Spatial Statistics And Its Application In The Analysis Of Spatial Pattern

Spatial statistics which keeps a close connection with geo- statistics having originated from geology is always widely used and based on classical statistics. The study areas of spatial statistics contain regular statistic ananlysis which include expectation of stochastic variables,variance , covariance,samples distribution, parameter estimate, hypothesis test, variance analysis , regression analysis and so on. Geo- statistics which studies variance function and its theory model, interpolation ,conditional simulation, is one part of spatial statistics .Multivariable- spatial statistics ,non-linear and non- stationarity spatial statistics, fractal theory and wavelet ,spatial regression,computing intelligence all pertain to spatial statistics. Spatial statistics aims to describe geo-phenomena and geo-process, interpret cause and effect and forecast the trend. Geo- statistics has been attached importance to and progressed greatly. But spatial regression, spatial simulation, spatial forecast spatial pattern and computing intelligence are all in their initial stages.This thesis focuses on spatial pattern. Spatial pattern lies between spatial prototype and spatial model and is the geographic object's reflection on the analysis space.So far the studies about spatial pattern have scattered and the structure characteristics haved been neglected and it cannot have been widely used.This article emphasizes the preliminary analysis of point pattern and area pattern.Chapter 1: introduction.At first this chapter interprets the meanings of spatial statistics and its study object which includes all the objects and phenomena which have spatial attribute.Then its developing progress is introduced. Quantitative geography is the rudiment of spatial statistics and geo-geography has pushed spatial statistics ahead. Spatial statistics is easily confused with other concepts such as spatial analysis,spatial data analysis,spatial statistic technique,spatial statistics and geo-statistics.Chapter 2: spatial datas' measurement scales and statistic analysis.Spatial datas could be classified as interval datas,ratio datas,order datas and nominal datas according reference standard.The measurement of central tendency and dispersion is the most basic statistic analysis of spatial pattern. Central tendency includes mean center,median center and Euclidean median which is the most widely used and has the most practical meaning. Public facilities are always sited at Euclidean median in certain area to maximize the interests.But it is very difficult to find the center in an irregulararea.The dispersion describes the extent of all points dispersing from the central point and it can be classified as absolute dispersion and relative dispersion. Both absolute dispersion and relative dispersion are influenced by atypical or extreme points greatly.Relative dispersion is still influenced by region area,region shape and boundary delimitation. The measurement of central tendency and dispersion only describe synoptically and interpret simply the complex spatial pattern ,so it is hard to avoid the windage and it is a cursory and not a precise measurementAt the same time ,the skewness coefficient and kurtosis coefficient are used to describe the distribution of density function of geographic features .Besides those,several relative parameters and spatial sampling types are discussed.Chapter 3: spatial point pattern analysis.The point pattern is classified as three types :clustered,dispersed and random.The judgement means are as follows:the nearest neighbour methods,quadrat analysis,area methods and function methods.As far as the nearest neighbour methods are concerned,the delimitation of area boundary is the key to the analysis result.Besides,standard nearest neighbour index R is not always corresponding to the type of point pattern,however,ordered nearest neighbour methods and area nearest neighbour methods can settle the problem.The two methods are not widely used.Quadrat analysizes the frequency distribution in different parts of the study area,and then confirm the ty...