Effects of forest roads on flood flows in the Deschutes River, Washington

The effects of forest roads on peak flows were examined through a combination of field data collection and modelling in the extensively logged 149 km2 catchment of the Deschutes River, Washington, USA. Based on a field survey, the connectivity of culverts to the channel network was related primarily to hillslope curvature and distance to the natural stream channel. Culvert crest stage recorders operated during winters 1996/97 and 1997/98 demonstrated that higher flows occurred in ditches draining clearcuts compared to forested areas. Contrary to expectation, road cutslope height did not seem to affect culvert peak runoff. A distributed hydrologic model was used to evaluate road effects on peak flows in nine subcatchments (2·2 to 21 km2) of the Deschutes River as well as the Deschutes main stem. The model-predicted increases in the mean annual flood due to forest roads alone ranged from 2·2 to 9·5 per cent, and from 2·9 to 12·2 per cent for the 10 year event. These increases are roughly equivalent to slightly smaller than those predicted for harvest effects alone. Simulated road effects on peak flows were independent of forest harvest state. However, at the hillslope scale, modelled as well as field-monitored road ditch response was dependent on harvest state. Modelled road effects generally increased with flood return period, while vegetation effects decreased.