Uncertainty Quantification of Soil-Water Balance Predictions, Using Fuzzy-Probabilistic and Maximum Likelihood Bayesian Averaging

Friday, November 8, 2013: 1:30 p.m.
Boris Faybishenko , Lawrence Berkeley National Laboratory, Berkeley, CA

Multiple semi-empirical formulae have been developed for soil-water balance calculations of potential evapotranspiration (PET), actual evapotranspiration (ET), and infiltration (I), using meteorological data and hydraulic parameters. Selection of one these models and corroboration with field observations of infiltration and evapotranspiration is a challenging problem. In this presentation, I will discuss several types of uncertainties affecting soil-water balance calculations, and will present the results of Monte Carlo and fuzzy-probabilistic simulations of PET, ET, and infiltration (I), based on meteorological data for the Hanford and Savannah River sites. Then, will provide a comparison of using a fuzzy degree of similarity index (FDSI) and Maximum Likelihood Bayesian Model Averaging (MLBA) for the selection of a subset of models to express the uncertainty of calculations of PET, ET, and I for each site.

Boris Faybishenko, Lawrence Berkeley National Laboratory, Berkeley, CA
Boris Faybishenko is a staff scientist of the Earth Sciences Division of Lawrence Berkeley National Laboratory, Berkeley. He conducted field site characterization, monitoring, and modeling of coupled water and gas flow and chemical transport (organics, radionuclides, and metals) in unsaturated (vadose zone) and saturated (groundwater) soils and fractured rock at organically and radioactively contaminated sites. His research interests include the development of innovative in situ site characterization and monitoring tools, conceptual and numerical modeling, using methods of nonlinear dynamics, chaos theory, fuzzy systems, and stochastic modeling of flow and transport in the vadose zone and groundwater, as well as time-series climatic analysis.