| Beyond the memory of men,fire has become one of the most permanent and severe disasters.Besides great casualties and losses of property,fire also endangers the fields such as energy,resource and environment.Effective prevention and control of fire rely on in-depth understanding of the behavior and characteristics of the fire system.Forest fire system is one part of the forest ecological system.There are many complex interactions between forest fire occurrence and its influencing factors, which may lead to the correlation between fires.To prevent and control fires,we need analyze the fire risk under different conditions and obtain quantitative effect of each parameter on fire system.The human activities also play an important role in the fire occurrence.Thus,we first focus on the effect of several factors on the forest fire probability,including the weather condition,population and so on,based on the fire occurrence principle.Especially,we quantify the influence of population density to the forest fire occurrence probability.Furthermore,impacts of some weather parameters on the urban fire probability are also studied.In 1998,Malamud et al compared actual forest fire data with simulation results of forest-fire model proposed by Drossel et al.and found that actual forest-fire system has self-organized criticality(SOC) characteristics.From that time, investigations on systemic law of actual forest fire have been carried on.But there are still some problems to solve,such as the applicability of SOC for different forest fire systems,the difference of SOC characteristics between different forest fire regions,etc.On the other hand,in contrast to the researches on forest-fire system,it seems that there have not yet many studies on other fire systems.Song et al investigate the SOC characteristics of the urban fire system of China.Is SOC characteristics of urban fire system universal? Therefore,in the second part of this dissertation,we study the spatial distribution of the fire system.Here,the spatial distribution is referred to fire intensity,i.e.fire area,for forest fires and fire loss for urban fires.Firstly,the "frequency-area" distribution of forest fires in Japan is found to satisfy power-law relation,which is similar to that of China and America.Steady power-law relation of "frequency-size" distribution is one of the important behaviors of SOC,so we investigate the power-law relation in different periods and different regions.The results validate that this power-law relation of "frequency-size" distribution is steady and scale-invariant.Then we investigate the "frequency-loss" distribution of urban fire system in Japan and several cities of China.It is found that the urban fire system also has SOC characteristics.Furthermore,a system has not only the spatial distribution characteristics,but also temporal distribution characteristics.So what is the temporal distribution of forest fires and urban fires? What is that of fires with different intensity? Aiming to solve these problems,in the last part,we analyze the temporal characteristics of fire system by means of several statistical methods,including coefficient of variation, "frequency-interval",Fano factor and Allan factor.It is found that "frequency-interval" of forest fires satisfies power-law relation with a periodical fluctuation,which satisfies exponential relation for urban fires.Results of Allan factor and Fano factor show that both forest and urban fire systems have self-similarity and scaling behavior with periodical fluctuation.The periodical fluctuation of fire time distribution is probably associated with and similar to human activities and main weather parameters(e.g.relative humidity).Results show the fractal characteristics of forest fire time sequence and the "fractal-rate" stochastic point process characteristics of urban fire time sequence.
|
PDF Full Text Request