Erosion at Stream Crossings: The Case for Restoration

Wildland roads are a major source of sediment in many water sheds. Although sedimentation is a natural process in many aquatic systems, large amounts can greatly impair the integrity of a water shed. For example, suspended sediments can negatively impact salmonid fisheries through direct mortality, hindering the development of eggs and larvae, disrupting natural movements and migration, and reducing food organisms (Newcombe and MacDonald 1991). Inadequate drainage structures on roads are a common cause of erosion, so if a road is no longer needed, removal of culverts through road decommissioning is recommended.

Literature Review

Several studies conducted in northern California address erosion and sedimentation from failed culverts. Best et al. (1995) examined 111 stream crossings on unpaved logging roads. They found that stream diversions at road crossings are the most important causes of fluvial erosion in the watershed. Such diversions typically occur when a culvert plugs and flow is diverted down the inboard ditch instead of breaching the road fill. Diversions are more prone to occur on insloped roads with inboard ditches than on outsloped roads. The most important factor in determining the probability of stream diversion is the gradient of the road at the point of crossings, and diversions are more common on steeper roads. There were 15 stream diversions noted in the study, and these diversions produced 64,000 metric tons of sediment. This volume was much greater than the volume of road fill in the 15 crossing prisms, and illustrates the fact that the total erosion due to problems at a stream crossing must be considered, not just the immediate problems associated with breaching the road fill at a crossing.

In addition to erosion due to stream diversions, 43 culverts plugged and caused erosion of road fill at the stream crossings. Some crossings were rebuilt and failed again, resulting in 52 total failures. These failed crossings contributed 11,200 metric tons of sediment, and had an average failure size of about 200 metric tons. In many cases the amount eroded from the crossing was greater than the amount of road fill in the prism because of incorporation of native material through bank erosion and knickpoint retreat.

Klein (1987) surveyed 24 stream crossings immediately following culvert removal and again after winter flows caused some erosion of the decommissioned crossings. Post-rehabilitation erosion was positively correlated with stream power, and inversely related to the boulder and cobble content of the stream bank materials. Average incision of the newly excavated streambed was 0.8 m3 per meter of channel length. (Average width of stream channel was not listed, nor was surface erosion measured in this study.)

Bloom (1998) surveyed 86 decommissioned instead of breaching the road fill. Diversions are more prone to occur road crossings before and after a 12-year recurrence interval storm in the Bridge Creek basin, northern California. A total of 117,500 cubic yards of fill had been excavated from these stream crossings during restoration work. In the 1997 storm, decommissioned stream crossings yielded an average of 5 cubic yards of erosion (much less than what would have eroded if the road fill had been left in place and the culvert failed).

Madej (2001) examined 207 road crossings in Redwood National Park that had been decommissioned during the previous 20 years. Bank erosion, channel incision and mass movements were measured at each road crossing (measurements did not include surface erosion or rilling). A total of 220,000 m3 of fill had been excavated from the crossings (which could be considered potential sediment input if the road prisms had not been decommissioned), and 10,500 m3 of sediment eroded from the crossings since decommissioning. The crossings eroded a median volume of 23 m3 since the crossings were excavated. A few trouble crossings produced the most sediment (20 % of the crossings produced 73% of the sediment). Because many of these crossings were removed in the early days of the restoration program, erosion was probably higher than we would expect in current day excavations with more experienced staff. Channel incision and bank erosion were the most common forms of post-treatment erosion in crossings. The volume eroded was positively correlated with stream power and the volume of road fill excavated (i.e., the more road fill that had to be excavated, the more bare slopes were exposed to erosive forces and subsequently they showed higher erosion rates). Surface treatments of the bare slopes varied, from no treatment to heavy mulching.

Without treatment, roads can eventually fail and contribute sediment to streams. Based on an inventory of 330 km of untreated roads in a northern California basin, Weaver and Hagans 1999) estimated past road-related sediment delivery to be 720 m3/km of road, and future potential sediment delivery without road treatment to be an additional 820 m3/km, for a total of 1540 m3/km. In a similar study based on 140 km of untreated roads in the Redwood Creek watershed (G. J. Bundros and B. R. Hill, unpublished data, 1997) past and potential sediment delivery from roads was reported to be 1450 m3/km of road. By removing culverts and restoring natural drainage patterns, restorationists have removed the risk of stream diversions (discussed previously under the Best (1995) study). None of the 207 excavated crossings examined in the Madej (2001) study had diversions or debris torrents related to road treatment. Although road restoration in Redwood National Park did not completely prevent sediment production from removed roads, it did substantially reduce the long-term sediment risk from abandoned roads.

Furniss and others (1998) examined the effects of the 1996 Oregon/Washington flood events on road-stream crossings in three physiographic regions. Fill erosion was found at 49% of the failed crossings, and diversion of streams occurred at half the failed crossings, resulting in erosion of the ditchline and road surface, and gullying of the sidecast road fill. In addition, 69% of the diversions left the originating watershed and delivered runoff to an adjacent watershed and stream. Cascading failures occurred where stream diversions were routed to adjacent crossing structures, causing them to fail.

Conclusion

These studies were conducted in the Pacific Northwest, but many of the results can be applied to other wildland roads. The studies show that when culverts fail (through hydraulic xceedance, plugging by woody debris or sediment, or destruction by debris torrents, for example) the amount of erosion may exceed the volume of material within the road prism overlying the culvert because of off-site effects. Also, based on many field observations, when the road prism is eroded, most commonly channel incision does not stop at the top of the culvert, but extends to the natural channel bottom. The culvert more frequently acts as a large roughness element that induces scour along the stream banks and channel bed rather than acting as channel armor to prevent further erosion. Although many problems with drainage structures are preventable through proper road design (construction of rolling dips to prevent diversions, for example) and regular maintenance (especially the use of storm patrols), culverts can still plug with debris and fail. Road decommissioning removes the threat of culvert failure. The removal of culverts during decommissioning does not prevent all erosion from the crossing site, but with adequate mulching and implementation of best management practices we have found that post-rehabilitation erosion is much less than the erosion associated with unmaintained culverts.

--- Mary Ann Madej is with the USGS-Western Ecological Research Center.

References

Best, D. W., H. M. Kelsey, D. K. Hagans, and M. Alpert. 1995. Role of Fluvial Hillslope Erosion and Road Construction in the Sediment Budget of Garrett Creek, Humboldt County, California. Chapter M in U.S. Geological Survey Professional Paper 1454. Geomorphic Processes and Aquatic Habitat in the Redwood Creek Basin, Northwestern California. K.M. Nolan, H.M. Kelsey, and D.C. Marron, eds.

Bloom, A. L. 1998. An assessment of road removal and erosion control treatment effectiveness: A comparison of 1997 storm erosion response between treated and untreated roads in Redwood Creek Basin, northwestern California. MS Thesis. Arcata, CA. Humbolt State University.

Bundros, G. J. and B. R. Hill. 1997. Unpublished data. Redwood National Park. Arcata, California.

Furniss, M. J., T. S. Ledwith, M. A. Love, B. C. McFadin and S. A. Flanagan. 1998. Response of Roadstream Crossings to Large Flood Events in Washington, Oregon, and Northern California. U.S. Forest Service San Dimas Technology and Development Center Report 9877-1806. 14 p. http://www.stream.fs.fed.us/water-road/w-r-pdf/floodeffects.pdf

Klein, R. D. 1987. Stream channel adjustments following logging road removal in Redwood National Park. Arcata CA: National Park Service; Redwood National Park Watershed Rehabilitation; Technical Report No. 23. 38 p.

Madej, M. A. 2001. Erosion and sediment delivery following removal of forest roads. Earth Surface Processes and Landforms. Vol. 26, No. 2, pp.175-190.

Newcombe, C.P., and D.D. MacDonald. 1991. Effects of suspended sediments on aquatic ecosystems. North American Journal of Fisheries Management. 11:72-82.

Weaver, W. E. and Hagans, D. K. 1999. Storm-proofing forest roads. In: Proceedings of the International Mountain Logging and 10th Pacific Northwest Skyline Symposium. Sessions, J. C. and W. C. Chungs, (eds.) Oregon State University Department of Forest Engineering and the International Union of Forestry Research Organizations. Corvallis, Oregon.