Block adjustment is one of the most important processing steps in topographic mapping. In order to achieve precise 3D photogrammetry points, the ratio of base-to-height(RBH) or the forward intersection angle(FIA) of stereo images should meet the requirement of specifications for topographic maps when the block images are used. In some fields, overlaps and ratios of base-toheight are often ignored, no gaps among the images are the basic requirements for block images. In this study we give a method to detect the forward intersection angle for homonymous points with RPCs, and create a indicator angler to reflect the level of weak convergence. In order to achieve stable and precise solver for block adjustment under the condition of weak convergence, this paper take the elevation value interpolated from DEM as observed values with errors, combine the height value and measured image coordinates to build adjustment model. The model can be steady solved in ill-pose situation and have little effects from the DEM errors. the model and the methods are validated by simulation and real data including optical ZY1-02C and TerraSAR images.