A Holistic Approach to Early Warning Systems Using an Agent-Based Model

Abstract

Developing an early warning system requires four key components: risk knowledge, hazard detection (including monitoring and forecasting), dissemination (involving decision making and warning issuance), and response (including action implementation). Early warning system (EWS) provides an integrated system to facilitate timely responses to hazards. To assess the effectiveness of an EWS, a systems-based approach that holistically captures its critical components is required. Therefore, this study used a system-based modeling tool, an agent-based model (ABM), to examine the factors influencing evacuation response in a flooding scenario. The model was tested for an area nestled within the Ba catchment in Fiji. Surveys, interviews, and previous literature underpin the development of the model. Evacuation response was examined across key social and physical factors, with the dissemination of warning information kept as the central focus. The findings indicate that timely warnings, coupled with training, substantially improve response outcomes. However, factors such as belief in the warning and flood velocity can undermine outcomes even when warnings are issued promptly. This study underscores the critical need to assess the effectiveness of EWS holistically by accounting for a range of factors, extending beyond forecast development and dissemination.