Abstract: Digital Elevation Model (DEM) is known as the important GIS foundational data used in geo-simulation and terrain analysis, and its accuracy especially local accuracy have gained widespread research ever since DEM concept was proposed in 1958. This paper proposed an accuracy field model based on stochastic process method to estimate the DEM grid accuracy. Compared with traditional models, the model proved two kinds of DEM errors (noise error and approximate error) can be evaluated with uniform mathematical description, which further revealed the mathematic meaning for each components that made up of DEM error; As to application aspect，the model has sound versatility that can adapt to all error assessing on condition that build grid DEM using interpolation method.