Mandelbrot B B
General Sciences Department, IBM T.J. Watson Research Center, Yorktown Heights, New York 10598.
Proc Natl Acad Sci U S A. 1975 Oct;72(10):3825-8. doi: 10.1073/pnas.72.10.3825.
The degree of irregularity in oceanic coastlines and in vertical sections of the Earth, the distribution of the numbers of islands according to area, and the commonality of global shape between continents and islands, all suggest that the Earth's surface is statistically self-similar. The preferred parameter, one which increases with the degree of irregularity, is the fractal dimension, D, of the coastline; it is a fraction between 1 (limit of a smooth curve) and 2 (limit of a plane-filling curve). A rough Poisson-Brown stochastic model gives a good first approximation account of the relief, by assuming it to be created by superposing very many, very small cliffs, placed along straight faults and statistically independent. However, the relative area predicted for the largest islands is too small, and the irregularity predicted for the relief is excessive for most applications; so is indeed the value of the dimension, which is D = 1.5. Several higher approximation self-similar models are described. Any can be matched to the empirically observed D, and can link all the observations together, but the required self-similarity cannot yet be fully explained.
海洋海岸线和地球垂直剖面的不规则程度、岛屿数量按面积的分布,以及大陆和岛屿之间全球形状的共性,都表明地球表面在统计上是自相似的。首选参数是海岸线的分形维数D,它随着不规则程度的增加而增大;它是介于1(平滑曲线的极限)和2(平面填充曲线的极限)之间的一个分数。一个粗略的泊松 - 布朗随机模型通过假设地形是由沿着直线断层放置的非常多、非常小的悬崖叠加而成且统计独立,对地形给出了一个很好的一阶近似描述。然而,该模型预测的最大岛屿的相对面积太小,并且预测的地形不规则程度对于大多数应用来说过高;分形维数的值D = 1.5也是如此。文中描述了几种更高阶近似的自相似模型。任何一种模型都可以与经验观测到的D值相匹配,并能将所有观测结果联系起来,但所需的自相似性尚未得到充分解释。