UW Masters Thesis by Florentiu Damian : Cross-drain Placement to Reduce Sediment Delivery from Forest Roads to Streams

A thesis submitted in partial fulfillment of the
requirements for the degree of

Master of Science

University of Washington
2003

Chair of the Supervisory Committee:
Professor Peter Schiess
Management and Engineering Division, College of Forest Resources

Ditch relief culverts can reduce road sediment delivery to streams by allowing infiltration and sediment filtering across the forest floor. Below the last ditch relief culvert, all the sediment routed by the ditch will be delivered directly to the stream. The last ditch relief culvert should be as close to the stream crossing as possible. If the ditch relief culvert is too close to the stream however, then there is little potential for sediment filtering. This tradeoff between minimizing the amount of water delivered directly to the stream and maximizing the distance for outflow filtration poses a question of where the last ditch relief culvert should be placed.

A model has been developed which allows a designer to place ditch relief culverts at various locations and subsequently evaluate their impact on sediment delivery to streams. The main feature of the model is its immediate feedback to the forest engineer in visual as well as quantitative form. It allows the designer to dynamically assess the sediment impacts associated with each culvert as it is placed on the road network. Sediment delivery and routing algorithms are based on accepted methodologies. Current as well as planned roads can be evaluated and the potential for improvements documented in a quantifiable and repeatable way.

The model was tested on a portion of the Tahoma State Forest, situated south of Mt. Rainier. Two existing road systems with 28 and 39 stream crossings and 82 and 86 cross drain culverts respectively, were analyzed. Interactively relocating 20 and 35 of the cross drains resulted in a three quarter reduction in sediment delivered to the stream system. The last culvert was usually placed about 100 - 200 ft of a stream according to local conditions, challenging one of the regulatory recommendations to place a cross drain within 100 ft of a stream crossing. Forest engineers and regulators now have a design tool to assess effectiveness of a cross drain system rather then simply relying on culvert spacing and count.

LIST OF FIGURES

Figure 01	Schematics of common cross drain types
Figure 02	Cross sectional prism types with their drainage patterns
Figure 03	Cross section of insloped road prism with cross drain
Figure 04	Cross drain system dispersing sediment
Figure 05	Screen capture of the WEPP:Road interface
Figure 06	Effects of moving cut-off culvert C along the road alignment
Figure 07	Analysis setup for a single cut-off culvert scenario
Figure 08	Sediment delivery rate to stream for various sediment production rates
Figure 09	Sediment delivery rates for various stream crossing angles
Figure 10	Setup for the two cross drain experiment
Figure 11	Example of a real case road layout with culverts
Figure 12	Flow chart of cross drain design workflow
Figure 13	Sediment delivery represented as proportional symbols
Figure 14	CULSED Model-View-Controller internal architecture
Figure 15	Logical network of the ditch model
Figure 16	Shaded relief model of the North Tahoma planning area
Figure 17	Initial sedimentation from East North Tahoma road network
Figure 18	Initial sedimentation from West North Tahoma road network
Figure 19	Composite map of a culvert placement investigation process
Figure 20	Sediment delivery from East North Tahoma road network after redesigning the cross drain system
Figure 21	Sediment delivery from West North Tahoma road network after redesigning the cross drain system
Figure 22	Topographic detail of North Tahoma, 6ft resolution DEM
Figure 23	The Component Category Manager Application for registering CULSED components
Figure 24	The CULSED toolbar
Figure 25	Start menu item
Figure 26	Set up road grade menu
Figure 27	Set up road intersections with more than one water routing choice
Figure 28	Sediment modeling parameters
Figure 29	Option menu. Default Sediment Model (left), Default Road (right)
Figure 30	Road Geometry Setup has been run. A road grade is computed
Figure 31	Flow Setup and Sediment Analysis have been run
Figure 32	Culverts have been moved to near optimal locations

Tables

The Road Sedimentation Problem

Road Prism Structure and Drainage Patterns
Forest roads are at the core of modern forestry, providing quick, economical access to the wooded areas. To facilitate forest operations in areas managed for timber production, a well developed network of roads must exist. Timber harvesting and forest management activities drive the expansion of the existing networks into the inaccessible parts of watersheds and sub-basins. As engineered structures on the landscape, forest roads have the potential to disrupt the drainage characteristics of the watersheds they traverse by altering their natural flow patterns. Water from precipitation fallen on the exposed road surface, moving under the influence of gravity, follows a new path dictated by the local road grade. The cuts required by the basic structure of a road prism across a hillside also capture overland flow in close vicinity of the road tread. The accumulation and movement of this water within the roadway is detrimental to the well functioning of the road (Schiess and Whitaker 1986). Problems ranging from erosion of the traveling surface to saturation of the sub-grade and mass failure stem from these circumstances. Such processes are usually accelerated during storm events as larger volumes of water move at higher velocities, developing more destructive energy. In order to keep the road prism in good condition and avoid structural damage, roads are outfitted with drainage features.

Three essential prism components determine the water flow within the roadway: the road crown, the road ditch and the cross-drains. The crown represents the side sloping of the road surface. Its main role is to disperse water laterally away from the tread and prevent harmful flow routing along the travel surface. The side ditch, when present, collects and routes water longitudinally along the road alignment towards the nearest stream crossing or cross-drain structure. The cross drains are routing elements that empty the side ditch and redirect its accumulated water across and away from the road prism, onto the side slope, where it will reenter the natural flow regime. Frequently, cross drains are implemented by drainage pipes built into the road bed but other, less common implementations exist as well (Figure 1).

According to their drainage configuration forest roads can be classified in the following categories: insloped with a ditch, crowned with a ditch, outsloped with ditch and outsloped with no ditch (Figure 2). Each of these types generates different drainage patterns and impacts the original watershed drainage accordingly.

The insloped roads with a side ditch are a common occurrence among forest roads due in part to their ease of maintenance and increased traffic security. The drainage system of an insloped road involves a side ditch, cross drains and stream crossings. The surface of the road, being sloped inwards toward the cut slope, reroutes all water it captures to the side ditch. Consequently, the surface runoff intercepted by the cut bank is concentrated with the rerouted road capture. This can potentially lead to the accumulation of high volumes of water in the ditch. If no cross drains are present this water spills out at the end of the ditch, often directly into a stream at the nearest stream crossing. In cases where the road grade is relatively high, the energy of the moving water can reach potentially destructive levels. Cross drains are placed along the road alignment at various locations to empty the side ditch and reduce the possibility of water induced road and environmental problems.

At the opposite pole of road prism drainage types are the outsloped roads with no side ditch. The outlsoped roads with no ditch disperse all the water they capture on their fill slope. In the absence of a side ditch, concentration and rerouting of large volumes of water does not take place. As the energy of the flowing water is kept to lower levels, its damaging potential is also reduced. Moreover, the lack of direct ditch drainage into a stream network induces less hydrologic disturbance. Harmful processes of erosion and sedimentation generated by moving water are minimal. From an environmental standpoint, the outsloped road type represents a more suitable design. However, traffic security concerns raised by log trucks and heavy equipment, critically restrict its usability. Outsloping is normally applied to minor roads and spurs.

Other intermediary road types inheriting physical characteristics from both the insloped and outlsoped roads do exist (Figure 2). Their hydrologic impacts on the environment are closer to one or the other base types described above, with variations dictated by their elemental differences.

Sediment Production and Delivery Mechanisms
Stream sedimentation is an environmental problem generated by the expansion of sediment into streams in excess of the natural amount from hill-slope erosion and soil creep. Forest roads have been identified as a major contributor to sedimentation. Fine sediment generated by roads is transported into adjacent streams, leading to degradation of water quality and damage to aquatic habitat (WA Forest Practices Board 2000). Sedimentation from roads can be approached from two major perspectives: sediment production and sediment delivery.

Sediment production refers to the physical generation of sediment, the process of detaching soil and parent material particles under the influence of various agents. Different premises govern the generation of sediment on each individual road component. The cut and fill slopes are subject to surface erosion occurring when detachable soils are exposed to erosion factors such as: overland flow, raindrop splash, freeze-thaw, dry ravel, and other biogenic processes (WA Forest Practices Board 1997). The litter cover, typically present in a forested environment, protects the soil against all these factors by dissipating erosive energy before it reaches the surface. Road construction operations however, expose the soils on the cut and fill banks, increasing the potential for particle detachment. New roads tend to generate significant quantities of sediment for the first few years of their existence but in time, as side slopes re-vegetate, the sediment production drops. The road’s running surface is another major sediment producer. Traffic is the destructive agent acting upon the surface material. The action of tires against the road surface will grind and dislocate particles into smaller units that can eventually be carried away by other agents like water and wind. Some researches consider traffic to be the single most essential factor influencing sediment generation, capable of increasing sediment amounts by one order of magnitude or more (Reid and Dunne 1984). The side ditch can also be considered as a source of sediment where the erosion is caused by moving water. In the cases where water gains sufficient energy to overcome the soil’s sheer strength, particle detachment is observed.

Sediment delivery refers here to the transportation of sediment generated by the road prism into neighboring stream networks. Fine sediment travels in suspension, carried by overland flowing water along its paths towards the river system. As an active component of watershed hydrology, the road drainage has a fundamental impact on the sediment delivery process. By determining the local water flow within and about the roadway, the road drainage implicitly controls the sediment routing (Section: Road Crowning and Drainage Patterns). The various road drainage configurations in conjunction with the actual road location drastically affect the amount of sediment reaching the stream. For example, in the cases of roads that have numerous water crossings, the presence of a side ditch establishes a direct connectivity between the road drainage and the stream network. The side ditch typically empties at stream crossings, unloading all accumulated sediment into the water. This configuration results in a high potential for stream sedimentation. Other high delivery scenarios include valley bottom roads that run parallel and within close distance to a stream. Large quantities of sediment can reach the stream laterally through surface runoff and water diverted from existing cross drain culverts. On contrast, roads with an outsloped surface geometry and ridge top roads located far away from any streams, have minimal impacts on sedimentation and can be considered as disconnected from the natural watershed drainage.

An important aspect when examining overland delivery on lateral slopes is the site’s filtering potential. Through filtering, parts of the sediment produced by the road are deposited before they have a chance to enter the stream network. Filtering is a complex process, fundamentally based on energy loss and infiltration. The carrying water’s transport capacity is proportional to its velocity. By reducing water velocity soil particles are dropped from suspension and settle down. The porosity of the soil surface influences the infiltration rates and further contributes to the deposition and absorption of fine sediment. In practice, a reduction of sediment delivery is obtained by diverting the water onto vegetated side slopes where the litter layer, plants, and gentle slopes can slow down the surface runoff enough to trigger these filtering effects. Various researchers have shown that the nature of the parent material and the local micro-topographical features have a major influence on the distance the sediment can travel downhill on a side slope. In some extreme cases these distances can be fairly significant (Ketcheson and Megahan 1996), but usually the filtering effects are seen within the first 200 ft from the road (WA Forest Practices Board 1997).

Current Management Techniques for Reducing Sediment Impacts from Forest Roads
Contemporary forest practices regulation requires protection of the neighboring streams from harmful road generated sediment. Although no quantification of minimum acceptable impacts is provided, it is desired that water quality and aquatic habitat not be affected. Ideally no sediment from roads would be delivered to streams (WA Forest Practices Board 2000). In order to meet regulation requirements several road management techniques are effectively applied.

A large part of these techniques address the sedimentation at the production end, acting directly onto the sediment producing factors:

Other methods are centered on the actual sediment transport, designed to decrease sediment impacts by obstructing the physical flow of sediment into streams:

Cross Drain Systems and Sediment Reduction
Cross drain systems have originally been devised as an engineered solution for reducing the adverse effects of excess water within the roadway. They are commonly composed of culverts placed at key locations along the road alignment that drain the side ditch of its accumulated water. Positioning and spacing of these culverts are essential for keeping the road prism in a well functioning state. With the recent evolution of road design into a more environmentally aware paradigm, a new challenge for the cross drain systems has surfaced. Due to their intrinsic properties of intercepting and rerouting sediment-laden ditch water, cross drain systems emerged as a potential solution to the stream sedimentation problem. Thus, in addition to the functionality dictated by the prism health state, another prerequisite was added: to reduce sediment delivery to stream networks.

Designing cross drain systems to meet this new functionality asks for an analysis of the potential amount of sediment delivered by each culvert. In order to reduce sediment delivery, a cross drain must divert the sediment-laden water from the road ditch onto the side slope where it can be dispersed and filtered prior to reaching a stream (Figure 3).

The filtering capabilities of a side slopes vary with location as micro-topography and vegetation conditions fluctuate (Section: Sediment Production and Delivery Mechanisms). As sediment accumulates continuously along the stretches of road between culverts, the amount available at each location is directly affected by the cross drain spacing. An effective reduction of the sediment delivery from a road network requires a certain number of culverts strategically placed to take advantage of local filtering capabilities (Figure 4). It is important to note that these locations may not always coincide with optimal locations for prism drainage but are not mutually exclusive. When the amount of sediment available for delivery at a particular culvert location exceeds the lateral slope’s filtering potential sediment could still reach the stream. These cases are often met where a road is located too close to the valley bottom or a cross drain is placed too close to a stream crossing.

Policy and Design Restrictions
The recent aquatic wildlife crisis in the Pacific Northwest in conjunction with an increased awareness of environmental issues led to the adoption of a new policy for the timber industry. As forest roads are considered potentially harmful to the environment a particular set of restrictions has been imposed on forest road construction that directly affects the cross drain system design. Designers are required to minimize entry of ditch water and surface sediment into streams by dispersing sediment onto the forest floor.

Cross drain spacing is seen as an essential aspect of sedimentation reduction. Generally, cross drains are to be spaced at regular intervals along the road. The distance between culverts is provided as a function of road grade, side slope, average distance above streams, road surface condition and use, precipitation, and soil erosion potential. The rules also specify that the distance between a stream crossing and the first upslope cross drain is important to the volume of sediment delivered and recommend that a culvert should be installed 50 to 100 feet above all stream crossings (WA Forest Practices Board Manual 2000).

Even though the site filtering potential is not explicitly referenced by the current regulation, the cases where culverts are too close to a stream crossing and diverted sediment could still reach the stream network are recognized. The manual recommends either avoidance of such situations or implementation of additional measures such as sediment traps or ponds, rock armored ditches, and vegetated ditches.

A sufficient number of cross drains should be installed in order to prevent ditch scour, over flowing cross drain capacity or erosion at cross drain outlets. The manual requires designers to make use of natural swales that the road crosses in order to avoid rerouting water along the ditch, where it can pick up and transport sediment.

Using these guidelines makes it possible for an experienced professional to design a functional cross drain system. However, one potential draw back of spacing cross drains at regular intervals is that the site filtering potential might not be fully exploited. In areas highly susceptible to sedimentation, a non-uniform spacing of culverts to take advantage of local terrain and lateral sediment retention capabilities might be more suitable. For a more rigorous analysis of the sediment production and delivery from forest roads some specific software tools exist.

Sediment Modeling and Existing Tools
Due to the complex nature of the sedimentation process a precise measurement of sediment impacts form forest roads is not always possible. Several software programs were created in order to model the road-stream sediment interaction and estimate the amount of sediment delivered.

The Water Erosion Prediction Project (WEPP) is a simulation program, originally developed for agricultural purposes as a replacement for the Universal Soil Loss Equation (Elliot et al. 1999b). It is a complex program that models the processes that lead to erosion including infiltration and runoff, soil detachment, transport and deposition, plant growth and residue decomposition. WEPP works with a given slope profile and runs simulations over a specified period of time under a multitude of customizable parameters. The Forest Service Moscow Lab has developed a set of specialized interfaces in order to simplify WEPP use for erosion and sediment delivery from forest roads.

X-DRAIN is a basic interface to accessing predicted sediment yields from over 130,000 WEPP simulations ran by soil erosion specialists (Elliot et al. 1999a). It was developed to simplify and speed up the application of WEPP for simple forest road settings. The end user has limited control over climate, soil, side slope and distance to streams parameters. Road geometry and distance to streams are assumed uniform along the analyzed road segment. The total sediment yield in lb/year is presented on a tabular form for a fixed number of combinations of road gradient and cross drain spacing values.

WEPP:Road is meant to be a more refined sediment modeler capable of modeling one road segment at the time. As inputs, it accepts road surface information and a customizable climate description. The modeling can be done over a user defined period of time (Figure 5). The sediment yield to the stream network in lb is reported as a single number together with the additional average precipitation, runoff, and the amount of sediment leaving the eroding portion of the road prism. There is also the option of an abbreviated hillslope output presenting a distribution of erosion and deposition, the presence of a sediment plume in the forest, and the particle size distribution of sediment delivered to the channel (Elliot et al. 1999c).

One major limitation of the WEPP based programs is that they are spatially non-explicit and thus incapable of distinctly placing the sedimentation processes within a road network. Other organizations have approached this problem within a more spatially aware context using Geographic Information Systems (GIS) as a base for their analysis. GIS have become a standard in environmental modeling. Their capacity of modeling overland flow is what makes them indispensable to spatially distributed phenomena involving streams and water routing.

SEDMODL is a GIS based, road erosion and delivery model developed by Boise Cascade Corporation in cooperation with the National Council on Air and Stream Improvement. The model identifies road segments with a high potential for delivering sediment to streams in a given watershed. It uses spatial information to determine the proximity of the roads to the stream network. Sediment delivery is then calculated for the roads that drain to streams using methods derived from the Washington Department of Natural Resources Standard Method for Conducting Watershed Analysis and WEPP. The program is designed as a flexible, multipurpose tool that can be used both for screening purposes or a more detailed sediment analysis. For more reliable results a set of specific road attributes is required. They can include: road use, surface type, road width, construction year, cutslope height, road geometry type, and road gradient. If culvert locations are known they can be inputted as GIS layer and will affect sediment computations accordingly (National Center for Air and Stream Improvement 2002).

The Washington Road Surface Erosion Model (WARSEM) is new software from Washington Department of Natural Resources intended as a long term road management planning tool. It can model sedimentation and drainage at four different levels from the broad basin scope to the individual road segment. An increasingly complex amount of road information is required with each superior level. The model stresses out the importance of accurate, field verified input data towards a successful sediment budget. The model is implemented as an Access database without a spatial component. SEDMODEL2 results can be imported to generate performance metrics in a long term best management practices analysis (Watershed GeoDynamic et al. 2003).

The Need for a Specialized Cross Drain Design Tool
The contemporary emphasis on modeling the processes that lead to erosion and sediment transport resulted in a well represented collection of computer programs. These software tools were created as analysis packages and perform well when used for general identification of sedimentation problem areas or when examining isolated parts of already built road networks. However, as multiple design alternatives are evaluated during the road design stage, these tools may fall short of functionality, prove slow and relatively ineffective. They are focused on accurate modeling of sedimentation processes but do not provide the means for minimizing total sediment delivered, especially important when designing cross drain systems. None of these tools actively address the question of how culvert placement impacts sediment delivery to stream networks. For example, as WEPP:Road works with only one road segment at a time, using it to determine best culvert locations and optimal spacing of a road drainage system would require multiple simulations for each potential culvert movement. A typical investigation involving many case scenarios, with culverts placed at various locations, requires repeated runs of the model for all road segments affected by a the presence of a cross drain, at each particular snap shot. This could become a very inefficient, time-consuming procedure especially for long roads with multiple cross-drains where sediment producing factors vary a lot. XDRAIN on the other hand, can only be used for roads with uniform conditions, being limited to a constant road grade and considering culverts to be uniformly spaced. Users cannot tell which one of the culverts has more potential to deliver sediment and it is impossible to know what placement would possibly reduce sediment impact to streams. SEDMODL automatically identifies and displays road segments that are probable to deliver sediment based on, among other factors, a GIS layer of known culvert locations. If trying to find the best possible locations for existing culverts or revise the number of culverts to reduce sediment impacts, the culvert layer has to be modified manually and the model rerun. This process, as in the WEPP case, can become inefficient if repeated many times.

There is presently no methodology that allows a road engineer to quantify the effects of varying culvert spacing or selecting specific culvert locations, on sediment delivery to streams. The absence of such a design tool makes it more difficult to take culvert spacing into account as an effective solution for reducing sediment impacts from forest roads.

Goals and Objectives

Concepts

The Cut-Off Culvert
In order to analyze cross drain systems design and minimization of sediment delivery from forest roads we introduce the term Cut-Off Culvert. The cut-off culvert is any culvert that directs water across a vegetated hillslope. This differs from standard drainage culverts that divert water into a stream channel. Typically cutoff culverts are placed solely to reduce ditch erosion and maintain ditch flow capacity. Concurrently they can also be used to reduce the direct delivery of sediment-laden water from the road’s ditch by diverting it onto the forest floor where a major part of the sediment will be filtered out and retained prior to entering the stream (Section: Cross Drain Systems and Sediment Reduction). From the sediment reduction standpoint, the cut-off culvert’s location is crucial to the well functioning of a cross drain system. Figure 6 shows the effects of placing a cut-off culvert at different locations along a road segment.

The volumes of overland delivery (OD) from culvert C after filtering and direct delivery from ditch (DD) are represented by the two colored areas. The total sediment delivered is given by the summation of these two areas and fluctuates with culvert placement. Assume that the amount of sediment produced by the road prism is constant in all three cases, the side slope has uniform filtering capabilities and the lateral filtering is proportional with the distance the sediment must travel downhill this side slope. It can be noticed that by placing culvert C closer to the stream intersection increases its lateral delivery potential while it decreases the direct delivery from ditch (Figure 6b). Conversely moving culvert C away from the intersection gives more filtering power but also leaves more contributing area for direct delivery (Figure 6c). The question of ideal placement of a cut-off culvert becomes a question of maximizing the filtered sediment (ND) at the expense of direct delivery (DD) and overland delivery (OD).

Minimization of Sediment Delivery for a Simple Case
To explore the optimal location of a single cut-off culvert a simplified case scenario was built (Figure 7). A straight 150m long road segment, of constant grade, crossing a uniform slope hillside intersects a stream at an angle α . The sediment generating factors along the road segment and the filtering characteristics of the side slope are invariable. A cut-off culvert is placed at a random location along the road alignment. A direct delivery distance through the ditch (d) and a sediment- filtering potential characterize every possible location.

The water dispersed by the culvert in question usually travels towards the stream on a sinuous path following the local topography (flow distance). For simplification purposes assume that the side slope’s topography is uniformly flat and this flow path is identical to a straight perpendicular to the stream (Euclidian distance). The generic term: overland distance (o) refers here to the Euclidian distance between culvert and stream (Figure 7).

The sediment filtering potential expressed here as the proportion of total available sediment delivered to stream (F), varies with the overland distance (o) as well as other site-specific factors like: soil types, local gradient and vegetation cover. Because the overland distance (o) is in this case proportional with the direct distance (d) it can be stated that the filtering potential (F) is also proportional with the direct distance (d). By expressing F as a function of d the total sediment delivered to the stream can be budgeted with Equation 1:

The optimal location for cut-off culvert C can be obtained by solving differential Equation 2 for d. It can be noticed that the optimal location of C is not dependent on the sediment production rate K but only on the road-stream geometry and the local filtering characteristics.

In actuality, absolute deterministic models of side slope filtering potential to take into account every combination of factors encountered on forested terrain are not currently available. As the fraction of sediment delivered F cannot easily be determined a completely analytical approach to culvert optimization is not feasible.
Various researchers have approached the problem on empirical bases, using statistical inference on their field measurements in order to model sediment delivery. Equation 3 describes the volume of sediment deposition as a function of the travel distance for particular conditions encountered in the Idaho Batholith (Ketcheson and Megahan 1996).

By plugging equation 4 into the sediment budget equation the total sediment delivered for our case scenario can be plotted as a function of the culvert location (Figure 8). A family of curves is shown for various sediment production rates. A minimum delivery is obtained with culvert C at 50 meters away from the stream intersection all across these different sediment production regimes. It is important to note that empirically derived equation 4 cannot be universally applied. However, assuming that in most cases overland sediment delivery follows analogous invert exponential distributions (Burroughs and King 1989) it is possible to effectively approximate near optimal locations for culverts.

Similar approaches have already been incorporated into some more advanced sediment modeling programs. Therefore in order to further investigate the optimal placing of a cut-off culvert, the F.S. WEPP:Road interface was called upon to provide the modeling for the sediment production and delivery for the proposed case scenario. An experiment was conducted where culvert C from the set-up in Figure 7 was incrementally moved along the road segment. The sediment impacts associated with each of these locations were quantified by WEPP simulations run at the Forest Service web site. The sediment production parameters were: 4m wide, insloped bare ditch road, 5m long fill slope at 50% gradient and silt loam soils. Olympia Station described the local climate. The forested buffer had a uniform 25% slope. All simulations were performed for a 1-year period. Multiple runs were completed for various stream crossing angles α . Figure 9 presents a graph of the results of this experiment. The total sediment delivered exhibits a similar behavior to the Ketcheson and Megahan approach presented in Figure 8. It follows a right skewed distribution with a minimum in the first 1/3 of the road segment. The optimal culvert location for this particular case is located 50-60 m away from the stream crossing. The geometry of the road-stream intersection has a major impact on the filtering distance and implicitly affects the optimal cross drain location.

The total sediment delivered was recorded again using WEPP:Road. All WEPP sediment-modeling parameters set in the previous experiment (one culvert scenario) were kept unchanged. The stream intersection angle ? was set at 45?. Starting with the cut-off Culvert1 in the optimal position previously determined at 60 m from the stream, the total sediment delivered was computed with Culvert 1 and Culvert 2 at several different locations. Due to the high number of possible combinations of 10m increments over a 150 m road segment only a few were fully explored. Table 1 present the results of this experiment. The sediment delivery to stream S was reduced by as much as 58 % with the cross drains at 30 and 90 m away from the stream crossing as compared to the case of a single cut-off culvert in optimal location. Other possible combinations may further improve this result. It can be inferred that the addition of a third culvert at a key location could decrease sedimentation even more.

Table 1:

: Reduction of total sediment delivered (overland and ditch) with 2 culverts; road is 150m long, 4 m wide; stream crossing angle is 45 degrees. Configuration 12, with Culvert 1 at 30 m and Culvert 2 at 90 m from stream crossing delivers the least sediment.

Config.No.	Culvert 1 location	Culvert 2 location	Sediment Delivered	Sediment Reduction
Config.No.	(m)	(m)	(kg / yr)	(%)
0	60	-	2661	0
1	70	100	1471	45
2	70	110	1497	44
3	60	100	1284	52
4	60	110	1349	49
5	60	90	1290	51
6	50	100	1217	54
7	50	90	1227	54
8	40	100	1125	58
9	40	90	1140	57
10	40	80	1219	54
11	30	100	1165	56
12	30	90	1112	58
13	30	80	1132	57

Full optimization of cross drain systems becomes more complex with the addition of each new culvert but the importance of the cut-off culverts closest to the stream crossing cannot be overstated. In reality, departures from the simple triangular geometry assumed here, further complicates this analysis. As road alignments and stream paths curve and weave randomly, the road’s proximity to the stream and implicitly the side slope’s filtering potential become more difficult to model analytically (Figure 11). The flow path of the water dispersed by the cross drain follows the local micro-topography and can be significantly different from the straight line distance between culvert and valley bottom. Moreover sediment generating factors, vegetation and filtering characteristics vary along the road alignment and side slope respectively adding further complexity to sediment delivery modeling.

In addition to their stream crossings areas, roads are also susceptible to sediment delivery in all cases where they run parallel, in close proximity to streams (e.g. valley bottom roads). Whenever the volume of sediment diverted onto the side slope exceeds the local filtering potential, sediment delivery will occur. The process of finding best culvert locations in these situations follows similar analysis principles with the evident omission of the direct delivery from ditch component.

Computer Modeling of Cross Drain Systems

Culvert Location Analysis for Design Purposes
Optimizing cross drain culvert spacing takes into account a composite sum of factors including terrain information, road layout, and existing culvert locations making it difficult to be performed by hand. As automated optimization software packages do not currently exist designers are limited to using various sediment analysis programs in order to estimate the validity of their design. The investigation of best cut-off culvert location using such analysis packages can become cumbersome and time consuming for longer roads with multiple stream crossings laid across non-uniform terrain. For example applying WEPP:Road for a such project requires division of the road network into segments of uniform characteristics. A culvert placed on one of these segments further divides the segment into two parts: upstream and downstream from it, each necessitating a separate run of the model. The forested lateral buffer distance at the current culvert location has to be determined. The simulation is run for each segment and the total sediment delivered to stream is computed. If one or more culverts are relocated new divisions of the original road segments and measurements of the associated buffer length are necessary. WEPP:Road is run over the Internet. Simulating sedimentation over short periods of time is a relatively quick process taking approximately 1-2 seconds per road segment. However the program can only work with one segment at a time. The manual updating of culvert positions and dynamic segmentation of the road segments are very inefficient and represent major drawbacks of this approach. Furthermore, treating road segments as stand alone entities excludes any interaction from neighboring segments. Ditch water cannot be easily carried along different segments as encountered in reality.
The complexity of such analysis restricts its usage mostly to research projects. The lack of an appropriate, easy to use, culvert location analysis tool deters professional road designers and engineers from widely devising cross drain systems for sediment reduction purposes (Schiess, personal communication).

Interactive Culvert Placement
From the perspective of cross drain culvert design, an ideal design tool would evaluate each design step as it is proposed such that a user could easily tell the effect of his/hers decision and improve upon it. The term interactive culvert placement is used here to designate the process of designing cross drain systems by placing culverts at various locations with the immediate support of a sediment modeler to evaluate potential impacts to the nearby streams. Decision support tools that instantly quantify proposed alternatives typically assist this kind of interactive design. Figure 12 depicts the workflow of a cross drain design process using both an analysis package and a decision support tool.

The major difference between the two cases is given by the feedback the user receives during design, as a validation of a placement decision. This feedback is essential to producing a good solution as users can quickly improve upon each proposed step and get closer to an optimum cross drain setup.

CULSED- A Decision Support Tool for Cross Drain System Design
To facilitate the investigation of sedimentation reduction from insloped forest roads in realistic road design circumstances, a specialized computer program named “Culvert Locator for Sediment Reduction” (CULSED), was developed. The program is a GIS based decision support tool focused on assisting engineers during cross drain system design. Its primary function is to assess the sediment delivery at each culvert location and graphically display it on the computer screen such as users can immediately ascertain the validity of a design decision. Using this program it is possible to identify near-optimal cross drain locations by exploring various permutations along the road alignment. CULSED gives users the ability to add, move and remove cross-drain culverts, dynamically evaluating the total sediment impact to the stream network from the analyzed road system. Culverts can be represented with graduated symbols proportional to their sediment delivery (Figure 13). The question of minimizing sediment delivery is thus transposed to a question of minimizing symbols on screen. The total sediment delivered by the road is displayed at all times during the analysis stages and changes with every modification performed to the cross drain system. The program has no automated procedures for achieving absolute minimization of total delivery but provides the users with the means to compare culvert locations and identify good solutions.

CULSED is implemented as an ArcGIS extension that seamlessly integrates with the standard ArcGIS package being able to access all the existing functionality and providing a familiar interface and ease of use. Running CULSED requires at the minimum a GIS road layer, a stream layer, and a digital elevation model. Additional information such as: road surface, road age, road grade, soil and parent material and side slope vegetation cover may also be used during the sediment modeling stages. Appendix A contains a more detailed explanation of this software’s capabilities.

Modeling Cross Drain Systems with CULSED
The implementation of CULSED follows the well known Model-View-Controller paradigm. The various modules composing this architecture actively interact with each other at run time to compute and display the sediment impact related to each user action of adding, removing or moving culverts (Figure 14).

The controller module’s role is to provide user input to various operations related to the culvert analysis. It essentially drives the activities of other modules that require user interaction such as: generation of network topology, cross drain movement and description of sediment producing factors. The controller module is a part of the ArcMap user interface on which CULSED is developed.

The view module is also part of this interface and is represented by the standard ArcMap graphical output. It is at this level where the results are presented to the user for evaluation and decision support. This module receives information from the ditch model, dynamically rendering all changes made by the end user.

At the core of CULSED is a suite of three modeler modules: the Road Ditch Model, Sediment Production Model and Side Slope Filtering Model. The Road Ditch Model is a simplified representation of the road’s drain ditch, modeling the flow of water along it. This module makes the following assumptions:

The road ditch is described by the interplay of a geometric network and its associated logical network. The geometric network stores the physical location and geometry of the roads alignments, stream crossings and cross drain culverts together with their geometric connectivity. These network components are the elements that the end user sees and interacts with on the computer screen.

The logical network portrays the sediment flow within the system, storing the directionality of water movement. It is conceptually similar to a generic graph, being composed of interconnected edges and nodes. There is a direct correspondence between the elements of the logical network and the geometric network. An edge is associated here to a one dimensional stretch of road of uniform sediment producing characteristics. Edges are sources of the sediment that flows along them. A node is the abstract representation of a connection point between edges. Culverts are always node points and play the role of sinks within this architecture. Sinks capture all physical flow routed to them. From a topological perspective each edge can be characterized by parents, children, sinks and a flow direction (Figure 15). A parent is a connected ditch segment located directly upstream on the flow path, routing its sediment-laden water into the current segment. For implementation reasons the maximum number of parents is restricted to 4. Similarly a child is a connected ditch segment downstream along the flow path, receiving sediment-laden water from the current segment. Because in reality ditch water is never split onto multiple roads at intersections, the model only allows one child per segment. The presence of a sink at the end of an edge implies no children connectivity as in reality a cross drain intercepts and re-routs all the water it captures.

The Sediment Production Module used by CULSED follows the methodology outlined in the Washington Department of Natural Resources Standard Method for Conducting Watershed Analysis to compute the amount of sediment generated by the road prism (Appendix A). However, in order to provide the means for other user-implemented sediment modelers to be used with the program, this module was developed on top of a generic sediment production framework. The framework enforces a certain common functionality needed by various methodologies to work together with the other program components (Appendix B). Its main function is to calculate the sediment produced by the road prism on a segment by segment basis according to their specific sediment producing parameters. These parameters have to be associated to each road segment prior to running the analysis by the end-users unless default parameters are used.

The third module in the modeler suite, the side slope filtering module, determines the sediment filtering potential associated with each potential culvert location in road the network. It is based on the work of Ketcheson and Megahan of USDA Forest Service describing the sediment deposition on a vegetated side slope as a function of proximity to streams (Ketcheson and Megahan 1996). The module computes the proportion of sediment that can reach the stream at any given distance along a vegetated side slope (Equation 3). The proximity to the stream is based on the physical flow path distance from the culvert to the nearest stream generated from a digital elevation model.

The sediment delivered by each culvert is calculated in response to various user triggered events as the sum of the sediment produced by all contributing road segments multiplied by the filtering potential at that particular location (Equation 4).

The total sediment delivered by the entire road drainage system is thus given by sum of all contributing culverts (Equation 5).

Results of a Cross Drain Redesign Application

In order to proof the concept of the cut-off culvert and demonstrate its applicability in a real world case scenario, several road settings were examined within the context of a forest harvest plan. The goal of these experiments was to reduce the total sedimentation from an existing road network by redesigning its cross drain system to make better use of sediment dispersion and filtering potential on vegetated side slopes. As the cost of the resulting drainage system had not to exceed the original cost, the total number of cross drains could not be increased. CULSED was used to compute the amount of sediment delivered by the entire road network at each design alternative during the design process.

North Tahoma Planning Area
The North Tahoma State Forest is an area of approximately 25,000 acres of forested terrain on steep topography, with a well developed road network. It is located along the Nisqually River, near Ashford WA, contained within T15N, R6E and T14N, R6E. This site was chosen for our cross drain design experiment due to its fragmented terrain with many streams and road-stream crossings. A sufficient amount of site descriptive information was available. Digital datasets of existing cross drains, roads and high resolution digital elevation models were obtained from WA DNR. We focused our study on Reese Creek Watershed, a central part of the North Tahoma State Forest (Figure 16). The existing road network in our study area was approximately 42 miles long with 76 stream crossings. The road grade varied between 1 and 19% with an average of 6%. The majority of roads were the typical one lane forest road, 12 ft. width, surfaced with gravel and maintained in good condition. As the roads were relatively old, the side slopes were mostly re-vegetated. A cross drain system was already in place and contained 168 culverts placed according to the standard WA DNR regulation. The relatively large size of our study area and its high number of existing cross drains severely limited the user interaction with the dataset on the computer screen. Therefore the site was divided it into two smaller, more manageable parts, based on the local topography and road structure. These roughly equal subdivisions were named the East and West North Tahoma respectively.

Cross-drain Placement to Reduce Sediment Delivery from Forest Roads to Streams

TABLE OF CONTENTS