This investigation was undertaken to assess the sensitivity of the Hazus-MH (v2.0) earthquake model to model parameters and to guide the selection of these parameters for realistic earthquake-loss assessment in the central USA. To accomplish these goals, we performed several sensitivity analyses and a validation assessment using earthquake damage surveys from the 2008 M5.2 Mt. Carmel, Illinois earthquake. We evaluated the sensitivity of the Hazus-MH earthquake model to the selection of seismic hazard data, attenuation function, soils data, liquefaction data, and structural fragility curves. These sensitivity analyses revealed that earthquake damage, loss, and casualty estimates are most sensitive to the seismic hazard data and selection of the attenuation function. The selection of the seismic hazard data and attenuation function varied earthquake damages and capital-stock losses by ±68 % and casualty estimates by ±84 %. The validation assessment revealed that Hazus-MH overpredicted observed damages by 68–221 % depending on the model parameters employed. The model run using region-specific soils, liquefaction, and structure fragility curves produced the most realistic damage estimate (within 68 % of actual damages). Damage estimates using default Hazus-MH parameters were overpredicted by 155 %. The uncertainties identified here are similar to uncertainties recognized in other Hazus-MH validation assessments. Despite uncertainties in Hazus-MH earthquake-loss estimates, such estimates are still useful for planning and response so long as the limitations of the results are properly conveyed to planners, decision makers, emergency responders, and the public. Copyright Springer Science+Business Media B.V. 2012
