Objectives: To examine patterns in the timing and location of incidents of mar-itime piracy to seewhether, likemany urban crimes, attacks cluster in space and time. Methods: Data for all incidents of maritime piracy worldwide recorded by theNationalGeospatial IntelligenceAgency are analyzed using time-seriesmod-els andmethodsoriginally developed todetect disease contagion.Results:At the macro level, analyses suggest that incidents of pirate attacks are concentrated in five subregions of the earth’s oceans and that the time series for these different subregions differ. At the micro level, analyses suggest that for the last 16 years (ormore), pirate attacks appear to cluster in space and time suggesting that pat-terns are not static but are also not random.Conclusions:Much like other types of crime, pirate attacks cluster in space, and following an attack at one location the risk of others at the same location or nearby is temporarily elevated. The identification of such regularities has implications for the understanding of mar-itime piracy and for predicting the future locations of attacks.

This paper presents an initial demonstration of a model to generate dynamic patterns of maritime piracy. Model details, outputs and a method for assessing the goodness of fit of model outcomes are presented. The model presented here is a tool to produce dynamic maritime piracy risk. The Gulf of Aden is considered as a case study, and data on pirates, vessels routes and flows through the Gulf of Aden in the year 2010 are used to build and calibrate the model. Agent-based modeling is employed to simulate pirate, vessel and naval forces behaviours. This approach might be used to test different strategies to fight maritime piracy, not just by testing naval behaviours but also by exploiting ABM to test social factors that might deter such a phenomenon. 1

Objectives: To examine patterns in the timing and location of incidents of mari-time piracy to see whether, like many urban crimes, attacks cluster in space and time.Methods:Data for all incidents of maritime piracy worldwide recorded by theNationalGeospatial IntelligenceAgency are analyzed using time-seriesmod-els andmethods originally developed to detect disease contagion. Results:At the macro level, analyses suggest that incidents of pirate attacks are concentrated in five subregions of the earth’s oceans and that the time series for these different subregions differ. At the micro level, analyses suggest that for the last 16 years (or more), pirate attacks appear to cluster in space and time suggesting that patterns are not static but are also not random. Conclusions: Much like other types of crime, pirate attacks cluster in space, and following an attack at one location the riskof others at the same locationornearby is temporarily elevated. The identification of such regularities has implications for the understanding of maritime piracy and for predicting the future locations of attacks.

Abstract not found