Sensitivity Analysis of Loss-Estimation Modeling Using Uncertainties in Earthquake Parameters

Abstract

The Federal Emergency Management Agency9s loss-estimation model, HAZUS, is a powerful tool in emergency management. Its accuracy, when used immediately following an earthquake to estimate likely damage and effects, can be partially evaluated with sensitivity analysis by varying input parameters within expected uncertainties. For earthquakes, uncertainties in location (epicenter and depth) and magnitude are major factors contributing to uncertainties in HAZUS results. In this investigation, three factors from HAZUS outputs (total economic loss, numbers of buildings with major damage, and numbers of fatalities) are compared for various permutations of epicentral location, depth, and magnitude. Given the typical uncertainties in these earthquake parameters, this analysis indicates that HAZUS estimates will generally be within a factor of five relative to likely actual values. Uncertainties in the location of epicenters, depths, and magnitude, when combined with changing population, uncertainties in building stock, and uncertainties in local effects (soil and rock types, assumptions about attenuation, basin geometry, liquefaction potential, and directivity), make loss estimates generally consistent within one order of magnitude.