Natural Resource Ecology Laboratory

Current Research



  • Roads - sedimentation, sanding

Sediment Production from Roads in the Central Sierra Nevada (Coe, DB and MacDonald, LH)


  • Urbanization and Development

 Measuring and Modeling the Effects of Development on Sediment Production and Delivery, St. John, US Virgin Islands. (Ramos-Scharron, CE and MacDonald, LH)



Effectiveness of BAER treatments in the Bobcat, Hayman, and Schoonover Fires, Colorado Front Range.
Daniella T.M. Rough and Lee H. MacDonald

Post-fire flooding and erosion are critical environmental concerns following high-severity wildfires. To reduce the risk of catastrophic events burned area emergency rehabilitation (BAER) treatments are often applied, yet few studies have quantified their efficacy. The primary objective of this study is to evaluate the effectiveness of BAER treatments in reducing post-fire erosion from the Bobcat, Hayman, and Schoonover Fires in the Colorado Front Range. The treatments being monitored include seeding, contour felling, mulching, scarification with seeding, and polyacrylamide (PAM).

Aerial and ground seeding did not significantly reduce erosion during any of the four years following the Bobcat Fire, but dry mulch treatments reduced sediment yields by more than 90% in 2001, 2002, and 2003. Contour felling treatments showed a moderate reduction in erosion rates, with the greatest treatment effect measured after new treatments due to the additional storage capacity of the logs. A paired-swale design applied after the 2002 Hayman and Schoonover Fires has enabled a more sensitive comparison between treated and control swales. Dry mulch and aerially-applied hydromulch reduced sediment yields by more than 95% in 2002 and 2003. However, ground-applied hydromulch did not significantly reduce sediment yields in 2003. Eleven kg ha-1 of PAM applied in ammonium sulfate solution reduced sediment yields by 66% in 2002, but did not significantly reduce sediment yields in 2003 on either new or old treated swales. Neither the seeding and scarifying treatments in 2002 and 2003, nor the dry PAM treatments (5.6 kg ha-1) in 2002 significantly reduced sediment yields.

The data from these three fires suggest that treatment effectiveness varies with time since application, storm intensity, and amount of ground cover. Percent ground cover is the most influential factor on treatment effectiveness, and this is attributed to its ability to reduce rainsplash and rilling induced erosion. This explains why mulch treatments have been most effective in reducing erosion, whereas treatments that do not immediately increase ground cover such as seeding, contour felling, and PAM, have showed limited effectiveness.

Hillslope Erosion Processes after High Severity Wildfires, Colorado Front Range
Pietraszek, J H and MacDonald, L H

High severity forest fires can increase soil erosion rates by two or more orders of magnitude, with potentially severe consequences to downstream water resources. The objectives of this study were to contrast hillslope and rill erosion in severely burned areas, and quantify the effects of different controlling variables. Sediment yields are being monitored from 34 sites that burned at high severity in early summer 2002 in the central Colorado Front Range. Twenty-three of these sites are in topographically convergent swales with contributing areas of 100 to 7000 m$^{2}$. The other 11 sites are on planar hillslopes with contributing areas ranging from 30 to 150 m$^{2}$. Rill formation is being measured at 117 cross-sections in 14 of the 23 swales. Tipping bucket rain gages are measuring rainfall intensities within 0.5 km of each site. For individual storms the sediment yields per unit area vary by an order of magnitude, and this is largely due to the spatial variability in the amounts and intensity of rainfall. In the swales storm sediment yields can exceed 1.0 kg m$^{-2}$, and on average the swales produce approximately 2-3 times more sediment per unit area than the planar hillslopes. The greater topographic convergence in the swales results in more concentrated overland flow, and this is evidenced by the rapid rill incision in the swale axes. The cross-sectional area of these rills ranges up to 0.05 m$^{2}$, and the cross-sectional area of these rills has been increasing in the second summer after burning despite the gradual increase in ground cover. Data from a storm of 8 mm shows that rill erosion in the swale axes can account for nearly 80% of the measured sediment yields. A better understanding of rill erosion is needed to accurately predict post-fire erosion rates and design effective rehabilitation treatments.

Measuring and Modeling the Effects of Development on Sediment Production and Delivery, St. John, U.S. Virgin Islands
Ramos-Scharron, C E and MacDonald, L H

Increases in sediment delivery rates into tropical marine environments may pose a serious threat to nearshore coral reef communities. The island of St. John in the U.S. Virgin Islands is at particular risk because the areas outside of the Virgin Islands National Park are being subjected to rapid development, 80% of its land surface has slopes greater than 30%, and the average annual erosivity is 13,500 MJ mm ha$^{-1}$ hr$^{-1}$. The objectives of this study were to: (1) measure erosion rates from a variety of anthropogenic and natural sources; and (2) develop a GIS-based model to estimate basin-scale increases in sediment delivery. Sediment production rates were measured from unsurfaced roads, road cutslopes, undisturbed hillslopes, streambanks, and treethrow. Unsurfaced roads had the highest erosion rates with values of up to 38 kg m$^{-2}$ yr$^{-1}$, while road cutslopes generated only about 0.8 kg m$^{-2}$ yr$^{-1}$. Since undisturbed areas generated only 0.001 kg m$^{-2}$ yr$^{-1}$, unsurfaced roads can increase hillslope-scale sediment production rates by up to four orders of magnitude. Sediment production rates from erodible streambanks were estimated to be nearly 13 kg m$^{-2}$ yr$^{-1}$, while uprooting of trees along stream margins was estimated to deliver 0.17 kg yr$^{-1}$ of sediment per meter of stream length. The St. John Erosion model (STJ-EROS) uses a series of empirical sediment production models and sediment delivery ratios to estimate watershed-scale sediment yields in a GIS-based system. Using this model, the estimated sediment yields for the highly-developed 6.0 km$^{2}$ Fish Bay basin was 41 tons km$^{-2}$ yr$^{-1}$, while the estimated sediment yield for the less-developed 4.3 km$^{2}$ Lameshur Bay basin was only 11 tons km$^{-2}$ yr$^{-1}$. The model simulations suggest that current sediment yields are seven and four times above undisturbed conditions in the Fish Bay and Lameshur Bay basins, respectively. The predicted sediment yields are consistent with measured sediment yields and sedimentation rates in both Fish Bay and Lameshur Bay. In both cases actively used unsurfaced roads were identified as the dominant source of sediment. Cutslopes and streambanks played a secondary role in total sediment yields, while undisturbed hillslopes and treethrow contributed only minimal amounts.

Sediment Production from Forest Roads in the Central Sierra Nevada, California
Coe, D B and MacDonald, L H

Unpaved roads are a major source of sediment in many forested landscapes. For the Sierra Nevada of California there are few data on road erosion rates or the factors that control road sediment production. This information is urgently needed to establish TMDLs and evaluate cumulative effects. Annual sediment production from 139 road segments was measured over three wet seasons using sediment fences. The first wet season had 1290 mm of precipitation as compared to 890 and 1057 mm in the second and third wet seasons, respectively. More importantly, the second and third wet seasons were dominated by colder storms with much of the precipitation falling as snow. Hence the annual erosivity (EA) in the first wet season was 847 MJ mm ha-1 hr-1, which was nearly twice the values from the second and third wet seasons. The mean sediment production rate from native surface roads in the first wet season was 0.64 kg m-2 as compared to 0.17 and 0.20 kg m-2 in the second and third wet seasons, respectively. Rocked roads produced only 0.01-0.03 kg m-2 yr-1. Recently-graded native surface roads produced twice as much sediment per unit storm energy as ungraded roads. The product of road surface area and road slope explained 40% of the variability in sediment production from native surface roads. Adding EA and a dummy variable for grading increased the overall R2 to 0.54. Road segments draining soils with rock outcrops or particularly shallow soils produced more than twice as much sediment as roads on or adjacent to all other soil types, and this difference is attributed to the higher runoff rates associated with the interception of shallow subsurface stormflow. The much lower erosion rates in the second and third wet seasons can be attributed to the lower annual erosivities and the role of the snow cover in reducing rainsplash and overland flow velocities, and possibly attenuating high-intensity bursts of rainfall. The results emphasize the variability of road erosion rates, the sensitivity of road erosion rates to site and climatic factors, and the difficulty of accurately predicting sediment production rates.

Factors Associated With Post-fire Sediment Yields From Hillslope Plots in the Colorado Front Range
Benavides-Solorio, J and MacDonald, L H

In recent years there has been a large increase in the number and size of wildfires in the mid-elevation zones of the Colorado Front Range. High-intensity rainstorms after these fires have increased erosion rates by several orders of magnitude and severely affected downstream aquatic resources. The objective of this study was to measure sediment production rates at the hillslope scale and determine the key controlling variables. To this end 48 sediment fences have been continuously monitored in three wildfires and three prescribed fires at elevations ranging from 1670 to 3050 m. The most intensively-studied area is the Bobcat fire, which burned 43 km$^{2}$ in June 2000. Within this fire sediment production rates exceeded 10 Mg ha$^{-1}$ yr$^{-1}$ for areas burned at high severity. Prescribed fires produced substantially less sediment than the corresponding wildfires. Sediment production rates from sites burned at high severity were nearly 200 times higher than sites burned at moderate severity. Nearly all of the erosion occurred as a result of summer rainstorms rather than winter snowmelt. Sediment production rates per unit area were 2-3 times higher in swales or small drainages than from planar hillslopes. Data from the older fires indicate that sediment production rates remain elevated for sites burned at high severity for at least three years after burning. When the data from all fires were combined, 77% of the variability in sediment production rates could be explained by fire severity, percent bare soil, rainfall erosivity, soil water repellency, and soil texture. A simpler model using only percent cover and rainfall erosivity had a R$^{2}$ of 0.62. Various models were tested against an independent data set from the Bobcat fire, and this showed that fire severity, percent bare soil, and rainfall erosivity could successfully predict post-fire erosion rates. The understanding and models developed in this project can help land managers predict the likely impacts from future wildfires and assist in the design of more effective post-fire rehabilitation techniques.

Detecting Cumulative Effects on Low-gradient Streams in the Central Sierra Nevada, California
MacDonald, L H Brown, N E and Coe, D B

Cumulative watershed effects (CWEs) result from multiple activities over time and space. The assessment of CWEs in the Sierra Nevada is severely limited by the lack of field data on the effects of a given action, the lack of models to predict the effects of multiple actions at the watershed scale, and the limited data relating stream channel conditions to measured or predicted changes in sediment yields. Since 1999 we have been measuring hillslope-scale sediment production rates from roads, timber harvest, wildfires, and minimally-disturbed areas. From these and other data we are developing models to predict sediment production rates from different land uses. The next, more difficult step in developing a reliable CWE model is to compare predicted changes in catchment-scale sediment production rates to stream channel conditions. Channel conditions were measured in 28 pool-riffle reaches in the American and Cosumnes river basins. Contributing areas ranged from 2.9 to 70 km2, and the elevations of the reaches ranged from 1200 to 1800 m. The basins were selected to encompass a wide range in the amount of natural and anthropogenic disturbance. Variables used to characterize the amount of disturbance include road density, number of road crossings, modeled road sediment production, percent forest harvest, and the percent burned by wildfire. The data collected in each reach included gradient, channel dimensions, grain size, amount of large woody debris, and pool size and sediment infill. Preliminary results indicate that basins with more than 20% of the area in granitic lithologies have significantly more fine sediment coverage in pools (r2=0.35; p=0.0002). The fine sediment coverage in pools also was correlated with the amount of large woody debris. After removing the variability due to lithology and basin area, there was a significant, positive correlation between the estimated sediment production from unpaved roads and residual pool infill volume (r2=0.14; p=0.02). Other than these relationships, there appear to be no significant correlations between the measured channel characteristics and the amount of disturbance. The limited number of significant correlations may be a consequence of a record flood event in early 1997. This event may have effectively "reset" the stream channels, resulting in a lower detectability of cumulative watershed effects.