Locally-Informed Cellular Automata for Emergency Response

Emergency managers often have well-designed initial plans for responding to disasters, but such plans need to be continually adapted as each disaster unfolds uniquely over time. Resources, such as police, emergency medical services (EMS), and fire need to be positioned in appropriate locations within the area managed by the emergency manager so they can be dispatched expeditiously. For larger disasters that span multiple municipalities or jurisdictions, resource sharing becomes even more critical, given that public budgets are continuously strained and personnel availability can be a driving factor in effective response. These constraints are particularly apparent in localities that depend on shared capacity. An appropriate model is needed for dynamic responses in larger disasters that span jurisdictions. Previous research has indicated that a model based on cellular automata could be useful when applied to resource sharing at the community and county level.This paper outlines the development of a cellular automata model using actual disaster data from 911 calls in Monroe County, NY. Using machine learning techniques, such as heatmaps, the paper examines how the cells could be interacting to capture resource sharing and allocation. It makes a compelling case for states to work on lessening jurisdictional autonomy to improve emergency response during natural and man-made disasters. Additionally, this paper explores how the model can be scaled to the scope of the emergency and its requisite needs.