A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science

Thesis (M.Sc.) - Asian Institute of Technology, 2005

The study focuses on the positional error of digitized geospatial data. In GIS, geospatial data are used to define information that geographically and characteristically identifies natural or constructed features and defines their boundaries on the Earth. They include point, line, and polygon. Three samples of line object were selected for the experiment namely; contour line representing natural and curve lines; road representing artificial and straight lines; and city block representing artificial or boundary lines. As their positional error was tested, the distribution of positional error at midline almost conformed to normal distribution while the positional error at end points conformed to no theoretical distribution. There was 77% of tested data at midline points and 68% at end points lying in - of distribution region. The correlation between length and positional error showed a negative relation. This paper evaluated by a positional error model simulating the - error model suggested by Shi (2000), this result proved that this model is realistic and appropriate with the positional error of geospatial data. As the study concluded, the error band model can be adopted to test the positional error of existing geospatial data in Thailand.