Warning Response and Risk Behavior in the 3 May 1999 Oklahoma City Long Track Violent Tornado

On 3 May 1999 a long-track violent tornado (hereafter referred to as the 'OKC Tornado') struck Oklahoma City and parts of its southwestern and eastern suburbs killing 40 of the 44 Oklahoma tornado victims of an outbreak of some 70 tornadoes. The OKC Tornado was on the ground for approximately 64 km (40 mi) and nearly one and a half hours. It destroyed nearly 10,000 structures and caused 700 injuries, with total damages in excess of one billion dollars. Warnings for this storm were exceptional in lead-time, accuracy, media coverage, and public reception. In the days and weeks following the disaster, field interviews were conducted and data collected from more than 300 persons living or working in or within 1.6 km (1 mi) of the damage path regarding their actions and the actions of those in their company. These data include respondents representing all manner of impacts from total to no destruction, the injured and uninjured, and locations where fatalities occurred. The purpose of the field operations were to collect geographical, demographical, behavioral, and attitudinal information from a large and statistically significant population impacted by the tornado. Data were then analyzed for significant trends in warning access, source(s), lead-time, and compliance, as well as for respondent actions, location, structure and shelter, and views on tornado risk and on the performance of the warning system for this event. The goal of this work is to complement rather than duplicate the various other research projects on the OKC Tornado that deal with structural engineering and damage or disaster epidemiology by attempting to characterize the warning landscape. Similarities and differences between the emerging findings from the OKC Tornado are compared to other warning research from relevant historical events, with an emphasis on data from our parallel work in the 1998 Birmingham, Alabama tornado, which was also violent and killed 32 people. Some unusual findings and speculations on their implications are also included on such things as the extraordinary media coverage, the role of multiple warning sources, the effectiveness of various NWS products (such as the Tornado Watch), shelter availability, and risk perception as indicators of warning response, action, and success.