地理信息空间分析的理论体系探讨

A Theoretic Framework for Spatial Analysis

空间过程是生态环境、社会经济和地理系统的基本运动形式之一。空间分析是指分析、模拟、预测和调控空间过程的一系列理论和技术。在空间信息极大丰富和ＧＩＳ空间数据管理能力日趋完善的形势下， 空间信息分析理论研究获得了前所未有的发展机遇， 有必要提出新的理论体系， 和谐地包容空间分析领域已取得的各方面成果。该体系包括空间统计指标、空间回归及自适应模型、空间机理模型、空间统计与空间机理相结合的模型和空间复杂系统模型， 以及解析解、数值解和解析＋ 知识解三种求解技术。

Spatial analysis refers to a series of techniques for analysing, simulating, predicting and coordinating the spatial processes. It is necessary to propose a new theoretic framework to integrate coherently the existing wide range of spatial analysis techniques, including spatial statistic indices, spatial regression and adaptive modelling, spatial mechanism dynamic modelling, integrated spatial statistic and spatial mechanism modelling, and spatial complicated system modelling, combining with analytic, numerical and analytic+knownledge solution. The unique characteristics of spatial data in its form and relationship among them require special tackling techniques. Spatial-temporal-attribute process summarizes the sub processes of spatial and related sub processes (Fig.1). Spatial-temporal process: diseases diffusion; morphology change: river course change; GIS space: disaster distribution and ore deposit; spatial pattern: distribution of river network and disease; time series: water level curve; and classic statistics: dicing. Based on the different theoretic frameworks of spatial analysis, a new theoretic framework, covering the wider range of techniques, has been setup (Tab.2), which consists with (1) the reasons of spatial association, in raw data and in residual of model. (2) the representation of spatial association: spatial structure and spatial interaction. (3) spatial data modelling: (3a) spatial point distribution (point pattern, spatial sampling, location choice); (3b) spatial linear regression equation (with spatial correlation structure in general form, in governing equation, and in residual), kriging, and filter; (3c) spatial nonlinear equation: regression, adaptive, and others; (3d) stochastic equation: distribution, and error propagation. (4) spatial mechanism dynamic modelling: (4a) human society (behavior, traffic network, etc.); (4b) physical (GCM, hydraulic, heat transfer, etc.). (5) spatial complicated system modelling: neural network, fractal, CA, etc. (6) spatial mechanism with stochastic modelling; (7) spatial graphic analysis; (8) spatial optimalization. (9) algorithms for spatial models: (9a) consequences of OLS (with spatial correlation in governing equation, and in residuals); (9b) spatial algorithms. (10) testing of performance (residuals, and governing equations).