abundance decreased with increasing channel width Channel width is dominant influence on number of pieces of LWD/m2 Beechie and Sibley 1997 Southeastern France 0.0012 to 0.0018 channel gradients 3700 km2 drainage area 6th order river LWD distribution on meander banks related to angle of bank to flow, height of flow, and presence of secondary channels Piégay and Marston 1998 Spain 0.08 to 0.1575 channel gradients 3 m to 8 m channel widths 0.4 km2 to 64 km2 drainage areas 1st through 3rd order streams Abundance and loading decreased in downstream direction Elosegi et al. 1999 Southeastern France 0.003 to 0.008 channel gradients 1650 km2 drainage area Most LWD in active channel on high bars LWD deposit controlled by deposit site morphology and proximity of LWD sources Piégay et al. 1999 Source: Lassettre and Harris (2001) Table 2.1. (Continued).

the debris. In steep, low order streams, woody debris mobi- lization and transport is accomplished primarily by debris tor- rents triggered by heavy rainfall and flood flows (Swanson et al. 1976, Keller and Swanson 1979, Nakamura and Swanson 1993). In intermediate and high order streams, flotation is the main mobilization mechanism (Lassettre and Harris 2001). Diehl (1997) suggests that the depth sufficient to float a log is about the diameter of the butt plus the distance the roots extend below the butt. He notes that this is roughly 3% to 5% of the estimated log length based on observations of typical large logs at drift study sites. Braudrick and Grant (2000) developed theoretical models of entrainment and performed flume experiments to exam- ine thresholds for wood movement in streams. They note that piece entrainment is primarily a function of the piece angle relative to flow direction, the presence or absence of a root- wad, the density of the log, and the piece diameter. They also note, contrary to previous studies, that piece length did not significantly affect the threshold of movement for logs shorter than the channel width. Although their model reasonably pre- dicted entrainment thresholds for pieces oriented at 45° and 90° to flow, they note that the model underestimated the flows at which pieces oriented parallel to flow moved. Because of the complexity of movement of wood in streams, they rec- ommend that a larger number of variables need to considered in predicting entrainment and that unequal forces exerted on different parts of the log, including the effects of flotation, need to be considered in predicting wood movement. Subsequently, Alonso (2004) characterized the transport mechanics of indi- vidual logs by examining information on a number of influ- encing factors and using a generalized modeling concept of log motion. Based on his study, he determined that log trans- port by unsteady streams can be characterized by estimating the hydrodynamic forces on a single cylindrical body and noted that drag and buoyancy are the main mobilizing forces. In general, most drift is transported as individual logs, which tend to move along the thalweg of the stream (Chang and Shen 1979, Lagasse et al. 2001). However, drift can commonly aggre- gate into short-lived clumps or jams, most of which are broken apart by turbulence as they move downstream or as they strike stationary objects. Using flume experiments, Braudrick et al. (1997) observed three distinct wood transport regimes: uncon- gested, congested, and semi-congested. They note that during uncongested transport, logs move without piece-to-piece interactions and generally occupy less than 10% of the chan- nel, whereas in congested transport, the logs move together as a single mass and occupy more than 33% of the channel area. Semi-congested transport is intermediate between these trans- port regimes. Flume experiments conducted by Bocchiola et al. (2008) showed that, even though logs tended to travel individ- ually rather than as a clump or large aggregation of material, wood pieces tended to travel further downstream when con- gested transport was observed. Diehl and Bryan (1993) note, based on work in the basin of the West Harpeth River in Tennessee, that curved pieces of debris were more likely to form intertwined jams. Even though the effect of shape was not examined in their flume studies, Lyn et al. (2003b) used natural twigs because it was decided that they represented a more “realistic” model log element. Diehl (1997) notes that most drift floats at the water sur- face in a zone of surface convergence (Figure 2.3) where flow is generally deepest and fastest and, as a result, is transported 14 X X Debris STRAIGHT CHANNEL REACH Debris Concave Outer Bank of BendConvex Inner Bank of Bend CURVING CHANNEL REACH Figure 2.3. Secondary flow patterns in straight and curving channels.

at about the average water velocity. In moderate radius bends, drift is observed more often along the thread of a stream (i.e., thalweg) between the center of the channel and the outside bank than in contact with bank vegetation (Diehl 1997). In contrast, submerged debris is transported near the bed by dragging, bouncing, or tumbling and is often deposited along the banks in straight reaches and on point bars in bends by the slower, diverging flow near the bed (Diehl 1997). Lyn et al. (2003b) also indicate that stable debris piles were more likely to develop in shallower flows or flow regions and to form at lower velocities. Contact with fixed objects also tends to strip the branches and longer roots from trees as they move downstream. Little physical research has been conducted to evaluate how far drift is transported and where it is deposited in streams. Braudrick and Grant (2001) note in a review of the literature that most studies inferred transport relations from mapped temporal changes in LWD distribution in first to fifth order streams; LWD moves farther and more frequently in larger (≥ fifth order) rather than smaller (< fifth order) streams; smaller pieces move farther than larger pieces; piece diame- ter, which strongly influences flow depth requirements for log entrainment and transport, influences travel distance; and channel morphology is also a factor in determining travel distance, because wide, sinuous reaches tend to promote deposition on the outside of bends and the head of bars and islands. Braudrick and Grant (2001) hypothesize that the distance logs travel may be a function of the channel’s debris roughness, DR, a dimensionless index incorporating ratios of piece length and diameter to channel width, depth, and sinuosity: where: Llog = Piece length wav = Mean channel width Rc = Radius of curvature db = Buoyant depth dav = Average channel depth a1, a2, a3 = Coefficients that vary according to relative impor- tance of each variable However, they note that while the terms in the model were significantly correlated with distance traveled for pieces, the results as indicated by the low R2 values were not particularly useful from a predictive standpoint. Finally, Lyn et al. (2003b) note that the delivery of debris to a given site “seems to occur in bursts, rather than contin- uously, even during a flow event of extended duration” and that a possible explanation for this is that “the debris is not generated in the vicinity of the site, and the bursts result from DR a L w a L R a d dav c b av (2.2)1 2 3α log log + + ⎛ ⎝⎜ ⎞ ⎠⎟ different travel times from different contributing areas.” They also conclude that “the transport of debris occurs rather intermittently with long periods of comparative inactivity punctuated by short periods of intense activity, generally on the rising limb of the hydrograph.” Table 2.2 presents a list of the mobility and transport mech- anisms of LWD for various geographic locations and channel positions as compiled by Lassettre and Harris (2001). 2.1.4 Debris Deposition, Accumulation, and Storage In intermediate channels (typically third and fourth order), drift is transported during major floods in large logjams con- sisting of large pieces and typically spanning the channel (Diehl and Bryan 1993). Thus, drift may be deposited in a variety of positions and accumulations throughout the chan- nel and may be sufficient to form blockages of the channel. In contrast, in most wide streams (typically fifth order and larger), very little drift is stored within the channel, instead drift accumulates most frequently and in the greatest amounts where the path of floating drift encounters obstructions and fixed objects (Diehl 1997). On meandering streams, debris is often deposited in wide, sinuous reaches where meander bend and alternate bar mor- phology promote frequent contact between the debris and the channel margins (Nakamura and Swanson 1994). In these types of channels, pieces tend to deposit on the outside of bends, at the heads of islands and bars, in flood plain forests, and in chutes and sloughs (McFadden and Stallion 1976, Sedell and Duval 1985, Malanson and Butler 1990, Chergui and Pattee 1991, Nakamura and Swanson 1994, Abbe and Montgomery 1996, 2003). Daniels and Rhoads (2003) con- ducted a field experiment to assess the influence of a debris obstruction located along the outer bank downstream of a bend apex on the three-dimensional (3-D) flow through a meander bend and identified pronounced influences on the flow structure through the bend. Debris jams form where pieces lodge against obstructions such as large boulders and other immobile features (Nakamura and Swanson 1994, Diehl 1997, Abbe and Montgomery 2003, Manners et al. 2007). D’Aoust and Millar (2000) noted dur- ing an investigation of the stability of single logs, single logs with rootwads, and multiple log structures that natural debris jams are initiated by large immobile logs that act as a stable key. When the rootwad remains intact on logs, deposited logs tend to be oriented with the rootwad upstream, likely as a result of the rootwad dragging as the log is transported down- stream. The rootwad can be an obstacle to other debris and can become the point of further debris accumulation, espe- cially where they become lodged on the apex of a bar (Abbe and Montgomery 1996, 2003). 15

Based on a study conducted in the Queets River basin in Washington, Abbe and Montgomery (2003) determined that individual pieces of woody debris in an accumulation or jam can be separated into key, racked, and loose members. They define three primary categories of woody debris jams (in-situ, transport, and combination jams) based on whether or not the constituent debris was fluvially transported. In-situ debris jams are composed of debris that has remained in place fol- lowing entry to the channel, although it may have rotated or the channel may have moved. Transport jams are composed of debris that has moved downstream by fluvial processes. Combination jams are composed of both in-situ key mem- 16 Location Channel Network Position Results References Western Oregon 0.10 to 0.30 channel gradients Ability of streams and rivers to move LWD depends on size of river and wood dimensions Large rivers transport most LWD, while small streams move only small debris short distances before deposition on channel obstructions Very large debris in small channels transported through debris torrents Torrents rare in >2nd to 3rd order streams, as steep gradients are required to move debris Swanson et al. 1976 Western Oregon 0.02 to 0.50 channel gradients Debris moves through flotation at high flows or through debris flows Debris flows originate in 1st and 2nd order channels (>50% gradient, <0.2 km2 drainage areas) 3rd to 5th order streams (4 km2 to 60 km2 drainage areas) wide enough to float large debris and debris jams at high flows Swanson and Lienkaemper 1978 Eastern Washington 0.015 channel gradient 11.5 m bankfull width LWD movement depended on length and diameter of wood Distance traveled by LWD inversely related to piece length Anchored pieces more stable in high flows Bilby 1984 Indiana, North Carolina, Oregon 0.001 to 0.40 channel gradients Debris torrents main transport mechanism in small, high gradient streams Flotation main transport mechanism in large, low gradient streams Keller and Swanson 1979 Western Oregon 0.03 to 0.37 channel gradients 3.5 m to 24 m bankfull widths 0.1 km2 to 61 km2 drainage areas 1st through 5th order streams Wood movement occurred in larger streams during monitoring period Distance moved depended on piece length in relation to bankfull width All pieces moving >10 m were shorter than bankfull width Lienkaemper and Swanson 1987 Western Oregon 0.03 to 0.21 channel gradients 7 m to 25 m bankfull widths 2nd through 5th order streams Debris flows redistributed LWD in steep, low order streams Floods redistributed LWD in medium to high order streams Nakamura and Swanson 1993 Western Oregon 0.019 to 0.028 channel gradients 9 m to 71 m channel widths 5th order streams Most transported pieces shorter than bankfull width 20% of untransported pieces longer than bankfull width LWD length to channel width useful measure of susceptibility to transport Nakamura and Swanson 1994 Wyoming 0.041 to 0.055 channel gradients 6.4 m to 7 m low flow channel widths LWD less stable in burned drainage due to increased runoff and peak flows, and decreased bank stability Young 1994 NA NA Uncongested, semi-congested, and congested wood transport based on ratio of log input (Qlog) to stream discharge (Qw) Proposed ability of channel to retain wood is function of debris roughness, which varies with channel and log dimensions Braudrick et al. 1997 Spain 0.08 to 0.1575 channel gradients 3 m to 14 m bankfull widths 0.8 km2 to 69 km2 drainage areas 1st through 3rd order streams Loading (m3/m2) decreased in downstream direction Log mobility increased in larger streams Elosegi et al. 1999 Source: Lassettre and Harris (2001) Table 2.2. Mobility and transport mechanisms of LWD.

bers and fluvially transported racked and loose members. Within these categories, Abbe and Montgomery (2003) fur- ther define 10 types of woody debris accumulation based on the mode of recruitment and the orientation of the members (Table 2.3). They found that key members usually account for a small fraction of individual logs in a jam, but they com- pose most of a jam’s volume, and that many jams have one to four key members and 10 to 100 times as many racked pieces. They also determined that the frequency of jams generally increases with drainage area up to about 116 mi2 (300 km2), above which frequency gradually decreases. Montgomery and Piégay (2003) indicate that tree type is important in the character, distribution, and type of debris accumulations. They note that widely spreading or multiple- stemmed hardwoods are more likely to form snags rather than accumulating as racked members in large log jams, whereas coniferous debris tends to produce cylindrical pieces that are more readily fluvially transported and routed downstream resulting in local concentrations or log jams along a river system. Once a debris jam forms, fine particles and small material such as brush, leaves, twigs, tree bark, small branches, grasses and weeds, sediment, and man-made materials can fill in the interstitial spaces within the jam (Parola et al. 2000). This reduction in the permeability of the debris jam matrix may increase the scour potential of the jam. Laursen and Toch (1956) suggest that permeability may be just as important as the size of the debris mass in evaluating scour. Observations indicate that the matrix of most debris jams is at least partially filled with fine material. Still, the study conducted by Manners et al. (2007) on the Indian River in New York indicates that debris jams are still highly porous structures, even with large volumes of soil and organic matter in them. 2.1.5 Debris Accumulation at Bridge Piers The structural components of a bridge can trap debris and may result in significant accumulations. Diehl (1997) notes that there are two classes of debris accumulation observed at bridges: single-pier accumulations and span blockages. Downs and Simon (2001) suggest that the overall “tree-trapping” potential of bridges is a function of tree height, trunk diame- ter, canopy or root bole diameter (whichever is greater), and pier span distance. Debris characteristics and, lacking detailed debris information, channel width upstream of the bridge site can be used to estimate the probable maximum width of debris accumulations and blocked bridge spans. Debris accu- mulations are generally deepest at the piers that support them and widest at the surface (Diehl 1997). Depth of Debris Accumulation at Piers Because flow depth controls the mobilization and trans- port of debris, it stands to reason that the depth of debris accumulations will be dependent, in large part, on flow depth at the time of the debris accumulation. Diehl (1997) suggests that, in the process of formation, single-pier accumulations often take on a form roughly resembling an inverted half-cone shape. He notes that drift accumulation begins at the surface, but may grow downward through accretion—“The accumu- lation can grow toward the river bed through accretion of logs on the underside of the raft as they are washed under it by the 17 Types Distinguishing Characteristics In-Situ (autochthonous) Key member has not moved downstream. Bank input Some or all of key member in channel. Log steps Key member forming step in channel bed. Combination In-situ key members with additional racked woody debris. Valley Jam width exceeds channel width and influences valley bottom Flow deflection Key members may rotate, jam deflects channel course Transport (allochthonous) Key members moved downstream Debris flow/flood Chaotic debris accumulation, key members uncommon or absent, catastrophically placed. Bench Key members along channel edge forming bench-like surface. Bar apex One or more distinct key members downstream of jam, often associated with development of bar and island. Meander Several key members buttressing large accumulation of racked debris upstream. Typically found along outside of meander bend. Raft Large stable accumulation of debris capable of plugging even large channels and causing significant backwater. Unstable Unstable accumulations composed of racked debris upon bar tops or pre-existing banks. Source: after Abbe and Montgomery (2003) Table 2.3. Basic wood debris accumulation typology.

plunging flow at the upstream edge . . .”—and that “drift accumulations are typically deepest at the piers that support them, and widest at the surface.” Although he states that under some circumstances debris accumulations can reach from the water surface to the riverbed, observations and photographs from a number of locations suggest that debris accumulations often extend to the river bed and in many cases the debris may become embedded in the channel bed during high flows and flow recession. Because most debris accumulations slide down the pier and settle on the bed during flow recession and do not re-float during subsequent floods, the debris mass may poten- tially form a base for further accumulations. Diehl (1997) indicates that an accumulation developing on top of a pre- existing accumulation can grow more rapidly through inter- action with the underlying debris and that uncleared debris has the potential to promote bar or island growth. Diehl (1997) also notes that no limit to the vertical extent of debris accumulations has been established other than the depth of flow. The maximum vertical extent of debris accumulation observed in his study was more than 40 ft (12 m). Diehl (1997) indicates that piers on the banks have less like- lihood of trapping debris than those in the channel. He found that bridges in Tennessee with one pier in the channel were several times more likely to have single-pier debris accumu- lations than bridges with two piers on the banks and none in the channel. Diehl (1997) and Lyn et al. (2003b) also note that large single-pier debris accumulations seem to be associated with the formation or presence of islands and mid-channel bars or the accumulation of exceptionally large debris. Width of Single-Pier Accumulations The channel width influences the length of debris delivered to the bridge site and consequently plays an important role in determining the accumulation potential and characteristics. The length of the longest debris pieces determines the maxi- mum width of the common types of debris accumulations. Debris accumulations resting against a single pier often con- tain one or more logs extending the full width of the accumu- lation perpendicular to flow. Large debris accumulations are held together by long logs that support the jam against lateral hydraulic forces. The full-width logs may not be visible because of the accumulation of many smaller logs and other debris, which may conceal the full-width logs. Single-pier accumu- lations often have a curved upstream appearance when viewed from above, and the center of the downstream side, which rests against the pier, contains the thicker part of the accumulation. Plots compiled by Diehl (1997) of debris width versus upstream channel width for sites in Tennessee and Indiana (Figure 2.4) suggest that most debris accumulations on sin- gle piers are usually less than 50 ft (15 m) wide. The data (see Figure 2.4) also suggests that debris accumulation widths decrease with channel widths greater than about 50 to 60 ft (15 to 18 m), probably as a result of the increased hydraulic forces associated with a larger channel, which would have a greater tendency to break up or deter formation of debris rafts and large accumulations. Width of Span Blockages Debris accumulations between piers typically occur where the length of transported logs exceeds the effective width of the openings between piers resulting in the potential for logs to come to rest against two piers. The effective span width is dependent on the skew of the bridge piers to the approaching flow. Span blockages from debris accumulations can also occur where debris rests against and spans debris accumu- lated on adjacent piers. According to Diehl (1997), most pier- to-pier blockages are the result of a single log bridging the span. Where spans are short enough to be easily spanned by debris, multiple-span or nearly complete bridge-wide block- ages are possible. 18 (a) Indiana (b) Tennessee Source: after Diehl (1997) Figure 2.4. Width of inferred single-pier drift accumulations at scour potential sites in (a) Indiana and (b) Tennessee. Dashed lines represent curves for maximum expected width of single-pier debris accumulation.

In addition, accumulations between the bank or abutment and an adjacent pier are similar in structure to accumulations between piers (Diehl 1997). Often these accumulations extend diagonally upstream from the pier to the bank. Bank-to-pier accumulations are the result of one end of a large log becom- ing lodged against a tree or other fixed obstruction on the bank and the other end rotating into the pier. In all cases of pier-to-pier–spanning debris accumulations that were examined for areas outside the Pacific Northwest by Diehl (1997), the effective width of the blocked openings was short enough to be spanned by a single log. Based on the data collected by Diehl (1997) and shown in Figure 2.5, it would appear that span blockages decrease significantly for spans greater than about 80 ft (24 m) for areas outside of the Pacific Northwest. Contrastingly, Diehl found that current design practice favors spans in the range of 70 to 100 ft (20 to 30 m). Design/Key Log Length Site investigations conducted by Diehl (1997) indicate that log length, which is closely related to channel width, is the most important factor influencing the width of debris accu- mulations at bridges. Key logs are those logs that extend the full width of a debris accumulation perpendicular to the approaching flow. Because single-pier drift accumulations are based on logs extending the full width of the accumulation and spans are often blocked by logs extending from pier to pier, Diehl (1997) suggests that the maximum width of these types of accumulation is about equal to the maximum length of sturdy logs delivered to bridges. Diehl recommends that the log length that should be used in the design of bridges be inferred from the width of the largest single-pier accumulations and the longest blocked spans when estimating the potential for debris accumulations. Diehl notes that the design log length does not represent the absolute max- imum length of debris pieces, but instead represents a length above which logs are insufficiently abundant or insufficiently strong throughout their full length to produce drift accumu- lations equal to their length. This length is also described as the sturdy-log length. Diehl (1997) states that “the design log length, which is also the minimum effective span length for trapping potential, is intentionally set at the highest level justified by the confirmed pier-to-pier accumulations” and defines the design log length at a given site by the smallest of the following: • Width of the channel upstream of the site • Maximum length of sturdy logs • In much of the United States, 30 ft (9 m) plus one-quarter of the width of the channel upstream from the site For narrow streams, the minimum width immediately upstream of a bridge can be used as an estimate of the length of the longest logs that can be delivered to the bridge. Although the height of mature trees on the banks of wider streams deter- mines the maximum length of the logs that may be delivered as debris to a bridge site, tree height is not identical to the max- imum sturdy-log length. Diehl (1997) found that the max- imum sturdy-log length seems to reach about 80 ft (24 m) throughout much of the United States, and especially much of the eastern United States, and may be as long as about 150 ft (45 m) in parts of northern California and the Pacific Northwest. Diehl (1997) suggests that for those areas where the maxi- mum sturdy-log length is 80 ft (24 m), it can be assumed that the design log length is less than either the upstream channel width or the maximum sturdy-log length over an intermedi- ate range of channel width from 40 to 200 ft (12 m to 60 m). Therefore, he recommends that the design log length over this range of channel width is 30 ft (9 m) plus one-quarter of the channel width (Figure 2.6). Where the average sturdy-log length exceeds 80 ft (24 m), as in the Pacific Northwest, the design log length is equal to the lesser of either the upstream channel width or the regional maximum sturdy-log length as shown in Figure 2.6. 2.1.6 Modeling Debris-Induced Hydrodynamic Forces and Scour Parola et al. (2000) report that a common cause of dam- age to bridges subjected to forces caused by transported debris was due to the flow constriction resulting from debris accumulation. Additionally, while Parola et al. indicate that debris accumulations were typically considered to be float- ing rafts, observations identified conditions where debris accumulations developed through the entire flow field. 19 Design Log Length Source: Diehl (1997) Figure 2.5. Effective width of debris-blocked spans outside of the Pacific Northwest.

Examining the drag forces identified by Parola et al. (1998b), one can see the potential for both static and dynamic fluctu- ations in the hydraulic conditions around a structure as debris accumulates. Young (1991), Cherry and Beschta (1989), Melville and Dongol (1992), Parola et al. (2000), Wallerstein et al. (2001), Lyn et al. (2003b), and Bocchiola et al. (2008) report physical model results associated with transported debris. With the exception of Parola et al. (2000) and Bocchiola et al. (2008), studies were conducted in flumes 2 ft or less in width. While numerous data can be collected from a physical model study conducted at a small scale, applying small-scale model results to prototype applications often results in overly conservative designs. Wallerstein et al. (2002) at the University of Nottingham, U.K., developed a dynamic model for “constriction” scour caused by large woody debris. This one-dimensional (1-D) model was developed for predicting the rate and total depth of scour associated with a channel constriction. Wallerstein applied it to determine scour rates and depths associated with LWD jams and to assess debris impacts on local channel morphology. Several studies (Prasuhn 1981, Cherry and Beschta 1989, Raudkivi1990,AbbeandMontgomery 1996, Parola et al. 1998b, Wallerstein and Thorne 1995 and 2004, Wallerstein et al. 2001, Wallerstein 2002, Huizinga and Rydlund 2004, Manners et al. 2007, Bocchiola et al. 2008) address scour related to flow over, around, and under debris in channels. For example, studies by Cherry and Beschta (1989) and Buffington et al. (2002) suggest that scour hole formation occurs in the conver- gence zone off the tip of a debris jam while Manners et al. (2007) state that “only after a jam has accumulated enough material to substantially decrease its porosity will flow be sufficiently concentrated at the tip, thereby creating the pre- dicted scour hole.” They also note that localized areas of scour can develop downstream of the jam with increasing jam porosity because of flow acceleration through one of multiple holes under and/or through the debris jam. However, many of the studies indicate that the effects of constriction and obstruction can be addressed by adapting contraction and local scour equations, depending on the characteristics of the debris jam. Other studies—including Laursen and Toch (1956), Dongol (1989), Melville and Dongol (1992), Diehl (1997), Mueller and Parola (1998), Wallerstein and Thorne (1998), and TAC (2004)—provide information on scour caused by debris accumulations at bridges. Diehl (1997) provides information on Australian and New Zealand design practices for establish- ing the geometry of debris masses at piers. Laursen and Toch (1956) state that debris effectively enlarges the pier obstruc- tion, but that evaluating debris scour is very difficult because the permeability, position, and size are all important charac- teristics of the debris mass. Melville and Dongol (1992) indi- cate that a debris mass located at the water surface on a single circular pier can be evaluated using an effective pier diameter in a local pier scour equation as illustrated in Figure 2.7. Using three different idealized shapes [cylindrical, conical, and elliptical (in plan)] enveloping the pier, Dongol (1989) deter- mined that a cylindrical shape provided the maximum clear- water scour for uniform non-rippling bed material. Diehl (1997) and TAC (2004) also indicate that debris rafts increase velocity and bed shear, which increases general (pressure) bed scour. For local pier scour, the concept of effective pier diameter is included in the HEC-18 (Richardson and Davis 2001) complex pier scour approach. The complex pier scour approach includes the pier stem, pile cap, and pile group scour amounts as inde- pendent scour components and is illustrated in Figure 2.8 (Jones and Sheppard 2000). This complex pier approach is a more advanced treatment of the effective pier diameter approach suggested by Melville and Dongol (1992) because it addresses the pile cap separately from the pile group, whereas 20 0 15 30 45 60 75 90 UPSTREAM CHANNEL WIDTH (meters) 0 15 30 45 60 D ES IG N LO G L EN G TH (m ete rs) EASTERN U.S. OLYMPIC PENINSULA Design Log Length Equal To Maximum Log Length Design Log Length Equal To Maximum Log Length Source: after Diehl (1997) Design Log Le ngth = 9 m + (C W/4) De sig n L og Le ng th Eq ua l T o C ha nn el Wi dth Figure 2.6. Design log length and upstream channel width (CW) for the eastern United States and the Olympic Peninsula. Source: Melville and Dongol (1992) Figure 2.7. Debris raft equivalent pier diameter.

Melville and Dongol combine the debris mass with the pile to arrive at an effective diameter. For general scour, HEC-18 (Richardson and Davis 2001) includes live-bed and clear-water contraction scour equations as well as a pressure scour equation for submerged bridge decks. These equations may be suitable to the debris raft condition where debris extends across bridge spans or if the debris raft extends a significant distance from an individual pier. The contraction scour equations or their application may need to be modified for debris raft conditions. A modification to an equation may involve an adjusted coefficient or exponent while maintaining the original form of the equation. Appendix D of HEC-18 (Richardson and Davis 2001) also includes an interim procedure for calculating scour for debris accumulations at the leading edge of a pier combined with a flow angle of attack. The method does not include an effec- tive width adjustment and, therefore, appears to assume that the debris accumulates over the full depth of flow. It is important, however, to recognize that there may be combined effects of both debris accumulation and angle of attack, both of which may not be fully addressed in existing analysis and design procedures. Another aspect of debris accumulations is the topic of NCHRP Report 445: Debris Forces on Highway Bridges (Parola et al. 2000). The report provides guidance on computing the additional force acting on the bridge structure. The drag force of the flow impacting the debris obstruction produces this force. Although evaluating debris forces on bridges is not a topic of this research project, this force is an additional com- ponent of the loss (or force) terms that should be included in bridge hydraulic computations. Because debris alters the bridge hydraulics (through the added drag force) and the hydraulic conditions determine the scour potential, it is impor- tant that hydraulic analyses and scour analyses are consistent and compatible. From the Parola et al. (2000) study, it is apparent that the computation of drag force (for large accumulations) is not sim- ply a matter of identifying a drag coefficient and an obstructed area, but also requires use of a constricted velocity rather than an approach velocity and must include the upstream and downstream hydrostatic pressure forces. Therefore, standard one- and two-dimensional (2-D) modeling approaches may not accurately represent the hydraulic effects of debris. One- and two-dimensional models may be used to determine whether the observed (experimental) hydraulic conditions can be replicated by the models. For example, several different bridge modeling approaches are available [WSPRO, energy, momentum, Yarnell, pressure/ weir in HEC-RAS (Brunner 2008)]. One method may be more applicable than others to debris simulation. In the Finite- ElementSurfaceWaterModelingSystem(FESWMS) (Froehlich 2002, rev. 2003), pier drag and submerged deck routines may be applicable to debris simulation (see Section 3.8). 2.1.7 Managing Debris Accumulations at Bridges Management of debris problems at bridges has generally focused on debris removal at bridges and in upstream channels (Brice et al. 1978, Lagasse et al. 1991, 2001), bank clearing, channel modifications (Gippel 1989, Gippel et al. 1992), and more recently, the installation of debris deflectors, guides, and traps. However, it is now recognized that debris removal can have a detrimental effect on stream ecology (Gippel 1989, Gippel et al. 1992) and, combined with bank clearing, can cause channel instability and produce a wider and shallower channel with reduced conveyance (Thorne 1990). Nonetheless, prompt and complete debris removal is greatly dependent on the acces- sibility of the bridge piers and substructure. The most widely used method for dealing with debris accu- mulations is through removal. However, where debris removal is difficult or expensive, debris accumulations may remain in place and the potential for additional accumulations grows with the number of floods that occur. Piers with accumulated debris may not shed additional debris as effectively and new accumulations can develop more rapidly over existing accu- mulations as a result of the interaction with the underlying debris. Additionally, debris accumulations that remain in place for long periods of time can induce the formation of bars or islands. 21 T h1 h2 h3 y1 y2 ys = ys pier + ys pc + ys pg ++= h0 pier stem pile cap pile group f f FLOW y3 Source: Jones and Sheppard (2000) Figure 2.8. Definition sketch for scour components for a complex pier.

Piers The placement, type, and skew of bridge piers have a sig- nificant influence on the extent of debris accumulation at a pier. Piers placed in the path of drift movement, such as in the thalweg, will have a higher trapping potential. In narrow channels, piers placed in the center of the channel may have a greater risk of initiating a channel-wide debris accumula- tion. Piers placed in close proximity to the bank will have a greater trapping potential if the span between the bank and pier is less than the maximum sturdy-log length. Thus, piers placed on the banks are less likely to trap debris than those placed in the channel (Diehl 1997). Multiple columns, pile bents, or piers with clusters or mul- tiple rows of piles, which can result from the intentional place- ment of a footing above the channel bed or water surface or exposure due to erosion, can act as a debris trap unless exactly aligned with flow. Even then the gaps between columns or piles can be spanned by debris and debris can become entan- gled in the columns and piles, thus making removal extremely difficult and increasing the potential for damage to the bridge during clearing operations. Diehl (1997) indicates that skewed pile bents in the water were about twice as likely to have debris accumulations as those with unskewed bents in the water. Pier shape may also influence debris trapping potential. Round-nosed piers may be more favorable than square-nosed piers, which provide a flat surface against which debris may become lodged. Although it seems intuitive that round-nosed piers should be able to shed debris more easily than square- nosed piers, Diehl (1997) found that bridges in Tennessee with round-nosed piers were no less likely to trap debris than those with square-nosed piers. Diehl (1997) indicates that current design practice favors the use of single-column piers with rounded noses. The use of wall piers that extend upstream to the edge of the parapet are more easily cleaned of debris than other types of piers. Although hammerhead piers are an alternative to multiple-column piers, removal of debris from the top of the accumulation is more dif- ficult because the pier nose is well under the bridge. Spans The potential for span blockages also depends on type, place- ment, and skew of bridge piers as well as on channel curvature and other channel shape features. In general, the potential for blockage of pier-to-pier, pier-to-bank, or pier-to-abutment spans is low where the effective span width is greater than the design log length. The use of long spans provides the bridge designer the opportunity to place fewer piers in the channel and, conse- quently, in the path of debris or in the middle of the channel where access may be difficult. On narrow channels, the main span should bridge the entire channel in order to effectively reduce the potential for span blockage. The placement of piers at or near the base of the banks in narrow channels can create less potential for span blockage than placement in the channel (Diehl 1997). Since transported debris tends to follow the channel thal- weg, the placement of piers and spans should account for the thalweg location because debris is more likely to move along the outer bank in channel bends and along the middle of the channel in relatively straight reaches. Where the bridge crosses a bend or where the reach immediately upstream of the bridge is a long curve, the use of a span that extends from the outer (concave) bank and crosses much of the channel will have a lower trapping potential. Where a bridge is located down- stream of a long straight reach, the use of a main span that bridges the center of the channel will have a lower trapping potential. However, the movement of debris should be eval- uated under flood flow conditions, especially for conditions greater than bank full flow where the debris path may be short- circuited or deviate from the expected path as compared to the path during less than bank full flow. Where the bridge is skewed to approaching flow, the effec- tive width of horizontal gaps (span and pier-to-bank width) is reduced. The effective span or pier-to-bank width is the dis- tance between lines parallel to the approaching flow that pass through the nose of each pier (Figure 2.9). Diehl (1997) indi- cates that the spans can be bridged by debris accumulations where the size of the debris exceeds the effective horizontal gap width. However, if key logs come in contact with a pier, they can rotate before coming in contact with the adjacent pier and, therefore, the pier-to-pier width (i.e., span width) should actually be equal to or greater than the design log length, not the effective horizontal gap width. The effective pier-to-bank width should also be greater than the design log length since logs will more likely impact on the bank first and then rotate and intersect the pier at an angle equal to or greater than perpendicular to the approaching flow. Conversely, if a log impacts the pier first, it is likely that it will not intersect the bank when and if it rotates, and will proba- bly be deflected downstream. If the log rotates inward toward the bank before passing downstream, it may become wedged between the pier and bank (depending on the bank geometry) under the bridge. Countermeasures Other management tools currently being used to control debris accumulation on bridges include river training or chan- nel stabilization at the bridge site and the placement of some form of debris deflector, guide, or collector in the channel immediately upstream of the bridge. Debris deflectors, guides, or collectors have had mixed results because their effectiveness 22

is dependent on the amount and frequency of debris delivered to the bridge site as well as local site conditions and structure configuration (e.g., Lyn et al. 2003b). Structural measures need to be robust enough to withstand impact and static forces and scour if they become blocked (TAC 2004). Various methods of river training and channel stabiliza- tion at a bridge site have been well documented (Lagasse et al. 2001, Richardson et al. 2001). Although river training and channel stabilization can control the geometry of the channel at the bridge site with regard to the alignment of the structural features to the principal transport path of debris, they have little effect on the amount of debris delivered to the bridge components themselves. A variety of methods for collecting debris or guiding debris through bridge openings has also been identified by several sources (Reihsen and Harrison 1971, Brice et al. 1978, Chang and Shen 1979, Perham 1987 and 1988, Saunders and Oppenheimer 1993, Lagasse et al. 2001, Richardson et al. 2001, Bradley et al. 2005). Three general categories of debris con- trol structures are used on streams: (1) debris deflectors, (2) debris catchers or racks, and (3) debris fins (Reihsen and Harrison 1971, TAC 2004). The most prevalent debris con- trol structures used today are debris deflectors and debris fins. Debris deflectors are generally used where debris loading is moderate to heavy. They are placed immediately upstream of a bridge and are used to deflect the major portion of debris away from bridge piers or assist in guiding debris into and through wide bridge openings. They may consist of a single piling or a series of three pil- ings placed such that the group forms a v-shape in planview with the apex pointing upstream. These types of deflectors have had mixed success because debris can become lodged in the deflector and may accumulate on the deflector itself. Debris that accumulates on the deflector may become dis- lodged and travel downstream where it can impact bridge piers as a massive cohesive unit, which can have catastrophic consequences. However, permanently installed debris deflec- tors can be ineffective in reaches where flow direction changes as a result of either changing flood stages (i.e., flow path changes with stage) or because of changes in channel planform (i.e., changes in flow alignment) over time. More recently, lunate- shaped hydrofoil deflectors and rotating-drum deflectors have also been developed for use on bridges. The lunate-shaped hydrofoil deflector is designed to gen- erate counter-rotating streamwise vortices in its wake such that the vortices migrate to the surface of the water ahead of the pier and deflect the debris around the pier (Saunders and Oppenheimer 1993). However, the hydrofoil is also fixed in the river and its success is largely stage and flow direction dependent, and it can be damaged by debris that becomes lodged underneath it during lower flows. It is not known if this type of deflector has been installed or used at any bridge locations. Rotating drum deflectors are mounted upstream near the front of a pier and rotate such that debris that impacts on the deflector is turned away from the pier by the current and the force of the debris (Collier 2005). These types of deflec- tors can be mounted such that they can adjust with river stage and are generally not affected by changes in flow direction except at extreme angles. Because they are mounted near the leading edge of a pier, deflectors can be easily changed if they become damaged. 23 Effective pier-to-bank width Stream Channel Approaching Flow Pier-to-pier (span) width Pier Bridge Bankline Source: after Diehl (1997) Effe ctiv e wi dth of h oriz onta l gap bet we en pier s Figure 2.9. Definition sketch of the effective width of horizontal gaps.

Debris racks or catchers are placed across the channel and are used to collect debris before it reaches a bridge (Perham 1987, 1988). They may be vertical or inclined and may be square or skewed to flow. Inclined racks allow for debris to ride up onto the rack and skewed racks allow debris to be partially deflected toward the bank. However, debris racks and catch- ers can become clogged or blocked. Completely blocked racks or catchers can produce significant backwater upstream, poten- tially severe scour downstream of the structure or where flow reenters the structure, contraction and local scour if the struc- ture is immediately upstream of the bridge, and possible flank- ing or by-passing of the structure due to lateral erosion (Diehl 1997, TAC 2004). In addition, racks and collectors may collect significantly more debris than would be trapped by a bridge, so the cost of debris removal may be greater where these struc- tures are used (Diehl 1997). Cribs or timber sheathing is also used on bridge piers to deflect debris away from piers that consist of open pile bents. However, cribs with large mesh openings can trap debris and create problems and wooden cribs or sheathing can be severely damaged by large debris and ice (Chang and Shen 1979). Debris fins are walls or rows of pilings placed upstream of bridge piers or may be upstream extensions of the piers them- selves (Chang and Shen 1979, TAC 2004). The fins are aligned to flow and may be inclined to allow deflection of the debris through the bridge opening. However, as with piers, the effec- tiveness of fins is dependent on their alignment to flow and debris path, and they can contribute to span blockages if the span between fins is less than the design log length. Finally, a combination of channel dredging and bank revet- ment may be an option where sedimentation and stream aggra- dation play a significant role in debris accumulation problems (Lyn et al. 2003b). However, dredging by itself would not pro- vide a long-term solution under these circumstances because the underlying causes of sedimentation/aggradation would probably not be addressed. 2.2 Debris Photographic Archive A photographic archive was compiled to assess typical debris geometry relationships. Photographs of debris at bridges were acquired from a number of contributors. The photographs in the archive provided the primary source that was used for evaluating debris accumulation characteristics and debris geometry from a wide range of sites located throughout the United States. The primary contributor of debris photographs was Collier (2005), who provided almost 50% of the photographs. Numerous photographs were provided by state DOT person- nel in response to the Task 2 survey. The remaining photo- graphs were acquired from in-house sources, Internet sites, and from referenced publications. A total of 1,079 photographs at 142 sites in 31 states were acquired from the various sources, but not all the photographs are specifically of debris accumu- lations at bridges. The photographic archive is provided on the TRB website (see Appendix A). Table 2.4 presents a list of the sources of the debris photographs, site locations and the num- ber of photographs available for each site (see also Figure 2.14 for a map of the photograph locations in each of the five geographic regions). Directions for navigating the debris photographic archive are included in Appendix A. The photographs are first orga- nized by state and finally by the river or stream. Within each subdirectory, a Microsoft® PowerPoint file has been created containing the individual photographs for the given site. 2.3 Regional Analysis of Debris One might infer from previous studies, experience with debris problems nationally, the geomorphic characteristics of rivers in different Physiographic Regions, and the distinctly different characteristics of woody vegetation and river bank erosion processes that there should be some regional bias in debris characteristics and in debris impact on bridges. As a starting point, these regional characteristics were investigated in the Task 1 literature search and in the Task 2 site reconnais- sance and survey. An attempt was made to identify common (or typical) debris characteristics that might be distinguishable by major Physiographic Region or Ecoregion. Several studies have suggested the potential for regionalized debris generation characteristics. Chang and Shen (1979) developed a depiction of a national distribution of debris problems from severe to moderate to minor or no problem as shown in Figure 2.10. The distribution was based on a state-specific survey of debris problems at highway bridges nationwide and indicated that the Pacific Northwest and the upper and lower Mississippi River Valley experience the most severe debris problems. Diehl (1997) mapped debris sites reported by DOTs, distribution of debris field-study sites, and debris sites referenced in publications or personal communications. Diehl suggests that watersheds of high or low debris generation can be identified based on watershed characteristics such as proximity of vegetation to the stream, rate of bank erosion, and/or channel instability resulting from natural properties, climate change, fire, or human modification. Prediction of debris accumulation might potentially be regionalized by initially dividing the land area into subdivi- sions based on debris-contributing characteristics such as vegetation type and geomorphic factors. Two commonly recognized systems of geomorphic and ecological subdivi- sions or classifications are the Physiographic Regions classi- fication by Fenneman (1917) and Ecoregions classification by Bailey (1983). 24

25 Geographic Region State Stream Source Number of Photos East North Deep River Debris Free, Inc. (Mike Collier) 4 Carolina Deep River NCDOT (web site photos) 5 Tennessee Coles Creek Web Site Photos 4 Appomattox River Debris Free, Inc. (Mike Collier) 4 Virginia Dan River Debris Free, Inc. (Mike Collier) 2 James River Debris Free, Inc. (Mike Collier) 2 Nottoway River Debris Free, Inc. (Mike Collier) 4 West Mud River Debris Free, Inc. (Mike Collier) 2 Virginia Tug River Debris Free, Inc. (Mike Collier) 3 Midwest East Skokie Ditch Debris Free, Inc. (Mike Collier) 3 Iroquois River Debris Free, Inc. (Mike Collier) 4 Illinois Mackinaw River Debris Free, Inc. (Mike Collier) 3 Mississippi River Debris Free, Inc. (Mike Collier) 3 Unknown Debris Free, Inc. (Mike Collier) 6 Indian Creek Ayres Associates Inc. 1 Eel River Debris Free, Inc. (Mike Collier) 8 Eel River Debris Free, Inc. (Mike Collier) 4 Indiana Vermillion River Debris Free, Inc. (Mike Collier) 11 Wabash River Debris Free, Inc. (Mike Collier) 3 White River Timothy Diehl (1997) 1 White River USGS (Robinson 2003) 6 Cedar Creek Tim Dunlay, Iowa DOT (Survey) 14 Iowa East Nishnabotna River Tim Dunlay, Iowa DOT (Survey) 11 West Nodaway River Tim Dunlay, Iowa DOT (Survey) 3 Elk River Ayres Associates Inc. 32 Neosho River Ayres Associates Inc. 38 Neosho River Ayres Associates Inc. 28 Verdigris River Ayres Associates Inc. 14 Chikaskia River Brad Rognlie, P.E. KDOT (Survey) 1 Elk River Brad Rognlie, P.E. KDOT (Survey) 25 Kansas Elm Creek Brad Rognlie, P.E. KDOT (Survey) 7 Neosho River Brad Rognlie, P.E. KDOT (Survey) 19 Smoky Hill River Brad Rognlie, P.E. KDOT (Survey) 37 Unknown Brad Rognlie, P.E. KDOT (Survey) 6 Verdigris River Brad Rognlie, P.E. KDOT (Survey) 11 Republican River Debris Free, Inc. (Mike Collier) 6 Smoky Hill River Debris Free, Inc. (Mike Collier) 34 Kentucky Unknown Debris Free, Inc. (Mike Collier) 2 Minnesota River Debris Free, Inc. (Mike Collier) 9 S. Br. Wild Rice River Duane Hill, MnDOT (Survey) 2 Minnesota Minnesota River Larry Cooper, MnDOT (Survey) 1 Minnesota River Larry Cooper, MnDOT (Survey) 1 Mississippi River Larry Cooper, MnDOT (Survey) 2 Black River MnDOT Dist. 1 (web site photos) 2 Missouri Florida Creek Ayres Associates Inc. 1 Unknown Debris Free, Inc. (Mike Collier) 1 Chariton River John Holmes, MoDOT (Survey) 9 Big Creek Ken Foster, MoDOT (Survey) 2 Grand River Ken Foster, MoDOT (Survey) 1 W. Fk. Grand River Ken Foster, MoDOT (Survey) 8 Nebraska South Platte River Ayres Associates Inc. 1 Ohio Great Miami River Brandon Collett, ODOT (Survey) 9 Unknown Debris Free, Inc. (Mike Collier) 5 Washita River Debris Free, Inc. (Mike Collier) 3 Oklahoma North Canadian River Debris Free, Inc. (Mike Collier) 12 Harpeth River Debris Free, Inc. (Mike Collier) 8 Tennessee Harpeth River Jon Zirkle, TDOT (Survey) 11 Harpeth River Timothy Diehl (1997) 2 Texas Clear Fk. Brazos River Debris Free, Inc. (Mike Collier) 6 Clear Fk. Brazos River Jerry Conner, TxDOT (Survey) 6 Wisconsin Bad River Allan Bjorklund, WisDOT (Survey) 1 Pikes Creek Allan Bjorklund, WisDOT (Survey) 1 Table 2.4. Inventory of photographs of debris at bridge piers. (continued on next page)

26 Geographic Region State Stream Source Number of Photos Pacific Bear River Ayres Associates Inc. 2 Coast Harris Creek Ayres Associates Inc. 1 California Malibu Laguna Ayres Associates Inc. 1 Sacramento River Ayres Associates Inc. 6 Sacramento River Ayres Associates Inc. 2 Stony Creek Ayres Associates Inc. 1 California Arroyo Grande Creek Kevin Flora, Caltrans (Survey) 11 (Central & Hopper Creek Kevin Flora, Caltrans (Survey) 3 South Coast) Salsipuedes Creek Kevin Flora, Caltrans (Survey) 5 California Sacramento River Kevin Flora, Caltrans (Survey) 2 (Central Stony Creek Kevin Flora, Caltrans (Survey) 6 Valley) Thomes Creek Kevin Flora, Caltrans (Survey) 3 California (Mountain) North Fork Deer Creek Kevin Flora, Caltrans (Survey) 4 California Mad River Kevin Flora, Caltrans (Survey) 4 (North Trinity River Kevin Flora, Caltrans (Survey) 3 Coast) Yager Creek Kevin Flora, Caltrans (Survey) 5 California Butte Creek Debris Free, Inc. (Mike Collier) 9 (North) Navarro River Debris Free, Inc. (Mike Collier) 13 Callegas Creek Debris Free, Inc. (Mike Collier) 18 California San Antonio Creek Debris Free, Inc. (Mike Collier) 9 (South) Santa Clara River Debris Free, Inc. (Mike Collier) 26 Ventura River Debris Free, Inc. (Mike Collier) 29 Shitike Creek Ayres Associates Inc. 1 Oregon Calapooia River Debris Free, Inc. (Mike Collier) 1 North Santiam River Debris Free, Inc. (Mike Collier) 3 Cowlitz River Debris Free, Inc. (Mike Collier) 7 North Fork Skagit River Debris Free, Inc. (Mike Collier) 6 Washington North Fork Skagit River Debris Free, Inc. (Mike Collier) 2 Skykomish River Debris Free, Inc. (Mike Collier) 4 South Fork Skagit River Debris Free, Inc. (Mike Collier) 1 Queets River Timothy Diehl (1997) 1 South Arkansas St. Francis River Debris Free, Inc. (Mike Collier) 34 Florida Escambia River Debris Free, Inc. (Mike Collier) 6 Bayou Boeuf Ayres Associates Inc. 1 Louisiana Amite River Debris Free, Inc. (Mike Collier) 10 Red River Raft Web Site Photos 2 Abiaca Creek Ayres Associates Inc. 1 Mississippi Jack Creek Ayres Associates Inc. 2 Sykes Creek Ayres Associates Inc. 1 Coles Creek MDOT Bridge Design (Survey) 1 South Carolina Little River Debris Free, Inc. (Mike Collier) 40 Jackson District Var. Debris Free, Inc. (Mike Collier) 121 Tennessee Wolf River Debris Free, Inc. (Mike Collier) 3 Wolf River Jon Zirkle, TDOT (Survey) 3 Brushy Creek Blaine Laywell, TxDOT (Survey) 6 Little River Blaine Laywell, TxDOT (Survey) 7 Guadalupe River Carl O'Neil, TxDOT (Survey) 7 Rocky Creek H.C. Schroeder, TxDOT (Survey) 6 Pedernales River Jeff Howell, TxDOT (Survey) 6 Texas San Marcos River Jeff Howell, TxDOT (Survey) 1 Cocklebur Creek Jon Kilgore, TxDOT (Survey) 7 San Antonio River Jon Kilgore, TxDOT (Survey) 13 Elm Fork of Trinity River Tim Hertel, TxDOT (Survey) 2 Red River Tim Hertel, TxDOT (Survey) 12 Brazos River Timothy Diehl (1997) 1 Unknown Timothy Diehl (1997) 5 Table 2.4. (Continued).

27 Geographic Region State Stream Source Number of Photos West Arizona New River Ayres Associates Inc. 2 Colorado Bijou Creek Ayres Associates Inc. 4 Idaho Pack River Ayres Associates Inc. 2 Teton River Ayres Associates Inc. 1 Boulder River Russell Brewer, PE MtDOT (Survey) 12 Montana Jefferson River Russell Brewer, PE MtDOT (Survey) 14 Regis River Russell Brewer, PE MtDOT (Survey) 3 Carson River Ayres Associates Inc. 2 Humboldt River Ayres Associates Inc. 1 Nevada Carson River Chris Miller, NDOT (Survey) 15 Carson River Chris Miller, NDOT (Survey) 6 Carson River Debris Free, Inc. (Mike Collier) 6 New Mexico Rio Grande Ayres Associates Inc. 1 Unknown Unknown Ayres Associates Inc. 3 Colorado River Debris Free, Inc. (Mike Collier) 4 San Rafael River Debris Free, Inc. (Mike Collier) 7 Santa Clara River Tim Ularich, UDOT (Survey) 6 Virgin River Tim Ularich, UDOT (Survey) 5 Utah Unknown Various Web Sites (incl. KSL Channel 5 TV, City of St. George) 11 Beaver Dam Creek Various Web Sites (incl. KSL Channel 5 TV, City of St. George) 4 Santa Clara River Various Web Sites (incl. KSL Channel 5 TV, City of St. George) 7 Virgin River Various Web Sites (incl. KSL Channel 5 TV, City of St. George) 1 Table 2.4. (Continued). Source: Chang and Shen (1979) Figure 2.10. Debris problem distribution.

Fenneman (1917) proposed the concept of physiographic subdivisions of the United States based on regional geomor- phologic characteristics in the early 20th century. Leighty (2001) defines a physiographic unit as “an area of the Earth’s surface within which the major topographic features have a single geomorphic history, a definite general structure, certain physical characteristics, and a predictable general pattern of lower order landforms.” Fenneman created a three-tiered classification system consisting of divisions, provinces, and sections derived from terrain texture, rock type, and geologic structure and history (Barton et al. 2003). In the United States, there are 25 physiographic providences and 82 physiographic sections. Figure 2.11 presents a map of the Physiographic Regions of the United States (Barton et al. 2003). Ecoregions subdivide the earth’s surface into identifiable areas based on macroscale patterns of ecosystems (Bailey 2005) and the environmental factors that most likely acted in creating variation in ecosystems (Bailey 1983). The intent of the ecologically based Ecoregions classification system is to provide a foundation for estimating ecosystem productivity and potential response to varying management practices (Bailey 1983). Ecoregions are based on a three-tiered hierarchy system. The two broadest tiers, domains and divisions (within domains), are delineated principally on the large ecological climate zones identified by Köppen (1931) and modified by Trewartha (1968). Climate is emphasized because of the prevailing effect it has on composition and productivity of an ecosystem (Bailey 1995), where moisture and temperature characteristics drive soil formation and surface topography and limit plant growth. The third tier, provinces, allows an Ecoregion to be described in terms of its dominant physical and biological characteris- tics such as land-surface form, climate, vegetation, soils, and fauna. Bailey’s Ecoregion map contains 34 provinces and 29 mountain provinces, within 24 divisions and 4 domains. Figure 2.12 presents a map of the Ecoregions of the United States (Bailey 1995). Ecoregions and Physiographic Regions define areas where one might expect to find similar types of vegetation and geo- morphic conditions, thereby possibly providing a method of predicting regional debris generation relationships. To more closely examine the relationship among the two classification systems and general debris characteristics, the site locations described in the photographic archive and presented in Table 2.5 are delineated on the Physiographic Regions and Ecoregions maps shown in Figures 2.11 and 2.12. An examination of the debris characteristics and accumu- lation geometry from the various debris site photographs relative to their locations on the Physiographic Region and Ecoregion maps reveals that there is no identifiable or well- defined relationship with regard to the individual Physio- graphic Regions or Ecoregions. Instead, the typical debris accumulation geometries that have been identified appear to be common throughout the United States, and any river or stream with a riparian corridor along its bank is susceptible to debris problems. For example, Figure 2.13 shows a com- mon debris accumulation geometry found at five different sites from various parts of the country that are in distinctly different Physiographic Regions and Ecoregions. With regard to vegetation types, only general relationships exist. In most of the eastern half of the United States, large woody debris delivered to streams and rivers is primarily from riparian forests or corridors along those streams and rivers. In the arid southwest and west, vegetation can be limited to small trees and shrub-like vegetation, and the debris accumulation sizes and geometry in these areas often reflect this type of vegeta- tion. In the Pacific Northwest, large conifers are the dominant tree species and are the controlling factor in the size of the debris accumulations and the debris geometry in this region. However, there does appear to be a simple geographic rela- tionship with regard to the distribution of debris problems as suggested by the site locations of the debris photographs in the photographic archive. The distribution of the photograph sites can be grouped into five general geographic regions of the United States: East, South, Midwest, West, and Pacific. The boundaries for these general geographic regions are roughly similar to some of the major physiographic bound- aries with some deviation. Figure 2.14 shows the debris sites from the photographic archive in relation to the five general geographic boundaries. As seen in Figure 2.14, the majority of the photograph sites are distributed within three of the five geographic regions: the Midwest, the Texas coast and lower Mississippi Valley por- tions of the South, and the Pacific. The large number of sites located in the Midwest is closely related to distribution of the major tributary drainages of the Mississippi River. The distri- bution of the sites in the South appears to be tied to the major tributary drainages in coastal Texas and the lower Mississippi Valley. Many of these major drainages still contain rivers and streams with abundant or extensive riparian corridors, active meanderbelts, erodible bank materials, and some form of dis- turbance within their watersheds that is driving channel or bank instability. The distribution of debris sites in the Pacific is probably tied to climate and geology, since most of the sites are located within or in close proximity to forested mountains and hills. The poor geotechnical competence of the hills and moun- tains of the Pacific region combined with major or extreme precipitation events can result in the delivery of large volumes of debris through mass movement processes. In the valleys of the Pacific region, the delivery of debris is more closely tied to the erosion of stream and riverbanks as a result of instabil- ity induced by watershed disturbances and channel migra- tion. Considerably fewer sites are in the West, probably as a 28

Source: after Barton et al. (2003) LAURENTIAN UPLAND 1. Superior Upland ATLANTIC PLAIN 2. Continental Shelf (not on map) 3. Coastal Plain a. Embayed section b. Sea Island section c. Floridian section d. East Gulf Coastal Plain e. Mississippi Alluvial Plain f. West Gulf Coastal Plain APPALACHIAN HIGHLANDS 4. Piedmont Province a. Piedmont Upland b. Piedmont Lowlands 5. Blue Ridge Province a. Northern section b. Southern section 6. Valley and Ridge Province a. Tennessee section b. Middle section c. Hudson Valley 7. St. Lawrence Valley a. Champlain section b. Northern section (not on map) 8. Appalachian Plateaus Province a. Mohawk section b. Catskill section c. Southern New York section d. Allegheny Mountain section e. Kanawha section f. Cumberland Plateau section g. Cumberland Mountain section 9. New England Province a. Seaboard Lowland section b. New England Upland section c. White Mountain section d. Green Mountain section e. Taconic section 10. Adirondack Province INTERIOR PLAINS 11. Interior Low Plateaus a. Highland Rim section b. Lexington Plain c. Nashville Basin 12. Central Lowland a. Eastern Lake section b. Western Lake section c. Wisconsin Driftless section d. Till Plains e. Dissected Till Plains f. Osage Plains INTERIOR PLAINS (continued) 13. Great Plains Province a. Missouri Plateau, glaciated b. Missouri Plateau, unglaciated c. Black Hills d. High Plains e. Plains Border f. Colorado Piedmont g. Raton section h. Pecos Valley i. Edwards Plateau j. Central Texas section 14. Ozark Plateaus a. Springfield-Salem plateaus b. Boston "Mountains" 15. Ouachita Province a. Arkansas Valley b. Ouachita Mountains ROCKY MOUNTAIN SYSTEM 16. Southern Rocky Mountains 17. Wyoming Basin 18. Middle Rocky Mountains 19. Northern Rocky Mountains INTERMONTANE PLATEAUS 20. Columbia Plateau a. Walla Walla Plateau b. Blue Mountain section c. Payette section d. Snake River e. Harney section 21. Colorado Plateaus a. High Plateaus of Utah b. Unita Basin c. Canyon Lands d. Navajo section e. Grand Canyon section INTERMONTANE PLATEAUS (continued) 21. Colorado Plateaus (continued) f. Datil section 22. Basin and Range Province a. Great Basin b. Sonoran Desert c. Salton Trough d. Mexican Highland e. Sacramento section PACIFIC MOUNTAIN SYSTEM 23. Cascade-Sierra Mountains a. Northern Cascade Mountains b. Middle Cascade Mountains c. Southern Cascade Mountains d. Sierra Nevada 24. Pacific Border Province a. Puget Trough PACIFIC MOUNTAIN SYSTEM (continued) 24. Pacific Border Province (continued) b. Olympic Mountains c. Oregon Coast Range d. Klamath Mountains e. California Trough f. California Coast Ranges g. Los Angeles Ranges 25. Lower California Figure 2.11. Physiographic regions of the United States and photographic archive sites.

Source: after Bailey (1997) 100 POLAR DOMAIN 120 Tundra Division 124 Arctic Tundra Province 125 Bering Tundra (Northern ) Province 126 Bering Tundra (Southern) Province M120 Tundra Division – Mountain Provinces M121 Brooks Range Tundra – Polar Desert Province M125 Seward Peninsula Tundra – Meadow Province M126 Ahklun Mountains Tundra – Meadow Province M127 Aleutian Oceanic Meadow – Heath Province 200 HUMID TEMPERATE DOMAIN 210 Warm Continental Division 212 Laurentian Mixed Forest Province M210 Warm Continental Division – Mountain Provinces M212 Adirondack-New England Mixed Forest - Coniferous Forest – Alpine Meadow Province 220 Hot Continental Division 221 Eastern Broadleaf Forest (Oceanic) Province 222 Eastern Broadleaf Forest (Continental) Province M220 Hot Continental Division – Mountain Provinces M221 Central Appalachian Broadleaf Forest – Coniferous Forest – Meadow Province M222 Ozark Broadleaf Forest – Meadow Province 230 Subtropical Division 231 Southeastern Mixed Forest Province 232 Outer Coastal Plain Mixed Forest Province 234 Lower Mississippi Riverine Forest Province M230 Subtropical Division – Mountain Provinces M231 Ouachita Mixed Forest – Meadow Province M240 Marine Division 242 Pacific Lowland Mixed Forest Province M240 Marine Division – Mountain Provinces M242 Cascade Mixed Forest – Coniferous Forest – Alpine Meadow Province 250 Prairie Division 251 Prairie Parkland (Temperate) Province 255 Prairie Parkland (Subtropical) Province 260 Mediterranean Division 261 California Coastal Chaparral Forest and Shrub Province 262 California Dry Steppe Province 263 California Coastal Steppe, Mixed Forest, and Redwood Forest Province M260 Mediterranean Division – Mountain Provinces M261 Sierran Steppe – Mixed Forest – Coniferous Forest – Alpine Meadow Province M262 California Coastal Range Open Woodland-Shrub- Coniferous Forest – Meadow Province 300 DRY DOMAIN 310 Tropical/Subtropical Steppe Division 311 Great Plains Steppe and Shrub Province 313 Colorado Plateau Semidesert Province 315 Southwest Plateau and Plains Dry Steppe and Shrub Province M310 Tropical/Subtropical Steppe Division – Mountain Provinces M313 Arizona – New Mexico Mountains Semidesert – Open Woodland – Coniferous Forest – Alpine 320 Tropical/Subtropical Desert Division 321 Chihuahuan Semidesert Province 322 American Semidesert and Desert Province 330 Temperate Steppe Division 331 Great Plains-Palouse Dry Steppe Province 332 Great Plains Steppe Province M330 Temperate Steppe Division – Mountain Provinces M331 Southern Rocky Mountain Steppe – Open Woodland – Coniferous Forest – Alpine Meadow M332 Middle Rocky Mountain Steppe – Coniferous Forest – Alpine Meadow Province M333 Northern Rocky Mountain Forest Steppe – Coniferous Forest – Alpine Meadow Province M334 Black Hills Coniferous Forest Province 340 Temperate Desert Division 341 Intermountain Semidesert and Desert Province 342 Intermountain Semidesert Province M340 Temperate Desert Division – Mountain Provinces M341 Nevada-Utah Mountains Semidesert – Coniferous Forest – Alpine Meadow Province 400 HUMID TROPICAL DOMAIN 410 Savanna Division 411 Everglades Province M410 Savanna Division – Mountain Provinces M411 Puerto Rico Province 420 Rainforest Division M420 Rainforest Division – Mountain Provinces M423 Hawaiian Islands Province Figure 2.12. Ecoregions of the United States and photographic archive sites.

31 Geographic Region Source State Stream No. of Photos Physiographic Region Ecoregion California (Central Arroyo Grande 11 24 260 and South Coast) Hopper Creek 3 24 M260 Salsipuedes Creek 5 24 M260 California (Mountain) N. Fork Deer Creek 4 23 260 Kevin Flora, Caltrans California Mad River 4 24 260 (Survey) (North Coast) Trinity River 1 24 260 Yager Creek 6 24 260 California Sacramento River 2 24 260 (Central Valley) Stony Creek 6 24 260 Thomes Creek 4 24 260 California Malibu Laguna 1 21 260 Ayres Associates Inc. Harris Creek 1 25 M262 Pacific Stony Creek 1 24 260 Coast Oregon Shitike Creek 1 24a 242 California Butte Creek 9 23 260 (North) Navarro River 13 24 260 California Callegas Creek 18 24 260 (South) Santa Clara River 26 24 260 San Antonio Creek 9 24 260 Debris Free, Inc. Adams Creek 3 24 M260 (Mike Collier) Ventura River 29 24 260 Oregon Calapooia River 1 23 M240 North Santiam River 3 23 M240 Washington Cowlitz River 7 24 240 N. Fork Skagit River 6 23 M240 N. Fork Skagit River 2 23 M240 S. Fork Skagit River 1 23 M240 Skykomish River 4 24 240 Timothy Diehl (1997) Washington Quetts River 1 24 M240 Debris Free, Inc. Nevada Carson River 6 22 340 (Mike Collier) Utah Colorado River 4 21 340 San Rafael River 7 21 340 Russell Brewer, PE Montana Jefferson River 14 19 M330 MtDOT (Survey) Boulder River 7 19 M331 St. Regis River 3 19 M332 West Chris Miller, NDOT Nevada Carson River 15 22 340 Tim Ularich, UDOT Utah Santa Clara & Virgin Rvs. 11 21 340 New Mexico Rio Grande 1 22 310 Ayres Associates Inc. Colorado Bijou Creek 2 13 330 Idaho Teton River 1 18 M330 Various Web Sites (incl. Utah Santa Clara River 8 22 340 (KSL Channel 5 TV, Beaver Dam Creek 4 22 340 City of St. George) Unknown 11 22 340 Illinois East Skokie Ditch 3 12 220 Debris Free, Inc. Iroquois River 4 12 250 (Mike Collier) Mississippi River 3 12 250 Mackinaw River 3 12 250 Unknown 6 Indiana Eel River 8 12 220 Eel River 4 12 220 Vermillion River 11 12 220 Midwest Wabash River 3 12 220 Kansas Republican River 6 12 250 Smoky Hill River 34 13 330 Minnesota Minnesota River 15 15 250 Missouri Unknown 1 Ohio Unknown 5 Kentucky Unknown 2 Texas Clear Fork Brazos Rv. 6 12 310 Tennessee Harpeth River 8 11 220 Oklahoma Wild Horse River 3 12 250 North Canadian River 12 12 310 Table 2.5. Inventory of photographs of debris at bridge sites with Physiographic Regions and Ecoregions. (continued on next page)

32 Geographic Region Source State Stream No. of Photos Physiographic Region Ecoregion Midwest Brad Rognlie, P.E. Kansas Verdigris River 12 12 250 (continued) KDOT (Survey) Smoky Hill River 73 13 330 Chikaskia River 1 13 330 Elm Creek 25 13 330 Neosho River 19 12 250 Elk River 25 12 250 Unknown 6 Ayres Associates Inc. Kansas Neosho River 38 12 250 Neosho Rv. (US 400) 28 12 250 Verdigris River 14 12 250 Elk River 32 12 250 Missouri Florida Creek 1 12 250 Nebraska South Platte River 1 12 250 Indiana Indian Creek 1 12 220 MnDOT District 1 Minnesota Black River 2 1 212 (web site photos) USGS (Robinson 2003) Indiana White River 6 12 220 Iowa Cedar Creek 3 12 220 Tim Dunlay, IDOT Iowa E. Nishnabotna River 6 12 250 (Survey) Iowa West Nodaway River 3 12 250 J. Holmes, MoDOT (Sur) Missouri Chariton River 9 12 250 Ken Foster, MoDOT Missouri Big Creek 2 12 250 (Survey) W. Fork of Grand Rv. 8 12 250 Larry Cooper, MnDOT Minnesota Minnesota River 2 12 220 (Survey) Minnesota Mississippi River 2 12 220 D. Hill, MnDOT (Sur) Minnesota S. Branch, Wild Rice Rv. 2 12 250 B. Callett, ODOT (Sur) Ohio Great Miami River 14 12 220 Allan Bjorklund, WisDOT Wisconsin Pikes Creek 1 1 210 (Survey) Bad River 1 1 210 J. Conner, TxDOT (Sur) Texas Clear Fork, Brazos Rv. 6 12 310 T. Hertel, TxDOT (Sur) Texas Red River 12 12 250 J. Zirkle, TDOT (Survey) Tennessee Harpeth River 13 11 220 Timothy Diehl (1997) Tennessee Harpeth River 2 11 220 Indiana White River 1 12 220 Debris Free, Inc. North Carolina Deep River 4 4 230 (Mike Collier) Virginia Appomattox River 4 4 230 James River 2 3 230 Meherrin River 3 4 230 East Dan River 2 4 230 Nottoway River 4 4 230 West Virginia Mud River 2 8 220 Tug River 2 8 220 NCDOT (web site photo) North Carolina Deep River 5 4 230 Web Site Photos Tennessee Coal Creek 4 8 220 South Carolina Little River 40 4 230 Debris Free, Inc. Florida Escambia River 2 3 230 (Mike Collier) Tennessee Wolf River 3 3 230 Jackson District Var. 193 11 220 Arkansas St. Francis River 34 3 230 Louisiana Amite River 10 3 230 Louisiana Bayou Boeuf 1 3 230 Ayres Associates Inc. Mississippi Abiaca Creek 1 3 230 Jack Creek 2 3 230 South Sykes Creek 1 3 230 Web Site Photos Louisiana Red River Raft 2 3 230 MDOT Bridge Des. (Sur) Mississippi Coles Creek 1 3 230 Jon Zirkle, TDOT (Sur) Tennessee Wolf River 3 3 230 J. Howell, TxDOT (Sur) Texas San Marcos River 1 3 250 Jon Kilgore, TxDOT Texas Cocklebur Creek 7 3 310 (Survey) Texas San Antonio River 13 3 250 Blaine Laywell, TxDOT Texas Little River 7 3 250 (Survey) Texas Brushy Creek 6 3 250 H. Schroeder, TxDOT (S) Texas Rocky Creek 6 3 250 Timothy Diehl (1997) Texas Brazos River 1 12 310 Unknown Unknown 5 Table 2.5. (Continued).

33 (a) (c) (b) (d) (f) (a) Pacific—Arroyo Grande Creek, CA; (b) Midwest—unknown location, KY; (c) Midwest—Verdigris River, KS; (d) East—Appomattox River, VA; (e) West—Santa Clara River, UT; and (f) South—St. Francis River, AR. (e) Figure 2.13. Photographs of similar debris geometry from sites in different regions of the United States.

result of the more arid climate, which produces less widely distributed debris. The paucity of debris sites in the East may be tied to the greater overall stability of the streams and rivers in the region. 2.4 Survey and Site Reconnaissance 2.4.1 Survey Task 2 required determination of typical debris accumula- tions by surveying state DOTs and other agencies. Since the late 1970s at least three studies related to debris conducted surveys and/or visited state DOT and other agency bridge sites. FHWA’s 1979 study “Debris Problems in the River Environment” by Chang and Shen (1979) presents their lit- erature review, a survey of debris hazards for FHWA regions and state DOTs, a debris hazard map, and a statistical analysis and observations resulting from the survey (see Figure 2.10). Responses from each DOT are compiled in an appendix. Diehl’s 1997 study for FHWA compiles detailed information and maps on sources of drift (debris). Diehl also identifies debris field study sites and presents a generalized map of debris sites based on publications and written and oral com- munications. While Parola et al.’s study for NCHRP (2000) did not include an independent survey, they relied on and interpreted the results of Diehl’s work. NCHRP Report 417: Highway Infrastructure Damage Caused by the 1993 Upper Mississippi River Basin Flooding (Parola et al. 1998a) contains specific information on debris problems and bridge failures for that region. Under Task 2, a survey on debris problems was developed and distributed to DOTs and other agencies. At the sugges- tion of a panelist, a copy of the survey was sent to all panel members. The survey also included a request that recipients forward the survey to others within the agency who may be more knowledgeable on specific topics (e.g., bridge inspec- tors, maintenance personnel). The response to the survey was outstanding. Eighty-eight responses representing 30 states and Puerto Rico were received. Appendix B contains a summary of respondents and their cor- responding state and agency. Survey responses were entered into a Microsoft® Access database. A copy of the survey and instructions for viewing the survey responses are included in Appendix B. The database is available for download on the NCHRP Project 24-26 web page (http://144.171.11.40/ cmsfeed/TRBNetProjectDisplay.asp?ProjectID=725). The Microsoft® Access relational database is designed to allow rapid screening and organization of a data set based on predetermined attributes of each document. The system allows a user to examine the database using the Structured Query Language (SQL) format to obtain, in this case, a listing of all survey responses meeting the user’s query criteria. From this 34 Figure 2.14. General geographic regions of the United States and photographic archive sites.

framework, a sublist of survey responses can be developed. The database allows the user to specify multiple query criteria, so refinements to the sublist can also be readily performed. This approach was used in the evaluation of survey responses in the next section. 2.4.2 Analysis of Survey Responses Surveys were returned from 88 respondents, represent- ing 30 states including Alaska and Hawaii. In addition, two responses were returned from Puerto Rico; therefore, 84 responses representing the continental United States were received. A breakdown of responses in accordance with the five geographic regions is shown in Table 2.6 and Figure 2.15 (some states may be split into more than one geographic region). The survey was partitioned into seven categories. Five of the seven categories asked respondents to rank debris-related questions on a scale of 0 to 5 in terms of importance/severity of problem. Responses to all the questions in each of these five categories were examined using the Analysis of Variance (ANOVA) method to determine if statistically significant differ- ences in the responses could be assigned to geographic regions. In general, responses from different geographic regions tended to be similar in nature, indicating that there are rela- tively insignificant regional differences regarding the nature of drift accumulations at bridges. Similarly, the types of drift- related problems reported tended to show very few regional differences. In some instances, however, regional differences were found to be statistically significant. A discussion of the responses is provided in the following subsections. The dis- cussion is organized to correspond to the seven categories of the survey. Part 1. Source Areas that Produce Drift and Debris Nationwide, respondents reported that unstable stream- banks were unquestionably the single most important con- tributor of drift material that accumulates at bridges. ANOVA testing indicated that at a significance level α = 0.05 (95% confidence level), there was no significant difference among the responses gathered from each of the five geographic regions. Watershed management issues were ranked second in impor- tance, and landslides were found to be least important. Table 2.7 provides a summary of the rankings by region, and shows the nationwide results as well, for the continental United States. Part 2. Bridge Characteristics Responses to Part 2 were mixed. This part of the survey was designed to determine if certain substructure types were more or less prone to debris-related problems. Nationwide, respon- dents indicated that pile bents were the substructure type most likely to be affected by debris and that skewed piers were more problematic than piers aligned with the flow. ANOVA results indicated that some regional differences exist in the mean values of the responses to this section at a significance level α = 0.05. Therefore, to discriminate between regions, further analysis was conducted using the two- tailed Student’s t-test, also with a specified significance level α = 0.05. Respondents from the East region considered wall-type piers most likely to have debris problems. There was a 95% confidence level that the mean value of the responses from the East was different from all other regions. In contrast, responses from the South region were exactly the opposite: respondents from the South region reported that wall-type piers were the least likely to have drift accumulations. In addi- tion, respondents from the East region along with those from the Pacific Coast region considered pile bents with cross brac- ing to be least likely to have problems with debris, at a level that is significantly different compared to the other three regions. From the data provided, it is not possible to tell whether these regional differences are due to certain types of substructures being more prevalent in some regions com- pared to others. Table 2.8 provides a summary of the rankings by region. Part 3. Debris Accumulation This part of the survey asked respondents to rank common elements that are typically used to characterize drift accumu- lations. Without question, debris clusters on a single pier were considered the most common problem found in all regions of the continental United States. In addition, floating material was reported to be of more concern than submerged material. In the West region, respondents considered drift buildup between the pier and abutment to be more of a problem than did their counterparts across the rest of the country, to a sta- tistically significant degree. However, respondents from the West region did concur that accumulations at a single pier were still the number one problem. Although not statistically significant, it is interesting that respondents from the West region, where many streams are 35 Geographic Region Responses States Represented Pacific Coast 8 2 West 10 7 Midwest 29 11 South 20 5 East 17 8 Table 2.6. Breakdown of responses to survey.

ephemeral or intermittent, reported that submerged ma- terial was equally as common as floating material. This phe- nomenon may be the result of accumulations that slide to the streambed during the falling limb of a flood event. If not removed, these materials may become incorporated into the streambed sediments and are then submerged by the next event. Table 2.9 provides a summary of the rankings by region. Part 4. Debris-Related Scour at Bridges Respondents were asked to consider the nature of the scour problem that is created when drift accumulates at bridges. Some variability was noted in the responses, although there were no statistically significant differences between the five geographic regions. In general, local scour at the pier where the drift cluster has accumulated was considered the most common scour issue. Bank instability in the vicinity of the bridge, caused by re- direction of the flow due to debris blockage, was considered the next most common problem. Table 2.10 provides a summary of the rankings by region. Part 5. Maintenance This part of the survey asked respondents to provide an esti- mate of the percentage of the average annual maintenance budget that was expended on removing debris, or repairing 36 Figure 2.15. General geographic regions of the United States and survey sites. Region Q1. Unstable Streambanks Q2. Landslides in Watershed Q3. Watershed Management Pacific Coast a b c West a c b Midwest a c b South a c b East a c b Nationwide a c b Note: An “a” ranking is highest in importance/severity Table 2.7. Survey results ranking source areas that produce drift.

damage caused by debris at bridges. Only about half of the respondents answered this question. Many accompanied their estimate with a comment such as “Unknown,” “Minimal,” or “After large floods.” Estimates for maintenance costs associated with drift and debris ranged from 0% of the annual maintenance budget up to 15%. One respondent reported that 100% of the annual main- tenance budget was spent on debris; this particular response is most certainly an outlier. The most typical responses were “1%” or “less than 1%.” This part of the survey had the most uncertainty in the nature of the responses. This uncertainty may be due in large measure to the sporadic nature of drift recruitment in water- ways and its subsequent accumulation at bridges. For exam- ple, the magnitude and frequency of flood events over a period of wet years compared to dry years, coupled with the timing of debris removal activities, can create a virtually infinite com- bination of scenarios for maintenance crews and their budgets as related to debris at bridges. Part 6. Characteristics of Debris This part of the survey requested information on the type of drift that triggers the beginning of the process of debris accumulation at an initially clean pier. The intent here was to follow up on concepts developed by previous investigators, such as the “key log” mechanism described by Diehl (1997), as discussed in Section 2.1. 37 Region Q1. Drilled Shaft Piers Q2. Pile Bents Q3. Pile Bents with Cross Bracing Q4. Pile Cap/ Pile Group Q5. Wall-Type Pier Q6. Pier Skewed to Flow Direction Pacific Coast e a f b d b West f b c c e a Midwest f a b e d c South c a b e f c East d c f e a b Nationwide f a d e c b Note: An “a” ranking is highest in importance/severity Table 2.8. Survey results ranking substructure types that accumulate drift. Region Q1. Cluster on a Single Pier Q2. Raft Spanning Two or More Piers Q3. Debris on Superstructure (deck/railing) Q4. Debris Spanning from Pier to Bank Q5. Submerged Debris Q6. Floating Debris Pacific Coast a d f c e b West a f f b b b Midwest a c f e c b South a c f e d b East a d d f c b Nationwide a d f e c b Note: An “a” ranking is highest in importance/severity Table 2.9. Survey results ranking characteristics of drift accumulation. Region Q1. Local Scour at One or More Piers Q2. Local Scour at Abutment(s) Q3. Contraction Scour Caused by Debris Constriction Q4. Overtopping of Approach Abutments Due to Flow Blockage Q5. Streambank Instability Due to Flow Redirection Pacific Coast b e b e a West b d c e a Midwest a b c e d South a d b e b East a c d e b Nationwide a d c e b Note: An “a” ranking is highest in importance/severity Table 2.10. Survey results ranking scour and stream instability problems caused by accumulated drift.

Respondents reported that key logs less than 25 ft (7.6 m) in length are most often the initiators of drift accumula- tion. Shrubs and brush-type vegetation was ranked second nationwide, and in fact, respondents from the West ranked shrubs and brushy vegetation as being the most important contributor initiating debris accumulation. This observa- tion was confirmed by t-test at α = 0.05 level of signifi- cance. Table 2.11 provides a summary of the rankings by region. Part 7. Potential Sites for Monitoring Many respondents provided information on specific sites in their regions that exemplified either typical or chronic drift-related problems. In most cases, these respondents also provided supporting information, which usually con- sisted of bridge plans accompanied by photographs, and sometimes inspection records as well. These types of addi- tional information were noted in the database; when photo- graphs were provided, they were added to the photographic archive. Respondents in 16 states offered one or more sites for field study, as indicated in Table 2.12. In particular, Kansas DOT provided detailed information on 21 bridges, primarily located in eastern Kansas, where drift has been a documented and recurring problem. As a result of this information, four sites in southeastern Kansas were selected for a field pilot study (see Section 2.5). Summary Based on the 84 surveys received from five regions repre- senting the continental United States, a qualitative interpre- tation of the survey responses, combined with an analysis of statistical variance of the numerical rankings by region, indi- cates the following: • Drift material is derived primarily from unstable stream banks. • It is most common for drift problems to occur at pile bents compared to other pier configurations. A possible excep- tion may be in the East region, where wall-type piers are reported to have the most problems. • Large logs are not necessary for drift to begin accumulat- ing at bridge piers. Most commonly, drift accumulation is initiated by logs 25 ft (7.6 m) or less in length or, in the West region, by shrubs and brushy vegetation (e.g., willows, Tamarisk). • Nationwide, the most common drift configuration at bridges is that of an individual cluster of material on a single pier. • Drift is most likely to exacerbate scour at bridges by caus- ing either (1) local scour at an individual pier or (2) stream instability due to flow redirection. • Highway departments typically do not collect cost data asso- ciated with drift removal and associated bridge repairs. Maintenance needs are sporadic in nature and are most often related to larger flood events. A reasonable range of cost esti- mates might be 0.5% to 1% of the total annual maintenance 38 Region Q1. Key Log Less Than 25 ft (7.6 m) Long Q2. Key Log Greater Than 25 ft (7.6 m) but Less Than 75 ft (23 m) Q3. Key Log Greater Than 75 ft (23 m) Q4. Primarily Shrubs and Bushy Vegetation Q5. Construction Debris/Manmade Material Pacific Coast a c d b e West b c e a d Midwest a b d c e South a b e c d East a c e b d Nationwide a c e b d Note: An “a” ranking is highest in importance/severity Table 2.11. Survey results ranking drift characteristics that initiate the accumulation process. Geographic Region Contacts States Represented Pacific Coast 3 CA West 4 AZ, MT, NM, UT Midwest 11 IA, KS, MN, MO, OH, TN, TX South 6 LA, TX East 3 NY, PA, VA Table 2.12. Sites offered for field study by region and state.

budget in years with little flooding and perhaps 2% to 5% in years that experience one or more significant events. • Relatively high standard deviation from the mean numer- ical responses within each region indicate that it is possible to experience the full range of drift-related problems in any given region; however, ranking the responses does provide an indication that some problems or issues are more com- mon than others, as noted in the preceding subsections. 2.5 Site Reconnaissance 2.5.1 Equipment and Preparation The objectives of the site reconnaissance were to obtain field measurements, photographs, geomorphologic informa- tion, channel types and flow patterns, associated bridge and pier geometrics, and historical scour depths, where available. These data were then used with the photographic archive to characterize debris accumulations at bridge piers in the field as a basis for Phase 2 laboratory studies and prediction meth- ods. For the field pilot study in southeastern Kansas, the follow- ing equipment resources were used: • Articulated-arm truck developed under NCHRP Project 21-07 (Schall and Price 2004) to deploy portable scour- monitoring instrumentation (probes, sonar) at bridges under flood flow conditions including high sediment con- centrations and floating debris. • Hydrographic survey equipment and experienced engineer/ survey personnel, a Differential Global Positioning System (DGPS), and survey-grade fathometer. The Ayres’ hydro- graphic survey boat is equipped with this equipment as well as a computer and software to accurately and efficiently col- lect bathymetric data. • Standard surveying instruments, including a total station and laser range finder. • Interphase Twinscope™, a forward-scanning sonar using phased array technology, originally tested for NCHRP 21-07. • Mounting brackets for a single-beam sonar to permit backward-looking and side-looking underwater measure- ments obtained by the articulated-arm truck. The goal was to assemble the best available equipment using current technology for the field pilot study to determine the feasibility and accuracy of both above water and underwater measurements of debris characteristics. 2.5.2 Field Pilot Study In response to the survey (Section 2.4), many DOTs also identified possible field sites. Kansas DOT, in particular, offered support in the field, including identifying sites, coor- dination for access, and on-site support such as traffic con- trol. As a result, the research team coordinated with Kansas DOT to be in a flood-watch mode ready to mobilize the artic- ulated arm truck and equipment in the mid-April to end of May 2005 time frame for sites along the I-70 corridor in Kansas. Kansas DOT personnel began tracking numerous sites in Kansas and provided a continuous flow of information on debris conditions throughout the state. The field pilot study at bridge sites in southeastern Kansas was scheduled for the period April 25–28, 2005. During the week of April 25, 2005, close coordination was maintained with Kansas DOT person- nel for site access, traffic control, etc. A detailed trip report for the field pilot study is included in Appendix C. 2.5.3 Field Pilot Study Results The field pilot study was an expensive undertaking. Preparation for and completion of the 5-day field pilot study consumed approximately 35% of the Task 2 budget. A sub- stantial effort was required to assemble, recondition, or recon- figure the necessary equipment, particularly the NCHRP 21-07 articulated arm truck. The following paragraphs provide a brief summary of instrument performance during the pilot study. The articulated-arm truck, as modified, performed well. The truck has a telescoping articulated arm, which can be positioned over the side of a bridge deck to determine surface characteristics of a debris pile. The truck is a 1-ton dual wheeled Ford F-450 with a Palfinger articulating knuckle- boom telescoping crane mounted on the bed and chassis. The crane is operated from the flat bed of the truck and is able to take direct measurements from a water level just below the bridge deck downward to about 30 ft (9 m). The crane can be articulated to allow positioning around the perimeter of the debris pile. A method of recording data around the debris pile and tak- ing topographical shots on the actual debris pile was devel- oped for this study. In this application, a chain of known length was attached to the end of the articulated arm and was positioned at locations on and around the pile to take mea- surements and provide an above water map of the debris pile. The known length of the chain was added as a “rod height” to the vertical position. Another method that was developed for the field study used the same chain method as just described, but a prism cluster was hung from the chain. The prism was shot with a total station at each position, and the debris pile was mapped using the total station and data logger for positioning of the end of the chain. This procedure provided another relatively accurate method of mapping the debris pile in relation to the bridge. For underwater measurements of debris, separate stream- lined heads were built to house a wireless sonar. Instead of the 39

sonar having to be mounted to point straight down, the side- looking or backward-looking heads allow the transducer to be mounted to look horizontally under the debris pile. The head was attached to the articulating arm and positioned around the pile in the same manner as the downward-pointing sonar. Computer software for sonar measurements with the crane allowed point measurements or continuous recording. The user can select from two data collection modes: Point Data collects one data point at a specific location; Continuous Recording col- lects data at 1-second intervals continuously as the articulating arm is moved around the debris pile. The hydrographic survey boat was used for documenting geomorphic conditions, such as bank stability, debris source areas, overall channel type, and flow patterns. The boat is a 16 ft flat bottom Jon Boat with a 50 hp jet drive outboard motor. The jet motor is very maneuverable and allows the boat to be operated in shallow water without risk of damag- ing a standard prop. The boat and motor can be operated in less than 1 ft (0.3 m) of water. A Leica total station with data logger was used to measure horizontal and vertical angles of the debris pile from the bank. The total station displays a digital readout of the angles from the instrument to the points being surveyed. The total station was also used to take electronic distance measurements (EDM) of the bridge deck and adjacent bank line. Similarly, a laser range finder was used to measure distance, height, and width of objects. The range finder gives accurate distances of an object up to 2,000 ft (610 m) without a prism. It also has a built in inclinometer to measure vertical angles. The Interphase Twinscope™, used as a handheld measure- ment system, was tested. The Twinscope™ is a continuously scanning sonar that displays a 90° underwater view in front of the sonar. The Twinscope™ is designed to be mounted to the hull of a boat and has been used successfully in ocean appli- cations. In this application, the instrumentation was lowered into the water and pointed in the direction of the debris pile to get an underwater view of the debris pile shape. The results were inconclusive. Echos, scatter, and interference in a shal- low water environment rendered the interpretation of results speculative, at best. 2.5.4 Recommendations The field pilot study to southeastern Kansas provided a real- ity check on the cost and technical limitations, primarily instrumentation, of obtaining debris-related data in the field. It was concluded that debris accumulations are highly variable in the field and depend not only on watershed conditions, but also on bridge type and geometry. The temporal variability of debris accumulations in relation to the flow hydrograph adds to this complexity. Thus, no two debris clusters are alike and determining or classifying “typical” debris configurations based on limited field data would be difficult. A variety of instruments were deployed during the field pilot study, representing the range of technology currently available, with mixed results. While surveying and measuring the surface expression (above water) of a debris cluster is easily accomplished, obtaining detailed information on the underwater characteristics of debris is extremely difficult and time consuming, and the results are inconclusive. During the Task 1 literature review and the Task 2 survey, an extensive archive of photographs of debris at bridge piers was assembled (see Section 2.2 and Appendix A). The origi- nal effort to obtain these photographs extended over many years and covered the Pacific Coast, Western, Midwestern, Eastern, and Southern regions of the country. This excep- tional collection of debris photographs provides a resource that was not available when the original work plan was devel- oped (see Section 1.2.2). Consequently, the NCHRP 24-26 panel concurred with the research team that additional field work should not be conducted for this study. Rather, resources should be reallocated to additional analysis of the photographic archive of debris for interpreting regional characteristics of debris (see Section 2.3) and character- izing debris accumulations for laboratory testing (see Section 3.5). 40