Printer-friendly versionSend to friend

Command, Control, and Interoperability Center for Advanced Data Analytics (CCI CoE: CCICADA)

Witch Fire in Del Dios on Oct 23, 2007 (San Diego Wildfire Education Project, http://interwork.sdsu.edu/fire/ photo_gallery/photo-gallery.htm)

Warning systems play an essential role in notifying the population of the future danger and providing precautionary and safety information so that necessary actions can be taken. In addition to reliable warning systems, it is essential to make use of existing social communication networks in communities to diffuse emergency warnings in such a way that the warnings can reach a larger audience and people at risk will act on the information. CCICADA’s research in this area focuses on modeling the diffusion process of warnings in a dynamic network. An axiomatic framework is formulated that incorporates the concept of trust, which quantifies the likelihood that individuals will believe the message being conveyed to them. The network is dynamic in a sense that individuals may leave the network and disrupt the flow of information as warnings are being diffused. An agent-based model can to be used as a tool to gain insights and how social communities and demographics may respond to various warning technologies.

Steele Canyon High School evacuation site for Harris Fire. (San Diego Wildfire Slideshow on KPBS.org, http://www.flickr.com/groups/sandiegofires/pool/)

The Rensselaer Polytechnic Institute (RPI) team performed a case study using demographic and event data from the San Diego Firestorms that occurred in 2007. RPI applied its framework to model the diffusion process of the Reverse 911 evacuation warnings that were sent during the event. RPI mapped the network process and configure the parameters using survey data provided by Oak Ridge National Lab and the After Action Report released by the County of San Diego Office of Emergency Services following the event. RPI constructed a social network of one million households based on the demographics of San Diego County and configure the household attributes using summary statistics from the survey data. RPI adjusted the parameters so that the number of evacuated households from the model is close to the actual number reported. This procedure enables RPI to configure the parameters in the model and use the model to investigate questions such as how social group structure, distribution of trust, and existence of weak ties affect the dissemination of evacuation warnings.

San Diego map showing the estimated fire perimeters (in red) and evacuation areas (in yellow) for October 24. The dots represent respondents from the survey data.

The framework demonstrates the ability to construct a large-scale social network that can be used as a basis for simulation, and simulate the warnings broadcast and the potential eventual evacuation behavior of the community in total. The parameters in the model can be calibrated to perform various scenario analyses. The results show the value of dynamic social network analysis and simulation in studying the diffusive processes.