摘要
用地理信息系统(GIS)获取、描述和表达栅格地理信息的尺度问题,可抽象为在欧氏空间上进行最小几何图形单元划分,以及从划分的样本集合中抽取全部或小样本对总体参量估计的影响问题,以及尺度大小对真实世界几何图形表达的逼近问题,显然后面的问题是尺度越小越好。由于欧氏空间中的几何体划分的复杂性、最小几何图形单元定义的自由性,经典的概率论与数理统计理论还未能完全阐明自由个体形状、尺度、划分方法等对地理信息描述的影响。该研究以归一化植被指数为研究对象,借助地理信息系统软件,利用对常用的5 m×5 m、10 m×10 m、30 m×30 m栅格尺度表达的已知总体,分别进行了全部自由个体的统计估计和检验,同时也进行了小样本抽样估计和检验。结果表明,地理空间的栅格尺度或地面抽样调查的几何单元大小发生了改变,但对总体的总值和均值的估计或表达没有影响。该研究结果对地理信息系统的研发和建设,对野外空间抽样框的设计,具有一定的指导意义。
The issue of using geographic information system( GIS) to obtain,describe and express the spatial scale of raster geographic information can be abstracted as the division of minimum geometric figure in the Euclidean space,the impact of all or small sample extracted from the sample set on the estimation of parameters,the approximation of size of spatial scale to the geometric figure in the real word. Obviously,to the last problem,the smaller,the better. Owing to the complexity of geometry division in the Euclidean space and the freedom definition of the minimum geometric graphics unit,classic theories of probability and mathematical statistics have not yet fully illuminated the influence of the shape,spatial scale,partitioning method,etc. of free individual on the description of geographic information. In this paper,with the help of geographic information system software,taking the NDVI as the research object,we separately statistically estimate and test the freedom individual as well as sampling estimate and test the small sample,for commonly used known population expressed by 5m × 5m,10 m × 10 m,30m× 30 m. The results show that,the totals and means of the population are not changed. That is to say,the change of grid scale of geographical space or the change of the ground sampling geometry unit size does not affect the estimation and expression of the totals and means of the population. The results of the study have some guiding significance to the geographic information system development and the design of the sampling frame in the wild.
出处
《安徽农业科学》
CAS
2015年第36期371-373,378,共4页
Journal of Anhui Agricultural Sciences