Watershed and Hillslope Modeling

Swift Creek watershed and Swift Creek landslide, Whatcom County, WA

Graduate students and I have been modeling the impact landscape changes and climate change on streamflow and mass wasting in watersheds in the Pacific Northwest using the Distributed-Hydrologic-Soils-Vegetation-Model ( DHSVM) developed at the University of Washington and Pacific Northwest National Laboratory. The model simulates a water and energy balance at the pixel scale of a digital elevation model (DEM). It has been applied predominantly to mountainous watersheds in the Pacific Northwest. Our studies are summarized below.

Prediction of sediment yield from Swift Creek Landslide using the Distributed-Hydrology-Vegetation-Model: Curtis Clement MS Thesis topic (ongoing)

Swift Creek landslide is an active landslide in Northwest Washington composed of naturally occurring asbestos. Although the slide has not exhibited catastrophic movement, it does produce large quantities of sediment to agricultural land and residential areas which poses a human health risk. This sediment fills channels that frequently have to be dredged to prevent flooding and road damage. In previous years, the material was used in a variety of ways including fill dirt in newly constructed residential areas. Studies performed by the Environmental Protection Agency (EPA) show that the asbestos fibers found amongst the dredged deposits do pose health risks to those who live, work, and play on or near the material. Because of potential health risks, Whatcom County faces a challenge disposing of the material. Knowledge of what parameters impact the sediment flux being eroded from the slide and an accurate quantification of the sediment discharge is important when forming remediation plans. Monitoring Swift Creek has been difficult in the past due to the flashy nature of the stream. As a result, historic data regarding the watershed and the landslide are sparse in quantity and temporal consistency. Monitoring methods used in this study include turbidity threshold sampling (TTS), physical water samples, periodic discharge measurements, and stage data; each with varying degrees of success. These data were used to calibrate a numerical model developed by researchers at the Pacific Northwest Laboratory and the University of Washington known as the distributed-hydrology-soil-vegetation model (DHSVM) and its accompanying sediment module. The calibration process and numerical experiments provide insight regarding what hydrologic parameters and basin characteristics are having the greatest impact on surface erosion and mass wasting from the landslide.

Preliminary results indicate that the relationship between discharge and suspended sediment concentration must be analyzed as individual events rather than applying a common rating curve throughout the year. There are three common patterns: a paired increase in turbidity and discharge, an increase in discharge without an increase in turbidity, and an increase in turbidity without an increase in discharge. These patterns could be controlled by the level of precipitation, snow melt, and stability of the landslide toe; as well as erroneous readings from the sensors.

Predicting Slope Failure in the Jones Creek Watershed, Acme, WA, using the Distributed-Hydrology-Soil-Vegetation Model: Brandon Brayfield MS Thesis topic (ongoing)

Mountain watersheds in the Pacific Northwest are particularly susceptible to shallow landslides and debris flows during periods of intense precipitation. The Jones Creek watershed near Acme, WA, is a 6.7 sq-km basin that hosts several active landslides, the largest of which is the 1600 sq-meter Darrington landslide. One suspected triggering mechanisms for debris flows in the basin is landslide dam outburst flooding due to shallow mass wasting events on the Darrington landslide. There are approximately 100 buildings constructed on a 0.75 sq-km alluvial fan deposited by debris flows sourced in the watershed. Predicting the occurrence of mass-wasting events as they relate to the duration and intensity of antecedent precipitation conditions is important for land-use planning and emergency preparedness in the surrounding Acme community. We use the Distributed-Hydrology-Soil-Vegetation Model (DSHVM), coupled with an infinite-slope failure model to determine the probability of shallow mass-wasting events for a variety of hypothetical precipitation scenarios. The DHSVM uses meteorological data, paired with spatially distributed soil and land cover data, to simulate a water and energy balance at the pixel scale of a digital elevation model. The infinite-slope model is dependent on the DHSVM-simulated hydrology and uses a stochastic approach to predict the probability of slope failure on a cell-by-cell basis.

Stream flow measurements taken frequently during the 2011-2012 winter season are used to calibrate the DHSVM hydrology simulations, with favorable results. Ongoing research includes calibrating the infinite slope failure model to historical mass wasting events in the basin, and a sensitivity analysis of the soil mechanical properties that control slope stability (soil cohesion, angle of internal friction, root cohesion, and vegetation surcharge). We will use the calibrated infinite slope failure model to evaluate a precipitation duration-intensity threshold for the initiation of mass wasting events in the basin (preliminary results).

Previous Studies