Time-scaling properties of city fires

Time-scaling scale-invariant approaches have been used to feature the temporal distribution of city fire sequences recorded in the city of AnShan (China). Our findings reveal that the point process of the city fires is a fractal process with a high degree of time-clusterization of the events. The time-clustering phenomenon is visible for timescales larger than few hours. Furthermore, the fire process tends to be less time-clusterized with the increase of the loss. The application of multifractal detrended fluctuation analysis to the series of city-fires has revealed that the sequence is dynamically heterogeneous due to the different long-range time correlation properties for small and large interevent fluctuations. (C) 2011 Elsevier Ltd. All rights reserved.