Gravity Model Parameter Variation during a Long-Term Study using Mobile Phone Data in the Rio de Janeiro Metropolitan Area

Abstract

Large-scale events are usually difficult for transportation planners because most survey data are focused on regular workdays. Mobile phone call records are low-cost and valuable data sources for transportation planning and monitoring. This work presents a long-term study of mobile phone call records collected throughout 2014 in the Rio de Janeiro Metropolitan Area. A careful aggregation of antenna coverage areas generated the useful spatial partition of the geographic area of study into geographic units. This spatial partition was used to identify presumed domiciles and to estimate four gravity models per day in four time-shifts. The power law gravity model distribution concerns the home distance of the visitors detected in the studied destination, Copacabana Beach, which was chosen because it is a frequent venue for several large-scale events in the city of Rio de Janeiro. The model results show that the power law parameters vary according to the type of day and the number of visitors. The analysis of parameter values reveals clusters of similar mobility patterns according to the type of day and time shift. The proposed methodology can be used as a first approach to simulate the mobility demand of new events complementing other transport planning approaches at a tactical and strategic levels.