Estimating Sediment Losses Generated from Highway Cut and Fill Slopes in the Lake Tahoe Basin
AuthorStucky, Daniel Ling
AdvisorDennett, Keith E.
StatisticsView Usage Statistics
Lake Tahoe's famed water clarity has gradually declined over the last 50 years, partially as a result of fine sediment particle (FSP, < 16 micrometers in diameter) contributions from urban stormwater. Of these urban sources, highway cut and fill slopes often generate large amounts of sediment due to their steep, highly-disturbed nature. Therefore, understanding the erosion mechanisms (rainfall-runoff and dry ravel), the magnitude of erosion rates and the particle-size distribution (PSD) of the eroded material from these highly disturbed slopes, as well as quantifying the load reductions achieved through slope stabilization practices, is critical to reducing sediment contributions to Lake Tahoe. Furthermore, accurate predictions of soil losses from these cut and fill slopes are required to establish baseline sediment loadings, assess the effectiveness of slope stabilization improvements and track the progress towards achieving Total Maximum Daily Load (TMDL) reduction milestones in the Lake Tahoe Basin. The Revised Universal Soil Loss Equation (RUSLE), Tahoe Basin Sediment Model (TBSM) and the Road Cut and Fill Slope Sediment Loading Assessment Tool (RCAT) are the most common soil erosion models used to predict sediment yields from disturbed slopes in the Lake Tahoe Basin; however, limited comparisons of the predictions from the various models to actual field data exist. The primary objectives of this research were to (1) design and construct an inexpensive rainfall simulator capable of closely replicating the kinetic energy of natural rainfall and operating over steep terrain (2) use rainfall simulation data collected from a diverse set of 25 slopes adjacent to highways in the Lake Tahoe Basin to evaluate the predictive performance of the erosion models and determine significant correlations between the physical plot characteristics and the collected runoff and erosion data; (3) provide suggestions to improve the predictive performance of the models; and (4) use field measurements of dry ravel to quantify sediment yields and develop predictive equations to estimate this erosion phenomena. The comprehensive correlation analyses of the rainfall simulation data indicated that surface cover, of all the physical characteristics of the slope site, most directly influenced the magnitude of erosion. In terms of broad comparisons, the slopes with volcanic soils (sandy loams) typically generated greater amounts of runoff and erosion than the slopes with granitic soils (sand and loamy sands) and exhibited finer particle size fractions in the bulk soil and runoff, resulting in four to ten times greater amounts of FSP soil losses for comparable slopes. The fill slopes appeared to exhibit more noticeable and less predictable variations in the measured runoff and erosion parameters, presumably due to the unique characteristics of these slopes (e.g., foreign soil material, increased soil compaction and decreased surface roughness). Using the Nash-Sutcliffe Model efficiencies (R2eff) to evaluate the predictive performance of the selected models, the R2eff for the TBSM and RCAT were negative for both the total and FSP soil loss predictions, indicating that the mean of the observed soil losses from the rainfall simulations predicted the soil losses better than the TBSM and RCAT models. Conversely, the RUSLE model performed best in predicting both total soil losses (R2eff = 0.20) and FSP soil losses (R2eff = 0.16). The RUSLE performed most accurately in predicting the largest FSP sediment yields, while the TBSM performed best in predicting the smaller FSP sediment yields. Some potential improvements to the various sediment loss models include: using the bulk soil characteristics to estimate the FSP fraction of the runoff erosion (RUSLE), incorporating a slope-length factor to increase erosion rates on longer slopes (RCAT) and refine model soil parameters using calibration techniques and the soil, runoff and erosion data collected from the rainfall simulations performed during this research (TBSM). The dry ravel collected from the field traps indicated that sediment yields are primarily slope dependent and may significantly increase when slope gradients exceed approximately 60%. Additionally, the PSD analyses revealed that the amount of fines in the bulk soil was similar to the amount in the dry ravel collected from the sediment traps (1:1 ratio), thus differing from the ratio of the FSP fraction observed during the rainfall simulation erosion and the FSP fraction of the bulk soil (1:5:1).