伽师强震群区、西秦岭阿尼玛卿缝合带上部地壳精细结构

【作者】 徐朝繁；

【导师】 张先康；

【作者基本信息】 中国地震局地球物理研究所 ， 固体地球物理， 2005， 博士

【副题名】复杂结构的地震波前成像方法及其应用

【摘要】 地壳结构,特别是上部地壳结构的探测与研究,对于地球学科本身的进步、防震减灾事业的长远发展及不可再生资源的进一步开发都有着十分重要的意义。上部地壳结构的知识是连接已有的大量地质学成果与较深部地壳构造的桥梁,同时上部地壳也是我们深入理解中下部地壳,乃至整个岩石圈结构的第一道屏障。上部地壳,特别是复杂构造地区上部地壳的结构,像“厚厚的云层”,阻碍了我们用地震方法这只“天文望远镜”对造成巨大地震灾害的陆内地震孕震区构造背景及地壳的演化历史和深部动力学过程的深入理解,只有对上部地壳这层在地质历史长河中永远不散的“厚厚云层”的结构认识清楚了,我们才有可能透过这“厚厚的云层”去更进一步深入认识中下地壳的孕震机理与环境,去更加科学地对地震危险区未来的震害作出评估,去为已有的大地构造理论提供更加可靠的深部地震学证据。而要做得这一点,首先要弄清楚这些地区的地壳结构,特别是上部地壳的精细结构特征。 复杂地区精细地壳结构的探测与研究主要用地震方法。大量的地震测深结果表明,地壳内部的结构,特别是造山带地区上部地壳的结构,是十分复杂的。一般说来,上部地壳的结构较中下地壳要复杂得多,而且在地表结构复杂的地区更是如此。出于对研究成本和成果产出及探测效率的考虑,宽角反射/折射和高分辨折射己成为精细地壳结构探测的最主要方法之一。上部地壳精细结构的知识则主要来自于宽角反射/折射和高分辨折射人工地震探测Pg波资料的处理结果。一般来说,探测的越细,要求的分辨就越高,所面对的研究目标就越复杂,传统地震数据处理方法的不足和缺陷就逐渐暴露出来。在复杂地壳结构地区,传统的基于程函方程有限差分解的Pg波资料处理方法就存在着以下缺陷:①有限差分反演结果只能给出速度分布,不能给出上部地壳内的可能的界面结构特征,而地壳内的界面是客观存在的,尤其是上部地壳内的折射界面较为发育,其结晶基底具有全球性,近几年在城市活断层探测中大量的高分辨折射地震资料表明,海相沉积盆地的结晶基底一般为一强折射界面;②在程函方程的有限差分解中假设波阵面处处可微,但在一些结构模型特别是在较复杂速度结构介质中波阵面可能自身相交,意味着其梯度并不存在,将导致数值计算上的不稳定性;③计算速度较慢,更多还原

【Abstract】 The knowledge of earth crust, especially the upper crust, is very important for the progress of geoscience itself and the long term strategy of disaster reduction as well as further exploitation of non-regeneration resources. The learning of the upper crustal structures builds a bridge between the geological results and the deeper crustal tectonics, it is also a first barrier for us to deeply understand the lower crust and even the whole lithosphere. Especially, the upper crust in complex tectonic areas is just like the very thick "cloud" which prevents us to use the "astronomical telescope" of seismic methods to deeply understand the seisgenic background of interior continent earthquakes with great disaster and the crustal evolution history along with the deep geodynamic process. Only after clearly comprehending the upper crustal structure like very thick "cloud" which never disappears in geological history can we possibly further recognize the seisgenic background and its mechanics, make more scientific estimation for the future earthquake disaster and provide more reliable deep seismic evidence for geotectonic theories. For this reason, the first step should be to gain a clear idea about the crustal structures, particularly the fine structure of upper crust.Seismic methods are mainly used to investigate the fine crust structures in complex region. A large number of results from DSS (Deep Seismic Sounding) show that the interior crust structure, particularly the upper crust structure in orogenic area is very complex. Generally speaking, the interior structure of upper crust is more complex than the middle and lower crust, and even more just so in the areas with complex surface structures. Considering the balance among the study cost and achievements as well as investigation efficiency, the wide angle reflection/refraction and high resolution refraction profiles have been main methods for the fine crust structure survey. The learning of the fine upper crust structure mainly come from the results of Pg data of wide angle reflection/refraction and high resolution refraction lines. By andlarge, the more fine structure to be investigated, the more high resolution to be required, and the more complex the survey objective to be faced, the inadequacy and defects of traditional processing method for Pg data are gradually exposed. So far, there are following drawbacks in traditional processing method for Pg data, which is based on the finite difference solution of eikonal equation:D The finite difference results only provide velocity distribution and can not supply the possible interface structure features in the upper crust. In fact, the interior crust interfaces do exist, especially in upper crust the refraction interface well developed and the crystalline basement exists globally. A large number of high resolution seismic refraction dada from urban active fault survey in recent several years present that in generally, the crystalline basement of the deposit basin with marine facies is a strong refracting interface;D It is presumed that everywhere on wavefront is derivable in solving the eikonal equation with finite difference method, but the wavefront may be crossed by itself in some structural model, especially in the comparatively complex media model, which means that its gradient does not exist and causes the algorithm to be unstable;□ The computing speed is not fast, which can not well suit the needs of the survey for high resolution and fine complex structures and results in very consuming cost;D The model is parameterized with square meshes, which enable the eikonal equation to be solved conveniently with the finite difference approximation,but it is not reasonable. The complicated level of real crust is different from place to place and even along the same profile, it has comparative difference among the different segments, maybe some part of which is simple and another one is very complex. Furthermore, the data amount along the different part at the same line may be different while denser observation is carried out at the complex structure part. Therefore, whether the model mesh lines are dense or sparse should correspond to both the model complex level and the data amount.In the wake of the quickly progressing of seismic observationtechnique and continuously improving of survey system as well as accumulating with a great number of seismic data, especially the needs of high resolution and fine complex structure study in urban active fault investigation and basic researches for some important geoscience problems, the high resolution refraction method will be more and more widely used. For this reason, it is very necessary to improve the drawback in artificial seismic data processing discussed above and develop seismic data processing technique that can be applied to arbitrarily complex crustal structure, particularly the fine structure survey of complex upper crust.In this paper some problems mentioned above in processing seismic data have been settled and the complete process of Pg data interpretation along with some typically applied examples, such as in Jiashi strong earthquake swarm area and Animqing suit belt as well as its adjacent region, are given. Following contents are included in this thesis:(1) In this paper, the histories of seismic refraction investigation and wavefront imaging methods in artificial seismic survey were reviewed. The characteristics of wavefront tracing algorithm in complex crustal structures and its defects in processing the Pg data of artificial seismic investigation were expounded rather completely. Some drawbacks have been improved and a new method for Pg data processing, which had been used to process and to analyze the Pg data obtained from the fine crust structure investigation profiles for urban active fault survey and some key basic geoscience study projects, was proposed. Good results were achieved.( 2 ) The defects during the forward computing for the first arrival Pg wave of the artificial seismic survey in complex structure areas had been removed and the finite different wavefront imaging method based on Huygens principle for velocity structure was proposed. In comparatively complex structures, the ray trace of Pg wave is not just a single refracted path which is presumed in traditional data processing. It presents direct wave near shot points, and at the certain distances away from the shot points it maybe appears as refracted wave, head waved ordiffracted wave which depends on different upper crust and crystalline basement structure features. The improving scheme as follows: based on Hole (1992) finite difference algorithm, using the 5 operators of Lecomte et al.(2000) to simulate the advancing, expanding, evolving and propagating of Pg wave field and to achieve the tracking for the Pg wavefront, and applying backprojectkm of SIRT(Simultaneous Iterative Reconstruction Technique) to reconstruct the P wave velocity structures.(3) The RHA(Ray Hit Analysis) method for determining the interface patterns of upper crust basements was proposed and had been used to process the artificial seismic data obtained from the high resolving refraction profiles carried out in north Sichuan and south Gansu region(NSFC Project) and in Xinjian Jiashi earthquake swarm region. Based on the P velocity model determined by wavefront imaging method, the ray hit numbers for each mesh are computed and the ray hit distribution maps are drawn up, which can be used to determine possible interface patterns. This is because if the crustal interface exists, it is sure that there are denser seismic rays ,with emergent angles near to 90,close to the interface, and the larger velocity contrast between the two sides of the interface, the denser the rays to be distributed. So except the ray dense regions caused by shot point positions, the distribution features of comparatively dense ray hits, with emergent angles near to 90°, should reflect the crustal interface positions.(4) The drawbacks of HRIM(Hagedoorn Refractor Imaging Method) had been removed, and this method was applied in seismic data processing in Fuzhou, Zhangzhou, North Hainan and Jiashi strong earthquake swarm area(Xu Zhaofan, Zhang Xiankang et al.,2002,2005; Jia Shixue,Xu Zhaofan et al., 2005; Urban active fault survey reports of Geophysical Prospecting Center, CEA, 2002,2003, 2004, 2005).Hagedoorn proved that the interface is located at the point where the travel time is just equal to the sum of forward and inverse propagation times. According to this principle, the refractor can be reconstructed. Four conditions must be satisfied using Hagedoorn principle to image therefractors, i.e. U measured reciprocal travel time data from both forward and reverse layout; D accurate reciprocal time; D known the velocity structure above the refractor; D reconstruction method for seismic wave field. Condition □ and □ may be satisfied during data collection. The velocity structures can be determined by finite difference inversion or wavefront imaging in terms of Huygens principle. An applied example was given by Aldridge et al.(1992) and Lecomte et al.(2000), respectively. In their algorithm, the velocity structures above the refractor need to be provided by other method and, it is difficult in real applications. Based on wavefront expanding in terms of Huygens principle and applying the Lecomte’s (2000) five wavefront expanding operator to simulate the seismic wave propagating so as to realize forward computing of seismic wave field, and using Hole’s (1992) back-project method for inversion along with visualizing of the results by Matlab algorithms, a complete refractor imaging method in terms of Hagedoorn principle can be completed(Xu Zhao fan & Zhang Xiangkang et al., 2002) and has been used to process the seismic data from high resolving refraction survey for the exploration of fine upper crust structures in urban active fault investigations in some cities and seismic active areas, and good results are achieved.(5) The data of the high resolving refraction profile in Jiashi strong earthquake swarm area were processed in detail with wavefront imaging method in terms of Huygens’ principle, synthetical ray hit analysis and Hagedoorn wavefront refractor imaging principle. The fine P wave velocity structures of upper crust and the clear patterns of basement in Jiashi strong earthquake swarm area were obtained, and the possible faults were inferred. Based on the comparatively complete observation system in Jiashi strong earthquake swarm area, the relations among the inverse parameters (such as initial model, iterative times, smoothing scale etc.), wavefront imaging results and ray distribution characteristics were analyzed and discussed in detail. In Jiashi strong earthquake swarm area, the basement with the depth of about 3.0km , where the Pg velocity is about 4.6km/s, appears continuously and completely, which possibly isthe bottom boundary between the complete developed marine platform basin faces and carbonic acid rock build and under which it is light metamorphic rock in middle and late Proterozoic Group ; The interface located at the depth of about 9.0km shows comparatively large variation and its depth gradually deepens from NE to S W, the depthes of which are 9.0km at southwest end and 8.5km at northeast end. This interface is characterized by the crystalline basement where the propagating velocity of Pg is about 6.25km/s. The crystalline basement between the post numbers 38km and 65km is continuous and complete and its depth range is from 8.5km to 9.0km. From the post number 37km to 38km, its depth goes down abruptly and it is about 11.5km between the post numbers 25km and 37km, the location of which is consistent with the fault beneath Jiashi inferred from wide angle reflection/refraction profile JA1 (Zhang Xiankang et al. 2002).From the wavefront imaging results of the velocity structures we can find that the crustal velocity structure above the depth of 1 lkm is not complicated and in generally speaking, the uppermost crust is homogeneous in laterally and conspicuously layered in vertically. The shallow cover above the depth of 400m is comparatively loose weathered layer and its P velocity is about 1.65~1.8km/s; The velocity isogram above the depth of 3.0km is nearly horizon, it indicates that the structure is homogeneous in laterally. Though there is a strong velocity gradient at the depth from 400m to 3.0kmt, the velocity isograms are even, i.e. the constant velocity gradient structure appears. Its P wave velocity varies from 1.8km/s to 4.5km/s. It is the second layer, which possible is the bottom boundary of clastic-barbonic acid rock build of well developed marine platform - basin facies of Tarim in Pz; The third layer is located at the depth of about 3.0km~9.0km, the upper part of which shows different velocity structure features from the lower part. The P velocity isograms of upper part are comparatively even and its depth range is from 3.0km to 5.5km, and the P wave velocity gradient obviously decreases. The lower part is a relatively weak gradient layer and its velocity is about 5.1km/s~6.0km/s, and it possible is the lightmetamorphic rock in middle and late Proterozoic Group. The fourth layer locats beneath the depth of about 9.5km, which is relatively even and its P wave velocity is about 6.3km/s. This layer may be the middle and deep metamorphic rock in late Archean or early and middle Proterozoic Group.The upper part of third layer is almost even and the average P wave velocity is about 4.8km/s. There is a strong velocity perturbation in this layer, which implyes crustal faults exists at the depths from 4.5kmto 6.0km , and three faults Fl, F2, F3 are inferred. According to the petroleum-geological data, Maigaiti fault passes through the high resolving refraction profile near the post number of 32km and Xiasuhong - Maigaiti fault has two branches, which are located beneath the post numbers of 45km and 58km, respectively. Fl fault maybe just is the Maigaiti fault and its location is slightly close to North. F2 and F3 faults should be the parts of Xiasuhong-Maigaiti fault but there is no any evidence from our results which shows that the fault Fl, F2, F3 pass through the crystalline basement or stretch to the free surface. It is suggested that each of the faults Fl, F2, F3 is not joined with the deep crustal fault in the studied area i.e. the shallow and deep structures of the crust are independent from each other.(6) The Pg data from the high resolving refraction profile which nearly perpendicularly goes through Animaqing suit zone in North Sichuan and South Gansu province were processed in detail by means of Huygens principle wavefront imaging method and the ray hit analysis technique. The fine structures of the upper crust and complex crystalline basement patterns in Animaqing suit zone and its adjacent area are obtained. The Pg wave velocity structures of the studied area show that the velocity structure in the lower part is more complex than in the upper part if roughly take the depth of 2.0km as the boundary. The velocity near surface is lower in Ruoergai basin where it is 3.86km/s~4.4km/s and higher in west Qinling fold zone of South Guansu where it is 4.0km/s~4.9km/s. Above the depth of 2.0km or so, the gradient variation of P velocity is not obvious vertically and the velocity structure is nerally homogeneous laterally in Ruoergai basin, while in the west foldbasin of South Gansu, the velocity gradient variation is obvious vertically and the velocity presents rather strong inhomogeneous laterally. The P velocity structure above the depth of 2.0km is much more complex than the area beneath the depth of 2.0 km, and there is a low velocity zone in the upper crust near North Xianman where Kusehu-Marqing fault goes through. , which possibly is the north boundary between Ganzhi and Songpan block. Wudu-Diebu fault and Zhouqu-Liangdan fault pass through the high resolving refraction profile at 9km, 30km in north Lanmusi, respectively. Beneath the depth of 2.0km, Wudu-Diebu fault shows a low velocity zone with small dimensions in velocity structure figure, and its depth is less than 6.0km. At 30km or so away Lanmusi in north, there is seemly a more large scale a low velocity distribution beneath the depth of 2.0km, and its location corresponds to Zhouqu-Liangdan fault. Because of its closing to the edge of inverse model and lack of the data, we can not give more detail patterns of this low velocity distribution. Generally speaking, Pg wave velocity structure can be divided into three regions, i.e.,south, middle and north regions, if taking both Xiaman and Lanmusi as geological block boundies. The south part is located in Ruoergai basin where the velocity variation is comparatively small and the middle part is located in Animaqin suit zone where its velocity varies greatly, while the north part lies in the West Qinling fold zone and its velocity variation is comparatively large.In the studied area, the differences of crystalline basement structures among different regions are more obvious than that of velocity distributions. Xiaman and Langmusi are the structure variation boundaries. The southern part lies in Zoige basin. Its crystalline basement is deeper and the structure is comparatively complicated. The largest depth of the crystalline basement is about 3.5km there and the basement velocity is about 5.65~5.8km/s; The middle part is located in Alimaqing suit zone and the crystalline basement is more complex than the southern and northern segments, and its depth variation is also large. The deformation of the basement is strongest in the middle segment. There are seemly double layer structure features beneath the depth of 2.5km, one ofwhich is located at the depth of 3.0km or so and its basement velocity is about 5.65-5.8km/s, and the another lies in the depth of 8.5km and the Pg velocity is about 6.2km/s; The basement structure in northern part is relatively simple and its depth is about 2.2km, and the basement velocity is about 5.5km/s. In Alimaqing suit zone, the crystalline basement is destroyed by Kusehu-Maqing fault zone and characterized by relatively large low velocity zone. Another two fault zones, i.e. Wudu-Diebu fault zone and Zhouqu-Liandan fault zone, are much small comparied with Kusehu-Maqing fault zone in the studied area.(7) Based on Matlab tool packages, the visualized programs for all results had been coded.更多还原