Transactions in GIS, 2005, 9(3): 323–344 © Blackwell Publishing Ltd. 2005. 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA. Yukio Sadahiro Department of Urban Engineering University of Tokyo Research Paper Buffer Operation on Spatial Data with Limited Accuracy Abstract The quality of spatial data is limited by the accuracy of their sources such as paper maps and satellite images. It is also reduced by errors in map digitization and photo interpretation. Spatial operations performed on spatial data such as map overlay, buffer operations and spatial smoothing, do not always work correctly because their accuracy depends on the quality of spatial data. The present paper evaluates the accuracy of the buffer operation on a point and a line segment in relation to the positional accuracy of spatial data. Two quantitative measures are introduced to assess the accuracy of the buffer operation. Their explicit forms are derived for a point, and a condition is presented where the measures can be evaluated efficiently by numerical calculation. Applications provide empirical findings that deepen our understanding of the relationship between the accuracy of spatial data and that of the buffer operation. 1 Introduction Spatial data are digital models that represent spatial objects and phenomena in the computer environment. Unfortunately, they are usually inaccurate to some extent (Goodchild and Gopal 1989, Bolstad et al. 1990, Guptill and Morrison 1995, Burrough and Frank 1996, Shi et al. 2002, Zhang and Goodchild 2002). Quality of spatial data is limited by the accuracy of their sources such as paper maps and satellite images. Errors creep into data acquisition and manipulation processes such as map digitization, photo interpretation, and map transformation (Chrisman 1982b, Maffini et al. 1989, Bolstad et al. 1990). Spatial data quality limits the accuracy of the spatial operations and analysis imple- mented in GIS. This is often called error propagation (Heuvelink 1998), which has been discussed extensively in both theoretical and empirical aspects. Examples include map Address for correspondence: Yukio Sadahiro, Department of Urban Engineering, University of Tokyo, 7-3-1, Hongo, Bunkyo-ku, Tokyo 113-8656, Japan. E-mail: sada@okabe.t.u-tokyo.ac.jp