oa 3-D Velocity Images in Near Surface

A 3-D tomographic inversion approach for reconstructing 3-D near-surface model is presented in this paper. Direct, reflected and turning waves are used simultaneously to update the near-surface model. We analyze the characteristics of the first-break traveltime in complicated low-velocity layers. To improve the accuracy for the velocity model, the various first-break times from direct, reflected and refracted waves are considered for model inversion. A fractal algorithm is applied to pick first breaks, which overcomes the error caused by wavelet shape differences, and the leg-jump, of refractions. The method is capable of picking a large volume of first breaks automatically. The ray paths and traveltimes are calculated with a 3-D ray tracer which does not increase computation time for complicated geological models. Our method can determine the ray path associated with minimum traveltimes regardless of the wave mode (direct, refracted, or reflected waves). We use a least-squares approach in conjunction with QR decomposition to reconstruct a 3-D near-surface velocity model from the actual first-break times obtained from 3-D data.