Editor's Note: I'm preoccupied with this project lately. The problems we've chosen are facinating and massively interdisciplinary, like many in watershed management. We are having an unabashed intellectual ball with it. I am especially grateful to my team for getting me off my 'know-it-already' butt and into a learning mode that I 've not experienced since college. A large measure of our enthusiasm flows from the mentoring of Dr. Bill Trush, whose imagination and ability to see to the heart of problems and solutions is a continuing inspiration.

The engineering staff of the Forest Service Washington Office recently organized a water­p;roads interaction program, to be conducted primarily by the San Dimas Technical Development Center. One component of the program is an investigation of watershed-scale road stream crossing risk assessment. Early on, we realized that little was available in the way of existing methods, consistent terminology, or systems for accessing knowledge and techniques. We have found that some managers are doing informal risk assessments on various scales. However, their procedures and results are largely unpublished. Thus, our first approach is to determine the "state-of-the-practice" and "state-of-the-need" for new analytical tools. To do this, we have mailed aquestionnaire to persons who regularly deal with road stream crossings (see end of article). The product of the crossing assessment project will be the development of a set of methods for stream crossing risk assessment.

Past attempts at stream crossing risk analyses

Our recent in-house attempts at stream crossing risk assessment began with stream crossing inventories for entire watersheds (20-200 square miles). The data were organized into spreadsheets where each site has a series of attributes. From that, a variety of queries were conducted to see what types of analysis would be most informative. For example, in the Pilot Creek watershed, we learned that 57% of the stream crossings had diversion potential (Figure 1). That is, if the stream overtops the crossing fill, water will abandon its channel and flow down the road or ditchline. Similar results were found in other watersheds as well. This provides an initial site priority screen. However, with just this data we were unable to answer, "Which pipe(s) should we upgrade or remove?" This requires an assessment of the likelihood of overtopping. To address this question we calculated the discharge necessary to overtop the crossing fill. This required assembling a hydrologic data set, such as drainage area and precipitation depth-duration-frequency. For many of the smaller streams however, drainage area could not reasonably be defined on 7.5-minute topographic maps. Therefore, about half the crossings in the watershed were excluded from the hydrologic/hydraulic analysis. From an initial 141 stream crossings (inboard ditch-relief pipes excluded), we calculated the flood recurrence interval that would exceed the hydraulic capacity of the structure (RI*) for 79 crossings (Table 1).

Several questions arose when the data were examined. For example, how might we combine the probability of failure (implied by RI*), and the consequence of failure (implied by fill volume and diversion potential)? Similarly, how might we regard the risk of a low consequence-low RI* crossing versus a high consequence-high RI* crossing (Table 1) ? As a first cut, we concluded that crossings with diversion potential were a primary concern. Not only are they capable of producing catastrophic erosion upon diversion, but the diverted water, in many cases, would flow to an adjacent basin, increasing the risk of failures there (Figure 2). Modifying crossing to eliminate diversion potential is relatively inexpensive with the use of rolling dips. With a relatively small amount of money, many diversion potential crossings can be remediated, versus perhaps a single, costly crossing upgrade or removal. In this case, modifying the consequence of failure is much more cost-effective than modifying the probability of failure.

Current efforts

We have conceptualized a breakdown of the environmental mechanics of road stream crossings into four parts:

• Inputs: What is expected to come down the channel­p; water, sediment, wood, and fish; and up the channel­p;fish­p;and be presented to the crossing?
  
• Capacity: How will the crossing structure accommodate the inputs?
  
• Consequence: If inputs exceed capacity, what is likely to occur, such as erosion, channel changes, pulse flooding, migration barriers?
  
• Values: What beneficial uses, on-site and downstream, are susceptible to the potential consequences?

We can readily calculate RI* and potential erosional consequences (though with substantial uncertainties), but have a paucity of methods to enable us to calculate other important factors such as sediment loading, woody debris loading, fill-piping potential, time-distrubution of fish passage, or the performance of any given crossing in actually passing the combined components of peak flows. We believe that reasonable qualitative assessment of these factors is possible, and are searching for ways to do this as accurately as possible while remaining do-able at a watershed scale; and to combine such qualitiative results with the calculable results.

Defining downstream values at risk is an intensely interdisciplinary issue that must be factored into crossing assessment. We are unlikely to find a way to reasonably quantify this, but believe it should control the outcome of the assessment. Some factors are more obvious and satisfactorily defined than others. For example, the proximity of a crossing to the beneficial use is simple to define and interpret. A crossing with a moderate chance and consequence of failure that is just upstream of a domestic water supply intake might fall out as a higher priority site than a crossing with a high chance and consequence of failure that is remote from the intake. Other values at risk might be more difficult to interpret, such as that old bugaboo, the effect of fine sediment on fish habitat quality.

Combining all the components of crossing risk in a structured inventory and assessment technique is challenging. We have identified some components of the problem that must be addressed. In Pilot Creek for example, the risk of culvert plugging by debris was not evaluated, but most workers agree that this is a common mechanism of failure in wooded environments. Some assessments have been attempted where the amount of floatable wood in the channel is estimated as either high, medium, or low, with a "high" rating suggesting a high susceptibility to plugging. Recent work however has suggested the channel width and inlet configuration may be better field parameters for estimating the capacity a culvert to pass floating debris and thus its susceptibility to plugging. Larger channels are capable of transporting larger wood. Therefore, where the channel width is far in excess of the culvert diameter, the likelihood of debris lodgment is greater, and we think it might be possible to add some rigor to this relationship.

We continue to scratch our heads over the small streams where drainage area cannot be calculated, and over inboard-ditch-relief culverts. Ditch-relief culverts are small-scale features compared to most stream crossing installations. However, given their abundance, ditch-relief culverts must play an important role in sediment delivery and water yield effects. The challenge again is one of miscibility. How can we take disparate parts of the problem and fuse them into a single workable methodology that reasonably reflects reality?

We are focused on watershed scale techniques (all crossings in a watershed are evaluated): so a balance must be struck between detail and efficiency. We envision a multi-level menu of screening techniques, with a gradient of required investment, analytical rigor and potential products. Methods will stop short of site-scale design, but will provide a range of screening quality. Each level of assessment should produce a useful product­p; at least a list of priority sites with recommended treatments and estimated costs. Results of any of the assessment levels should be miscible with all other levels, such that databases can be aggregated and each level builds on the previous level's data.

Related efforts

Well under way is an annotated bibliography of water-roads interactions. This is a searchable Reference Manager® database of annotations, key words, and cross references. Currently, the bibliography includes over 250 papers, books, and documents. We expect to publish this within 8 months.

We are attempting to discover examples of useful existing examples of crossing risk assessment, and to find out what managers believe would be useful to develop. Watershed assessment methods of all sorts fall into disuse when they're too specialized, require too much expertise, or fail to produce worthwhile, usable products. It is becoming clear that many of us share similar ideas on crossing assessment. With little funding available for road maintenance, upgrade, or decommissioning, many watershed managers wrestle with defining where the hot spots are.

We are also trying to discover what experts look at and think about when they make field assessments. Everyone knows that a good, experienced field manager is likely to have the best assessment of hot spots for their area, and can probably beat any conceivable analysis that does not include their knowledge. But what kinds of information do they know and integrate in making their judgments? Can this knowledge be captured and conserved somehow? Can newcomers be taught how to gain expert discernment? What kinds of evidence of risk can be observed in the field?

When looking at a site, some physical features might clearly suggest an undersized culvert. For galvanized steel pipes, the rustline that forms in the bottom may provide a quick field assessment tool. Preliminary observations have shown rustline heights in excess of one third the pipe diameter indicate that it is hydraulically undersized. Rustline development is also a useful qualitative tool for assessing the age and remaining useful life of a culvert.

Similarly, channel width may be a useful parameter for assessing the susceptibility to failure. Culvert-diameter-to-channel-width ratio is probably a good measure and is probably part of any expert integrated field assessment. This ratio integrates inputs and capacity and might provide a single field indicator of exceedance probability. Our preliminary data (Flanagan thesis in prep.) from NW California show that culverts sized at 0.7 times the mean scoured-stream-bed width are capable of passing the 100-year design flow and 95% of fluvially transported debris greater than 30 cm long.

The success of this project lies in communicating with fellow "culvert nerds."We need to know what's out there, who's doing what, and what's needed. If you have experience and/or interest in this topic, please contact us soon. Also, we have prepared and mailed out a questionnaire to assess what the "state-of-the-practice" is in the assessment of existing stream crossings in a watershed. The questions we have posed are given below. Please take a moment to read these, and if you have something to contribute, please respond. We have also posted this questionnaire on the World Wide Web for online completion. Find it at: http://glinda.cnrs.humboldt.edu/crossings. Hitting the "submit" button will email your answers to us.

The culvert nerds can be reached at:
Six Rivers National Forest
Interagency Watershed Analysis Center 4886 Cottage Grove Ave.
McKinleyville, CA 95519-9433
707-839-6275
707-839-6272 (Fax)
DG: R05F10A

email:
furniss@watershed.org
saf1@axe.humboldt.edu
jill.ory@yale.edu

Part I. Past attempts/ ongoing methods for crossing risk assessment
Have you used or are you using a technique for stream crossing risk assessment?
If you have used or are using an assessment technique, please describe:
1) What the information is used for?
2) How much time is required for data collection and assessment?
3) What scale is used in the assessment (e.g., watershed, road network, District)?
4) What data are collected (can you provide sample field data sheet(s))?
5) What analyses are done?
6) How much interdisciplinary input is required during the process?
7) May we contact you to set up a more detailed interview? (Please give your name and phone number)
If you formerly used a technique(s)
8) Why is this method no longer used? (e.g., took too much time, no usable products, too much subjectivity, required too much expertise, etc.)
9) What are the principal mechanisms of stream crossing failure or potential failure in your area?
10) Is stream crossing failure an important impact in your area?
11) Have you attempted any stream crossing monitoring? (if so, any results yet?)

Part II. What is needed?
13) What are your information needs i.e., do you need a stream crossing database? If so, could you describe its structure?
14) Would a rigorous methodology for crossing assessment be useful to you? If so, should this be highly detailed, or a 'screening' method for rapid identification of 'hot spots' at a whole-watershed scale? Should such a method be integrated with other road-related inventory and analysis?
15) What are your other needs for risk assessment of road stream crossings ? What other technology, guidelines, instructions, might be useful to you in performing an assessment?