Abstract: Conventional first-arrival traveltime tomography faces challenges in the computational accuracy and efficiency of traveltime and inversion precision when dealing with large offsets in high-density data. To address these limitations, we propose a near-surface velocity modeling method with double-difference constraints for large offsets. A hybrid approach combining local operators with a global fast scanning method is implemented to solve the eikonal equation, significantly improving the accuracy of first-arrival traveltime calculation for large offsets. Furthermore, double-difference traveltime constraints are incorporated into the iterative inversion to better constrain velocity updates between adjacent rays and reduce error accumulation in large-offset traveltime computation. The test results of model and field data demonstrate the applicability and feasibility of this method in processing large-offset seismic data.