Identifying Sources of Eroded Sediment Using a Bayesian Framework

The area of land contributing overland and subsurface flows to a river network is known its watershed. Soil erosion that occurs within a watershed may result in the loss of rich arable land and/or the degradation of surface water quality due to the transport of excessive amounts of sediment, and associated pollutants, into the river network. Erosion management is therefore necessary for sustaining agricultural productivity as well as aquatic ecosystems. Appropriate tools are often needed to identify critical erosion sources, or “hotspots”, so that suitable economic management actions may be taken. Examples of such tools include statistical models that allow for direct quantification of uncertainty in erosion source identification, which is important in decision making. This study presents one such model formulated within a Bayesian framework. The original framework was developed by Fox and Papanicolaou (2007), who applied it successfully to a watershed in Idaho to predict the relative erosion contributions from agricultural and forest land uses in the watershed. The current study seeks to extend the capabilities of the original framework to include the prediction of the dominant erosion processes that occur within the watershed. The proposed extensions, which can accommodate prior knowledge of erosion processes in the watershed, will provide watershed managers with invaluable information that will enhance their decision making.