Fractal Characteristics Of Picea Schrenkiana Var.tianschanica Forest Of Xinjiang And Its Application In Forest Classification

Fractal characterizes the self-similarity between parts and whole,it is a geometry that really describes nature,which means it can more accurately describe the complexity,diversity and irregularity of natural things and expand the human cognitive domains.Picea schrenkiana var.tianschanica is a dominant constructive species of mountainous forest in Xinjiang,which plays an important role in water conservation and soil and water conservation.Clarifying the fractal characteristics of Picea schrenkiana var.tianschanica forest can not only reveal its self-similarity characteristics,but also provide new ideas and methods for forest zoning and classification.In this study,the Picea schrenkiana var.tianschanica forest in Xinjiang Agricultural University practice forest farm was taken as the research object,the fractal dimension of Picea schrenkiana var.tianschanica forest in different canopy density and different resolution scales was detected and analyzed,and the mountain forest classification with fractal parameters was carried out.The results are as follows :(1)In terms of fractal characteristics of Picea schrenkiana var.tianschanica forest under different canopy densities,different algorithms(differential box dimension method,double blanket covering model method and fractal Brownian motion method)were used to obtain the fractal dimension of Picea schrenkiana var.tianschanica forest under different canopy densities(low,medium and high).The fractal dimension of Picea schrenkiana var.tianschanica forest was basically maintained at about 2.4,and the average coefficient of variation was close to 0.0200,which was in a small range,indicating that Picea schrenkiana var.tianschanica forest had a clear fractal dimension and showed a relatively stable characteristic that did not change with canopy density.(2)In terms of the fractal characteristics of Picea schrenkiana var.tianschanica forest at different resolution scales,the fractal dimensions of Picea schrenkiana var.tianschanica forest at different resolution scales(UAV image,World View-3 image and ZY-3 image)are 2.4390,2.4256 and 2.0885,respectively,which are slightly reduced with the decrease of image spatial resolution.The fractal dimensions of Picea schrenkiana var.tianschanica forest obtained by different algorithms are mainly between 1.9 and 2.6,and also show relative stability in numerical value.(3)In terms of the participation of fractal parameters in mountain forest classification,the fractal parameters generated by different algorithms(differential box counting,double blanket covering model method and fractal Brownian motion method)and different resolution images(unmanned aerial vehicle image,World View-3 images and ZY-3 image)were used to carry out forest classification in the study area.The classification accuracy of the obtained results was 1.46%–8.49% higher than that of the original images,and the classification effect was improved to varying degrees.In particular,the fractal parameters generated by double blanket coverage model had the largest increase in the classification accuracy,reaching 6.06%–8.49%,followed by the fractional brownian motion.The increase rate was 3.00%–4.13%.It shows that fractal parameters can improve the accuracy of forest classification in some degree,but the effect will be different due to the different algorithms.