关键词: 沙漠植被, 地上部生物量, 分形, 自相似, BP神经网络模型

Abstract:

Function y=ax^b maybe can disclose the correlation between shoot fractal structure and above-ground biomass of hungriness plant in Taklamakan Desert compared with Turpan Basin. Desertification and salinized soil, the two serious environment problems, annoyed human in willful persecution. Especially, this phenomenon is more obvious in the second largest desert Taklamakan, which lies in Tarim Basin. This research note aims to use the theory of self-similarity to study the relationship between the shoot fractal structure and each biomass of hungriness plant in Taklamakan Desert, exert the fractal dimension (FD) to explain the capability of spatial occupation of these three plants. Three hungriness plants (Tamarix spp., Haloxylon ammodendron and Calligonum mongolicum) are chosen and the statistical self-similarity (SSM) characters among shoot, branch and above-ground biomass are analyzed in this study. Based on the close relationship of statistical self-similarity between the length of branches, crown width, external volume and shoot biomass, a fractal model on calculating shoot biomass is built. When the source data are not uniform, the results show that the simulative outcomes of SSM worse than BP neural network model (NNM) that the values of χ²-test are not up the confidence interval too. SSM can be used for one method in measuring the biomass with the data of small variance and imply the capacity of spatial occupancy with the FD. It is practicable that this growth model using biomass of some segments of one whole plant to estimate the shoot biomass in the arid and semiarid regions where vegetation is sparse, ecosystem is flimsy and building the man-made vegetation area is difficult. We emphasize that the ecological scale in this paper is of individual significance.

Key words: hungriness plant, above-ground biomass, fractal, self-similarity, BP neural network model