Consistency is a basic indicator of spatial data quality assessment. As an opposite of consistency, inconsistency of spatial data refers to the conflicts or contradictions between spatial objects at the same scale or different scales. Spatial data inconsistency is a common concern in the international geographical information community since 1990's, and will have a direct effect on spatial data integration, cartographic generalization and spatial data updating. Previous studies mainly focus on handling inconsistencies between line objects with the same scale. Less attention is paid to the problem of handling inconsistencies between objects from maps with different scales or related issues about “a building dropped in a river” in the process of cartographic generalization while deriving a smaller scale map from a larger scale map. Hence, this paper proposes an approach based on data assimilation principles for handling inconsistency between rivers and buildings. The proposed method first extracts boundaries of both rivers at different scales by means of spatial operations (e.g. split, intersection). Then, morphing transformations are performed from a larger-scale river boundary to the other one. Finally, with respect to topological and distance constraints between river boundary and a building, an optimal assimilated river boundary is determined. Experiments demonstrate that the proposed method is able to handle inconsistency between generalized river and buildings at a larger-scale effectively, providing a promising solution for eliminating topological conflicts while deriving consistently a smaller-scale map from a larger-scale map.