Study Of First-arrival Traveltime Tomographic Velocity Modeling Method

This paper studied the first-arrival traveltime tomographic velocity modeling method andwe mainly focused on the research of first-breaks picking, first-arrival path ray tracingmethod, establishing and solving of the tomographic equations, the technology route of ourresearch is from the horizontal surface to the irregular surface and from two-dimension tothree-dimension.As an inversion problem, the first-arrival traveltime tomographic velocity modelingmethod tried to infer the velocity model from the seismic first-arrivals. We first talked aboutthe rules of the propagation of seismic waves in the near surface through the types offirst-arrivals and its propagation mechanism. Comparing the common near surface velocitymodeling method, we find that first-arrival tomography is an accurate and efficient modelingtool.The core of tomographic inversion is the calculation of sensitive ker nel function andraytracing is the main tool for the computation of sensitive kernel function and traveltimeunder the ray theory. The realization of first-arrival tomography needs a raytracing methodthat can trace first-arrival raypath accurately. Based on the previous studies, we derive theformulation of parabolic traveltime interpolation and propose the shortest path raytracingbased on parabolic traveltime interpolation. The heart of the method is that we calc ulate timeof one node not only using Dijstra algorithm but also by parabolic traveltime interpolation atthe interpolation segment formed by the nodes having obtained the local traveltime, then takethe minimum as the final time of the node, and the process of raypath tracing is also theroutine of backward tracing from receiver to shot by parabolic traveltime interpolating. Modeltests prove the method is accurate, valid and applicable.The traveltime picked from seismogram by first-break picking is the target data oftomographic inversion, whose purpose is to make the first-arrival traveltime from the modeling velocity field and the first-break picking traveltime best match in some kind ofnorm. Therefore, picking first-arrival traveltime accurately is prerequisite for tomographicinversion. This paper used the edge-preserving smoothing for first-break picking, throughEPS processing on different seismic attributes and their derivative, which not only played asmoothing effect but also highlighted the takeoff point of first-arrival and make the first-breakpicking easily. Raytracing forward modeling can get the first-arrival traveltime and thesensitive kernel function, then generating the residual traveltime by subtracting the time fromfirst-break picking and forward modeling. Next, we established tomographic equations andsolved the slowness update values by least squares QR decomposition(LSQR)which mixedwith regularized. Then iteratively updated the initial model until the meet accuracy. Wecalculate the static correction values based on the to mographic velocity, and the tests onmodel and real data proved the correctness, feasibility and applicability of our approach.Finally, the shortest path raytracing based on parabolic traveltime interpolation andcorresponding tomographic velocity modeling method successfully extended tothree-dimension velocity modeling and tested on some simple models.