Modelling traffic during Lilac Wildfire evacuation using cellular data

Abstract

Southern California is prone to wildfire events that spark major evacuations of communities in the Wildland-Urban Interface. Highly developed regions such as Southern California have a number of transportation data sources to draw from that can support emergency managers’ decision-making processes. Up to date traffic sensors such as those found on the majority of California’s highways can inform emergency managers on current traffic densities, flows and speeds. Yet, in many wildfire prone regions of the United States, this is not the case. Despite this data shortfall, many regions do have robust cellular networks that inherently produce substantial amounts of location data. The location data produced by cellphone users can be used to predict vehicular densities on evacuation routes. This study examines how cellular data can be used to predict vehicular densities on evacuation routes. A mathematical model was developed to aid in the prediction of vehicular densities on evacuation networks. Correction factors were produced to adjust for the overestimation of users on roadways by cellular networks. Extrapolation factors were also developed for estimation of the number of cellular users based on a single cellphone counts data point. The Lilac Wildfire data in Dec 2017, was used to test and validate the developed model. This methodology may prove useful to transportation planners and emergency managers in planning evacuations in areas not served by a network of traffic sensors.