栅格数据是空间信息的主要表示形式之一,空间信息栅格化实质上就是根据一系列采样点,按照一定栅格大小(分辨率)对所描述对象进行重建,核心问题是格网分辨率大小的确定。然而,目前,流行的地理信息系统软件在栅格数据分辨率的确定上,具有一定任意性和盲目性。本文针对连续空间对象和不连续空间对象栅格化的需求,从数据本身特征出发,以奈奎斯特采样定理和最小可视元理论为依据,以计算机处理能力为约束条件,研究了连续空间对象、不连续空间对象栅格化过程中的栅格分辨率确定方法,并给出了常用的矢栅转换、地形数据、遥感影像数据、散点数据等栅格化过程中的适宜分辨率计算公式。在此基础上,开发了ArcGIS环境下的栅格分辨率计算器插件。实验结果表明,根据数据本身特性所确定分辨率要优于GIS系统所提供的分辨率。
Raster structure is widely used in GIS with its merits that simple structure,easy spatial analysis and perfect match with remote sensing data.The popular use demands higher accuracy when converting vector data into raster one.The grid resolution,which directly determines the closeness to spatial object,is the first and critical issue for rasterizing of spatial data.Low resolution will cause mixed pixel that can't express the object clearly,while higher one will induce geometric growth of the data quantity with a result of data redundancy.Choosing a suitable grid resolution for output maps is not an easy thing since many facts must be considered,such as the scope of the study areas,research purposes and distribution of the original data.In fact,in most GIS projects,grid resolution is selected arbitrarily and blindly,without specific and scientific justification.Efforts were made to determine the raster resolution considering the inherent properties of input data,the rules of cartographic,computer processing power and other statistical concepts concerned with the various sources.Nyquist sampling theorem and the Smallest Visible Object(SVO) theory,the main basic theories in communications and map generalization,were used to demonstrate the suitable resolution.They are helpful in determining the appropriate grid resolution for the vector-grid conversion,terrain data,remote sensing image and scattered points with the restriction of computer processing power.According to the methods for different sources,a raster resolution calculator was developed under ArcGIS environment.The category of the source data and necessary parameters were inputted in the raster resolution calculator.Three standard grid resolutions for output maps are recommended: the coarsest legible grid resolution,the finest legible grid resolution and the recommended one.Finally,three kinds of input data,polygon in vector,contour lines for terrain and sampled point data for pH,were verified in the raster resolution calculator.The results of exper