New methodology of representing the positional error of nonpoint features in GIS

Traditionally, RMSE is used to represent the positional error of GIS features including point, linear and area features. Although RMSE can reflect the error of point features effectively and accurately, there are still some insufficiencies for the RMSE to describe the positional error of linear and area features. In this paper, the limitation of the RMSE for linear and area features has been depicted. Then some new indicators to depict the positional error of linear and area features has been introduced. These indicators include the parameters of area, perimeter, gravity, orientation, shifting, and shape similarity. First, these five parameters are configured and discussed respectively supposing only one factor influent the accuracy of the objects. Second, formulas as well as measurement methods are defined and described. Third, after individual error been measured and estimated, the cumulative error of the whole surveyed area can be integrated and computed. Based on the importance or other reasons, each error can be assigned with a certain weight when integrated them together. Consequently, we can achieve the objective of describing the positional error of GIS data, especially linear and area features, with a method different from traditional RMSE.