Final Program - Society for Risk Analysis

P.125 Ridley CE, Frederick RJ*; frederick.bob@epa.govUS EPAA CRITICAL ANALYSIS OF ECOLOGICAL RISK ASSESSMENT END-POINTS FOR BIOFUEL FEEDSTOCKSThe activities associated with cultivation of biofuel feedstocks and their conversionto fuel is expected to result in a complex set of inter-related environmental impacts.One approach that the US EPA has taken to focus this complex information isto construct conceptual models which schematically represent hypothesized links betweenactivities undertaken throughout the biofuels production supply chain and theresulting environmental impacts. Six feedstock models (corn, corn stover, soybean,perennial grasses, short rotation woody crops and forest biomass) were developed torepresent the range of current and future biomass sources for converting to biofuels.Initially, environmental and ecological endpoints were selected based on soft criteria,i.e. those that could be generally informative to environmental decision-makers. Subsequently,a more systematic analysis of assessment endpoints has been undertakento understand the benefits and drawbacks of particular endpoints for various assessmentapproaches, (including ecological risk assessment and comprehensive environmentallife cycle analysis), for different kinds of decision-makers, and for comparingamong biofuels and between biofuels and other forms of transportation energy. Theendpoints were evaluated by experts within and outside the agency. The criteria usedfor these evaluations include measurability, capacity for detecting changes attributableto biofuels activities, and utility for environmental decision makers. Additionally, weasked which endpoints might be used to compare the impacts of biofuel productionto impacts of other forms of transportation energy (e.g., domestically produced petroleum).Where known, data sources were identified. We present the results of thisevaluation as comparison tables. The outcome of this analysis will be used to framethe risk assessment of diverse feedstocks grown in spatially disperse regions throughoutthe United States.164W2-F.3 Rivera Mancia ME; mancia@math.mcgill.caMcGill UniversityMODELING OPERATIONAL RISK USING A BAYESIAN APPROACHTO EVTModeling of operational risk has emerged as important risk component forfinancial and insurance institutions due to the severe losses that it has produced inthe last years. One of the main problems in the study of operational risk is the availabilityof data since many operational losses are not recorded or simply because oftheir low frequency. However, some data sets have become available and have allowedto analyze operational risk. In this study, an analysis of financial institutions internalloss data is performed using the Generalized Pareto Distribution by considering theuncertainty about the threshold. The proposed model considers the form of the distributionbelow and above the threshold, combining a parametric estimation with aBayesian approximation to perform inference about the unknown parameters in bothcases. The estimation is carried out using Markov Chain Monte Carlo (MCMC) methods,allowing posterior inference. After this, it is possible to determine the minimumcapital requirements for operational risk.W3-C.2 Rooney AA, Boyles AL, Harris SF, Thayer KA; andrew.rooney@nih.govOffice of Health Assessment and Translation, NIEHS, SRA International, Inc.SOFTWARE FOR DISPLAYING TOXICOLOGICAL DATA AT NTP:THE EXPOSURE RESPONSE ARRAY AND FOREST PLOT VIEWERPROGRAMSThe NTP Office of Health Assessment and Translation utilizes a variety ofanimal and human data to inform its conclusions on health risks from toxicants.Thoughtful and deliberative integration of health effects data is important both todata interpretation and a clear presentation of conclusions. Visualization of thesedata in a graphical format can expedite assessment of patterns across studies andmake the data integration process more efficient. This requires flexible software thatincludes sufficient text detail with graphed results and has a user-friendly interface.We developed two Java-based programs that fulfill these requirements. Exposure ResponseArray Viewer was designed to create figures that summarize and display healtheffects data from experimental animal toxicology studies in a form that is useful tocompare effect levels across variables of interest (e.g. exposure route or duration).The program graphically presents effect levels in the form of LOAELS and NO-AELS (or BMDLs) and the dose range tested along with text describing study details(e.g. species, strain, or sex). Forest Plot Viewer was designed to graphically displayepidemiology study results such as OR or βs with confidence intervals alongside textdescribing study-population specifics and key study details. By displaying data in agraphical output as well as presenting columns of text describing key findings, theprograms are designed to display data that cannot be graphed together. Clear presentationof these critical factors facilitates data interpretation and comparisons acrossstudies. One of the main advantages of these programs is the capacity to display upto 15 columns of text as well as use data columns that are not displayed to format,group, or filter displayed effects by relevant study details. Forest Plot Viewer and ExposureResponse Array are freely available on the NTP website.W4-D.1 Rose S, Jaramillo P, Small M, Grossmann I, Apt J; srose@cmu.eduCarnegie Mellon UniversityQUANTIFYING THE HURRICANE RISK TO OFFSHORE WIND TUR-BINESThe US Department of Energy has estimated that if the US is to generate 20%of its electricity from wind, over 50 GW will be required from wind turbines in shallowwaters off the East coast and Gulf Coast. Many of these will be at risk of damagefrom hurricanes. Turbine tower buckling has been observed in typhoons, but no off-