以四川省为实证区,借助齐夫(Zipf)定律,证明了国内旅游流和入境旅游流的流量位序—规模分布均满足齐夫法则。在此基础上进一步分析了1996~2005年各年旅游流流量位序—规模双对数坐标图上无标度区范围、标度分段特征和各标度区内部流量密度分布的变化规律。结果表明,无论是国内还是入境旅游流,其流量规模分布结构的无标度区范围都呈逐年扩大趋势,但其空间分形结构和分布形态变化却有较大差异,其中国内旅游流的分形结构由最初的局部分形演化为单分形,再由单分形退化为双分形结构;空间分布形态则由帕雷托(Pareto)分布模式向对数正态分布模式转变。入境旅游流的分形结构没有出现双分形退化,其空间分布形态也基本保持不变,仍呈帕雷托分布模式。两种旅游流规模分布不同步变化的主要原因可能是交通条件改善和新旅游景区(点)开发对国内旅游者和入境旅游者的目的地选择行为的影响不一致造成的。

The Zipf theory is the law to study the relation between word frequency and word sequence which has been widely used in the fields of urban system, transportation net and metrology. It is also involved in tourist flow, but only used for simple analysis on the ranking of tourist flow size, without deep discussion for the spatial distribution evolution rules of size structure. This paper studied the spatial distribution changing characteristics and the mechanism of tourist flow size by Zipf Rules. It tried to find the changing law of the size distribution to guide the tourism resource development sequence, and to provide new theoretical support for spatial distribution of tourism scenic sites and urban system of tourist nodes. Compared with traditional quantitative method, Zipf theory can be used to judge the non-scaling section of tourist flow size distribution more precisely, and to interpret the varying characteristic of tourist size distribution based on non-scaling section and Zipf parameters. This will provide new quantitative method for the study on the spatial distribution variation of tourist flow. In this paper, Sichuan Province is taken as the study area. From size distribution analysis of domestic and inbound tourist flow during 1996-2005, it proves that the size distribution structure satisfies Zipf law in each year. Based on this result, this paper further analyzed the variation law of non-scaling range, non-scaling section characteristic and internal structure of each non-scaling range on tourist flow rank-size ln-ln graph. The conclusion shows that non-scaling range for both domestic and inbound tourist flow vary identically on ln-ln graph, with a gradually expanding tendency. However, the variation of spatial fractal structure and distribution pattern is not synchronous. From spatial fractal structure, domestic tourist flow size distribution varies from local fractal in 1997 to single fractal of 1998-2003, then to local fractal in 2004 and finally double fractal in 2005, while inbound tourist flow varies from local fractal in 1998 to single fractal in 1999 and till now, without double fractal. The spatial distribution variation also shows the difference: domestic tourist flow is changing from Pareto distribution pattern to lognormal distribution, but spatial distribution of inbound tourist flow is still steady as Pareto pattern. Perhaps the main reason is that the transportation improvement and new scenic sites development has more effect on destination choice of domestic tourists than on that of inbound tourists. This leads to the more apparent distribution variation of domestic tourist flow size structure.