Quality Assessment for Attribute Data in GIS Based on Simple Random Sampling

The research of spatial data accuracy and uncertainty began in the 1960s,and the NCDCDS(National Committee for Digital Cartographic Data Standards,1988) identified several components of data quality as positional accuracy,attribute accuracy,logical consistency,completeness and lineage.At present,the issue of the representation of uncertainty in spatial data has become more and more of a concern.The description of error is take as a"function of information" and a"fundamental dimension of data" because of endemic nature of error in GIS widely recognized.Correspondingly,the development of what have been termed error-sensitive GIS has been deeply researched.The using of sampling method in the analysis of spatial data accuracy and its measurement has two reasons:① it conforms to the international standard that spatial data is taken as a product to be quality inspected;② the data quality can be deduced and described with low expenses and high efficiency.So,on the basis of the principles of the simple random sampling,the statistical model of rate of disfigurement(RD) is put forward and described in detail.Based on the definition of the simple random sampling for the attribute data in GIS,the mean and variance of the RD are deduced as the characteristic valve of the statistical model in order to explain the feasibility of the accuracy measurement of the attribute data in GIS using the RD.From the result of the deduced equation about the mean and variance of RD,the mean of RD is the measurement of the amount of the defects while the variance of RD is the measurement of the scatteration of the defects.So the mean of RD can be used to measure the attribute data accuracy while the variance of RD can measure the accuracy of the sampling and assure the sampling confidence.After that,the quality assessment method for attribute data of the singe or batch of vector maps during the procedure of the collecting is discussed based on the mean and variance of the RD.The RD spread graph is also drawn to see whether the quality of the attribute data is under control.The RD model can synthetic judge the quality of attribute data,which is different from other measurement coefficients that only discuss the accuracy of classification,so it can find its significance by realizing the measurement of the accuracy during the research of accuracy and uncertainty for attribute data in GIS.