Extreme Events and Stochastic Processes of Sediment Yield

Erosion and sediment yield monitoring programs are used to develop time series of soil erosion and sediment yield. These databases are subsequently used for a variety of purposes including estimation of mean annual soil erosion rates and mean annual sediment yield, simulation model parameterization and calibration, design of erosion/sediment control practices and structures, and to estimate rates of landscape evolution. Analyses of data from Arizona, Iowa, Missouri, USA and from New Zealand are used to examine the importance of large storm events in determining sediment yield from plots and small watersheds. Under a wide range of climate, environmental, land use, and management factors, extreme events (large storms) dominated sediment yield. Geospatial analyses and a coupled hydrologic-sediment yield model are applied to experimental plots, small watersheds, and to over 3,000 sub-basins comprising 132,000 ha on the Yakima Training Center in Washington State, USA. Analyses of time series (realizations of stochastic processes) from observed data and from model simulations illustrated similar results. Those simulation models which lack the ability to account for the significance of large storms in time series of sediment yield, are of limited value in determining required monitoring periods, estimation of frequency distributions for event or annual sediment yields, and thus, for design of hydraulic structures based on exceedance probabilities, or for evaluating management actions against typical environmental regulatory requirements.