Study On Refraction Traveltime Tomography

T₀ differential method and traveltime tomography are the most widely used methods in geophysical exploration of engineering and environment.T₀ differential method is a conventional refraction interpretation method,which is characterized by simplicity,economy and efficiency,and it can provide engineering geology with data such as near-surface fluctuation and longitudinal change of velocity.The traveltime tomography makes up for the situation that the conventional refraction interpretation method is only suitable for the slow change of layered structure,and can not detect the deficiency of the complex underground structure,which has high applicability in the construction of near-surface velocity model and underground detection process,can identify the gradient of velocity vertical and horizontal change,depict the geometrical morphology and structure of underground structure,also,it can directly invert the physical parameters of the underground medium.Its characteristics are not affected by the characteristics of the source and GeoPhone,and the calculation efficiency is very high.Its main goal is to determine the fine structure of the underground and local inhomogeneity.In this paper,the refractiontraveltime tomography is optimized mainly from two aspects of forward and inversion.Firstly,in terms of forward modeling,a new travel time calculation method is introduced to improve the accuracy and efficiency of forward calculation.Since traveltime tomography is a ray inversion method based on the assumption of infinite high frequency,ray tracing is particularly important,which determines the travel time accuracy of the forward modeling.We focus on increasing the precision and efficiency of the traveltime,the calculation methods of the rapid advance method?FMM?on the basis of the introduction of an improved walking forward calculation method,the method by solving Eikonal equation of multiple templates?Eikonal?,computing each grid point of the solution,and then select the solution meet the wind conditions,called template more rapid advance method?MSFM?.The template centers on each grid point and covers all the nearest neighboring points.In a two-dimensional space,two templates cover eight adjacent points,including rectangular coordinates and diagonal coordinates.Compared with the fast propulsion method,this method has higher computational accuracy and efficiency,and better adaptability to the complex undulation model near the surface.Then,in the aspect of inversion,a dynamic compound prior constraint method based on t0 difference method is proposed from the perspective of multi-scale inversion.The prior information is added into the inversion equation,which strengthens the correlation between model parameters and propagation operators and improves the precision of inversion.Since traveltime tomography is a mixed problem,regularization processing is required to reduce the multi-solution of inversion and improve the stability of the solution.The selection of regularization method is the focus of this paper.Firstly,the introduction of prior information in the regularization process can effectively replace the traditional external explicit constraint method,so as to improve the stability of the solution.At the same time,the traditional method of regularization constraint in the selection of regularization factor is immutable,it will cause the inversion solution tend to rely on the information or data information of one party,resulting in the inversion,the lack of information,therefore,we adopt dynamic regularization factor selection method,namely in the iteration process,the first large regularization factors,draw the outline of the large scale of model form,falling in the regularization factor,disturbance of slowness to describe small scale model,can make the inversion problem of the solution is more stable and reliable.Finally,the model data and the measured data are used for the trial calculation and analysis,which verify the effectiveness of the optimized travel-time tomography algorithm in this paper.