Abstract: Spatial outlier detection is one of the major data mining methods. Detection of outliers will contribute to the discovery of implicit knowledge, significant changes, surprising patterns, and meaningful insights. In the field of geography, a spatial outlier is an object whose non-spatial attribute value is significantly different from the values of its spatial neighbors. Most current spatial outlier detection methods primarily consider that all the objects for outlier detection are correlated. Actually, spatial correlation decreases with the increase of distance. At the same time, the objects could be potentially wrongly identified as spatial outliers when there are several real outliers in their spatial neighborhoods. From the viewpoint of the spatial data field, a similar Gaussian potential function is utilized to measure the degree of spatial outlier degree. Further a field-theory based spatial outlier detecting algorithm is proposed. Firstly, the spatial clustering is employed to extract the local autocorrelation patterns, called clusters. Then the clusters were utilized to construct the reasonable and stable spatial neighborhoods using the constraint Delaunay triangulation. Finally, a robust spatial outlier measure is proposed to determine spatial outliers in each cluster. Experimental results show that the proposed method is effective for determining detecting spatial outliers in spatial point datasets.