Parameter Importance of an Analytical Model for Transport in the Vadose Zone

• Main Author: Bushnell, Tanner Hans
• Format: Others
• Published: BYU ScholarsArchive 2007
• Subjects: importance; sensitivity; uncertainty; RBCA; contaminant transport; parameter analysis; risk assessment; groundwater; Civil and Environmental Engineering
• Online Access: https://scholarsarchive.byu.edu/etd/1039; https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=2038&context=etd

The American Society of Testing and Materials (ASTM) has established a three tier risk-based corrective action (RBCA) program for cleaning up petroleum release sites, which is supported by the Environmental Protection Agency. RBCA programs make the cleanup of spill sites more efficient by requiring additional site information only when a more accurate risk assessment is needed. For spill sites that do not pass the first tier general assessment, a Tier 2 evaluation involving site specific information and screening level models to assess the potential risk must be conducted. Screening level models generally require site specific input parameters. To increase efficiency it would be helpful to know which parameters have large affects on model output and which parameters do not affect the model output significantly. There have been many studies focused on model sensitivity to input parameters. For an input parameter to vary there must be uncertainty about the value. This research proposes a method of including parameter uncertainty with model sensitivity to quantify the importance of a parameter, where the term importance is a combination of parameter uncertainty and sensitivity. Using the method developed in this thesis, an importance assessment was conducted on an analytical model for vadose zone transport. It was found that for sites posing high risk, with large spill volumes and shallow water table depths, the input parameters of water table depth and spill volume were the most important. The input parameters of precipitation and contaminant biodegradation half-life showed high importance in lower risk situations; when the water table was deep. A comparison of sensitivity analysis to importance showed differences in their results. The sensitivity analysis identified those parameters that the model was sensitive to, while the importance assessment identified the parameters that were sensitive and whose range of uncertainty was large enough to affect model output values. This information could be used for resource allocation decisions when acquiring additional site specific information.