Reddi, L.N., 1988, Probabilistic analysis of groundwater levels in hillside slopes: Columbus, Ohio, Ohio State University, Ph.D. dissertation, 141 p.
A comprehensive unsaturated-saturated flow model was developed to predict porewater pressure in hillside slopes. The porewater pressures were predicted as a function of rainfall, evapotranspiration, flow in the unsaturated zone, and drainage in the saturated zone. Appropriate individual models were studied for evapotranspiration, unsaturated flow and drainage. The evapotranspiration model was chosen to require only regularly observed meteorological data. A mass balance model was developed to represent the flow process in the unsaturated zone. Drainage in the saturated zone was computed using a kinematic storage model. The flow model was applied to two field sites. The first site was the Kennel Creek watershed located on Northeast Chicagof Island, Alaska and the second site was the Perkins Creek watershed located in the Suislaw National Forest of the Oregon Coast Range. Model results showed reasonable agreement with the field observations at both the sites. In the case of the Kennel Creek watershed, the predictive accuracy of the present model was as good as that of a more sophisticated 2-d model developed by previous investigators. The stochastic nature of the flow problem was studied by identifying the uncertain parameters in the flow model. The uncertainty of groundwater predictions was obtained as a function of uncertainties in these parameters. The First Order Second Moment approach was chosen for this purpose. The stochastic model was applied to the Kennel Creek watershed site. Bayesian updating was used to study the reductions in prediction uncertainty given the input-output measurements of the groundwater system. Piezometric levels observed at Kennel Creek watershed were used to update the parameters and moisture profile at the site. Considerable reductions in variances were achieved due to updating.
Theses and Dissertations