All the reasons that the surface limestone is exposed in a large area, the signal-to-noise ratio of seismic data is low, the underground structure is complex, and the dip angle of strata is steep make Nanchuan area to be a typical double-complex seismic geological condition of surface and underground. As the single velocity modeling method cannot obtain the accurate velocity models, it in turn affects the seismic imaging accuracy and is difficult to effectively guide horizontal well drilling in this area. In order to solve this problem, the velocity inversion of layer by layer grid tomography is carried out on the basis of structural constraints by analyzing the advantages and disadvantages of two kinds of velocity modeling methods, such as horizon chromatography and grid tomography, and combined with the actual data characteristics of Nanchuan area. That is, by picking up the seismic landmark layer with obvious longitudinal velocity variation and layer-by-layer constrained grid tomographic inversion from shallow to deep, the effect of surface and subsurface complex seismic geological conditions on velocity inversion is better reduced, and a higher precision depth domain velocity model is provided for the front depth offset . Through practical application, the imaging accuracy of seismic data in Nanchuan work area is improved obviously and basically meets the demand of shale gas exploration in this area.