期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2004
卷号：XXXV Part B4
页码：1025-1030
出版社：Copernicus Publications
摘要：In this study, the accuracy of a geographical information system is investigated. The mentioned accuracy is not only the geometric accuracy, but also the thematic accuracy. It is known that the standards for geographical information system are already a popular and topic in both practical and academic world. This study gives an example for the investigation of accuracy. Geometric accuracy can be derived using the errors between the geometric results and the real values. The real values can be calculated precise surveys or taken from maps of larger scale than the scale which is used in the study. This choice depends on the dimensions of the study area and the scale of the study. Thematic accuracy can be derived from differences of individual defined and to database added objects. These differences are the errors of definitions. The originality of accuracy investigation within geographical information system is to decide, how many objects from itself are used for within this investigation. Ideally, all of the objects must be used for this thematic accuracy investigation. But it is nearly impossible to do this investigation with all the objects if a big amount of them are used within a GIS project. This disadvantage forces to used sampling methods. In this study, an accuracy investigation of a MOLAND project for Istanbul has been done. By geometric accuracy investigation, a sampling class size has been defined for all the object types within the project and these samples are chosen from database randomly. Afterwards, the true values are taken from 1:1000 digital maps for referring samples. The errors are calculated and from these errors, a global geometric accuracy values are derived. For the thematic accuracy investigation, sampling class sizes are determined for all different landuse types. This process gives a number of approx. 3000 for sampling class
