The Multi-Scale Spatial Pattern Recognition of Vegetation Based on Fractal Geometry

Abstract

Spatial heterogeneity of the vegetative will appear new change and new feature when we observe it with different scale. In order to understand the vegetative pattern and dynamic state well and profoundly, we must take into account the characteristics which vary with the different scales. Fractal geometry is a feasible tool to solve the problem. The spatial distribution of the vegetation is a typical fractal object and show details in different scales. The fractal dimension always embodies self-similar characteristics which mean they don't change with the scales. Consequently, comparing the spatial pattern and the fractal dimension between different scales will make us understand the spatial pattern roundly. Based on regionalized variable theories, geo-statistics is one kind of spatial statistical theory used to explore the correlativity and dependence between spatial variables. The first character of this method is its emphasis on the importance of spatial dependence of variables. In practical research, semi-variance values of ecological factors or other indices can be calculated from the semi-variance formulate according to the theory, and then, semi-variogram can be drawn, distribution character of the vegetation (such as clumped or uniform pattern) can be found from the graph. Mathematical models simulation should be used in quantification of this character. In this paper, we take the vegetation of XINJIANG province for example. The fractal dimension was calculated by the double-logarithm semi-variogram. The lower the value of the fractal dimension is, the higher heterogeneity the distribution of vegetation has.