Search

Disasters vary in many characteristics, but their amount of forewarning-the amount of time remaining until the disaster strikes-is a crucial factor affecting the dissemination of emergency warnings. People can be warned by public safety officials through broadcast channels, such as commercial TV and radio, that transmit simultaneous warnings to mass audiences. In addition, however, warnings are also transmitted by peers through informal warning networks that operate through contagion from one person to another. This paper establishes an interdisciplinary agent-based model with Monte Carlo simulations to assess the relative effects of these broadcast and contagion processes in a multiplex social network. This multiplex approach models multiple channels of informal communication-phone,word-of-mouth, and social media-that vary in their attribute values. Each agent is an individual in a threatened community who, once warned, has a probability of warning others in their social network using one of these channels. The probability of an individual warning others is based on their warning source and the time remaining until disaster impact, among other variables. We model warning dissemination using simulation parameter values chosen from empirical studies of disaster warnings along with the spatial aspects of the Coos Bay, OR, USA and Seaside, OR, USA communities. Results indicate that the initial broadcast size has a negative correlation with the critical percolation threshold, which varies from approximately 1-5%, depending on the size of an initial broadcast. A sensitivity analysis on the model parameters indicates that, along with initial broadcast size and sharing probability, forewarning and confidence in the warning significantly affect the total number of warning recipients. The results generated from this study identify areas for future research and can inform community officials about the effects of event and community characteristics on the dissemination of emergency warnings in their communities. [ABSTRACT FROM AUTHOR]