Targeted alerting in early warning systems

Early warning is one of the most effective measures for better disaster preparation and mitigation. Effective warning requires constant monitoring and precise detection of possible hazards, on the one side. On the other side, the information about a hazard has to be distributed to the people at risk for enabling appropriate response. In the recent decade, we witnessed worldwide an increasing effort to provide a better protection against natural, technical, biological and humanitarian hazards through the provision of Early Warning Systems (EWS). Existing IT research and development in the field of EWS has mainly concentrated on hazard detection and dissemination technologies on the global and national level. These approaches have their limits in regards to targeted and efficient warnings for individuals on the regional level (e. g., for urban areas, communities, and industries). Previous research indicates that the effectiveness of disaster response and damage mitigation depends not only on the accuracy and the earliness of warnings but also strongly on the target orientation and individual advice in the warning content. The research in this thesis concentrates on the information technologies in the domain of regional and local EWS that can provide this enhanced target orientation of warnings to individuals and user groups. The overall challenge is to find interoperable and flexible models and architectures for a new layer of local alerting systems that provide an effective dissemination of warnings in terms of coverage and adaptation to the needs of the recipients. It is expected that the solution to this problem will provide a significant contribution to target-orientation, reliability, accuracy, and cost-effectiveness of early warning infrastructures in the future. This thesis investigates the foundations, models, and reference architectures for EWS solutions that provide targeted alerting and information delivery to individuals. Within this research a model for situation-based alerting is elaborated that can increase the effectiveness of warnings significantly. The model is incorporated in an interoperable and flexible architecture for EWS based on paradigms of event notification and context-aware systems. Further, an evaluation model for measuring the effectiveness of targeted alerting and its practical application in a field tests is presented. The contribution of this thesis to existing research in the field of EWS is - besides a first systematic structuring of the domain from a computer science perspective in literature - the elaboration of a general model, methodology, and reference architecture for the implementation of targeted alerting in EWS