| The 3D-3C seismic wave-field is the most complicated wave field now, which contains a lot of geologic information, at same time the three-component seismic survey method uses the full wave-field shooting and receiving for elastic wave, so this method can express the subsurface elastic characters and the rock physical properties fully at the current seismic exploration area. In order to obtain more and effective stratigraphic information, we need to process 3D-3C seismic data accurately, while the most difficult task in converted wave processing is the static correction of converted wave. Because the converted wave vibrates along the horizontal direction, the attenuation coefficient of near-surface converted wave energy becomes higher, the signal-to-noise ratio becomes lower and the influence of near-surface strata inhomogeneity on converted wave imaging becomes stronger, as well as the residual statics of converted wave have increased too. The conventional static correction method cannot deal with converted-data processing correctly anymore, so we need to set about new research from the following aspects. In the 3D converted wave survey, we often use the compressional-wave component to calculate the shot statics, the coefficients of transverse dip and vertical dip, while use the converted wave component to calculate the receiver statics. In practical seismic exploration, we would set about the three-component seismic micro logging and the short refraction survey firstly, then use the micro logging points and short-refraction points to be the control points, establish the basic data base of 3D converted wave static correction by interpolation method, and make the tomographic inversion of first arrival among those control points to set up accurate near-surface velocity model; secondly we would make the field elevation static correction; finally we would calculate the converted wave residual statics by the nonlinear and linear mixture inversion and then carry the static correction processing of 3D converted wave seismic sections. A study of the conventional tomographic forward and inversion measures for the first arrival travel-time in the converted wave exploration area. In the study of ray tracing measure of first arrival wavefront, we have mainly researched a ray tracing method of travel-time interpolation, which disperses a model into even square units and sets some nods on the units boundary for calculating travel-time, with the hypothesis that the travel time of any point on boundary can be obtained by the travel-time linear interpolation between the two adjacent discrete points on the boundary. The calculation process involves two steps; the first step is to calculate the travel-time of each nod on unit boundary by the interpolation formula, with starting from the shot points; The second step is to define the ray path according to the Fermat principle and the interpolation formula, with starting from the receiver points. The ray path defined by above method is not a connecting line among those discrete nods on unit boundary, but the line that crosses through the point that is satisfied with the minimum travel time condition fitly, and the refraction angle on the boundary changes continuously with the incidence angle. This method has overcome thedisadvantages of the shortest path ray tracing method and improved the calculation accuracy. In above travel-time linear interpolation method, we have supposed that the travel-time is changed linearly with the distance, and that is a kind of approximate. Practically the seismic wave traveling through homogeneous medium spreads from the source point to the surround by means of spherical wave, and the wave-front shape of 2D homogeneous medium is circular. The wave-front equation with the velocity changing linearly in 2D medium is also a circular equation. Therefore a travel-time interpolation method with higher accuracy should be interpolated nonlinearly according to the actual surface wave front. Using this method we have taken the forward and inversion calculation of first arrival travel time in the converted wave exploration area. The tomographic measure study of the converted wave first arrival travel time in Fresnel zone. For utilizing only the travel time data that can be picked up easily and its uncertainty can be estimated reliably, the seismic travel time tomographic method basing on the ray theory has been widely used in reconstructing the subsurface media velocity distribution with its simple algorithm and high efficiency. But the hypothesis of this kind of methods is that the frequency of seismic wave is high enough, the wave-length is infinitely small and the seismic energy only transmits along the zero-volume rays as well as the travel time is the linear integral for the slowness (or the reciprocal of velocity) along the path between shot points and receiver points. This is akind of mathematical abstract. This kind of idealized rays is referred to mathematical rays too, which only can descript the kinematics characters approximatively for seismic wave propagation, but cannot reflect the real physical process. This kind of rays is nonuniform distribution in the model space. It would deviate as meeting the low velocity objects and accumulate at the high velocity areas, which would introduce the unsuitability seriously into the tomographic inversion and make great influence on the reliability and the resolution of tomographic results. The actual seismic wave is bandlimited and the seismic energy doesn't transmit only along the ideal rays. The other medium that the ideal rays don't set foot in also would influence the travel time seriously. The wave equation describes the dynamics characters of seismic wave transmitting perfectly. The waveform tomographic method basing on the wave equation has considered the frequency factors of seismic wave and all the media effects of whole media space including any diffraction phenomena. Although this method has been solved in the mathematics and has the potential high-resolution itself, the problem of convergence at local minimum brought by the higher nonlinear character in its inversion still needs us to overcome. Actually the seismic energy mainly propagates through a small range between shot points and receiver points, which includes the ideal rays (i.e. first Fresnet zone). Only the media of first Fresnel zone near ray path can make great influence on the travel time of receiver records, while other medium that is far away from the rays and besides the zone will play less role. In the paper we have put forward a tomographic method of first arrivaltravel time basing on Fresnel zone for reconstructing near surface converted wave velocity model and have realized the method, the good results have been achieved from this theoretical model experience. Comparing with the conventional ray tomographic method, the tomographic method of Fresnel zone has the following advantages: (1) Considering the frequency components of actual seismic wave and making the measure more suitable for the actual transmitting rule of seismic wave in the physics. (2) Reducing the instability of inversion course and the uncertainty of inversion results because of ray sparsity. (3). Improving the ability of anti-error; (4). Improving the resolution and reliability. The experience research of a fuzzy method for the tomographic inversion of converted wave first arrival time Comparing with the transversal wave, the 3D converted wave has the following characters that it is easy to be shot and can generate the converted wave and the compressional wave at the same time, with low additional cost; the absorption loss of the converted wave is half of that of the pure transversal wave and the frequency of converted wave is higher than that of pure transversal wave. But there are some inconvenient aspects in the 3D converted wave method, which mainly involves that the travel path of converted wave is asymmetry that will increase the complexity and difficulty for seismic data acquisition and seismic processing; because of the surface interference, the signal-noise ratio of 3D converted wave data is very low and the signal-noise ratio of firstarrival in the single-shot records is also very low, as well as the first arrival cannot be picked up correctly by the artificial method, therefore the 3D converted wave travel time picked up is a fuzzy item itself. In addition those signals of records that have been observed actually have some errors and have some blur characters, especially in the complex areas, the lower the signal-noise ratio is, the higher the blur level is. The first arrival time is the basic data for the first arrival inversion and the static correction. Because the operation of picking up the first arrival time automatically exists large errors, while artificial picking up may cost much time and is not accurate too, as well as there are some fuzzy components in the first arrival records, so we can say that the first arrival time data used here is not a certain quantity, but a blur quantity. So far the most of first arrival inversion and static correction methods depend on the accurate first arrival time, and the inaccurate first arrival time will introduce much more errors into the final calculation results. So many examples express that the first arrival inversion or the static correction are failure due to the errors of first arrival time. Because of the lower signal-noise ratio of converted wave data, the data of first arrival time is not a certain quantity but a blur quantity, and the lower the signal-noise ratio of converted wave data is, the higher the fuzzy level is. The inaccurate first arrival time will bring larger deviation in inversion results. Therefore we have considered the fuzzy logic tomographic method that sees the first arrival time as a blur quantity, made a modeling experience and actual data processing. The final result shows that the fuzzy logic method has good prospects at the noise pressing, making the iterative inversion process to beconvergence stably etc. Obtaining the large residual statics by the nonlinear and linear mixture inversion measure. In this paper, basing on analyzing the measures of the stimulated annealing and the linear automatic residual static correction, we have reformed the two steps method used to solve the nonlinear automatic residual static correction. Firstly we define the residual static correction space depending on the apparent frequency and the apparent cycle of seismic data, and make this model space to be small enough. Basing on the rougher residual static correction model, we obtain the rougher results of automatic residual static corrections by the stimulated annealing method. During the course of obtaining results, we have chosen the second-order tendency forecast, for the reason that a two-order trend surface with free dimension has only one extreme point. In the condition of two dimension, the basic method that using the trend analysis to accelerate the stimulated annealing is expressed as following: Beginning the stimulated annealing process, giving a certain initial guess and searching the model space randomly again and again. As the model points researched are more than 6, constructing the two-order trend surface by the least square method at a appropriate time with the former m minimum object function values recorded and their corresponding model points. Using the extreme point of trend surface as the initial solution of linear inversion, then finishing the inversion process as the optimum solution of...
|
PDF Full Text Request