Large increases in runoff and erosion rates have been observed after several recent wildfires in the Colorado Front Range, but there are insufficient data to quantitatively evaluate the role of different factors on sediment production rates. The objectives of this project were to: (1) measure hillslope-scale sediment production rates from prescribed and wild fires of different ages; and (2) develop models to predict post-fire erosion rates. Since spring 2000 we have been monitoring sediment production rates from 48 sediment fences on six different fires in the central and northern Colorado Front Range.

For the Bobcat fire in the first two years after burning, sediment production rates for the sites burned at high severity averaged 8-10 Mg ha-1 yr-1. For sites burned at moderate or low severity sediment production rates averaged only 0.1-0.2 Mg ha-1 yr-1. Summer convective storms generated about 90% of the annual erosion, with most of the sediment resulting from only a few of the most intense storms. In the case of the intensively-monitored Bobcat fire, similar erosion rates were observed for the first and second summers after burning, but the rainfall erosivity in the second summer was 60% higher. Sediment production rates in the first year after burning were approximately 40 times larger in sites burned at high severity than for sites burned at moderate or low severity; this difference increased to about 140-fold in the second year. Sediment fences in swales produced three times more sediment than fences on planar hillslopes, implying that concentrated runoff and rill erosion are at least as important as rainsplash and sheetwash. Percent cover is the dominant control on sediment production rates. Data from all sites indicate that approximately 4 to 5 years are required for erosion rates to return to background levels.

Multivariate models confirm that the key controls on sediment production rates are percent bare soil, fire severity, rainfall erosivity, and hillslope position. Soil hydrophobicity, slope, and soil texture were generally not significant controls on sediment production rates. The best models yielded R2 values of around 0.8, and validation of these models against independent data from the Bobcat fire yielded R2 values of around 0.6.