华南陆缘是我国重要的矿产、地热资源区.晚中生代以来，在太平洋板块西向俯冲，地幔热对流活动共同作用下，该区出现多期岩浆-热事件和大规模爆发式成矿作用.在前人研究基础上，本文利用地表热流观测资料、地震剪切波资料、重力位球谐系数，计算了壳-幔温度结构，分析了动力学背景.计算结果表明：华南陆缘东南沿海地带，地壳10 km以浅温度达200℃以上，居里点温度475℃，莫霍面平均温度550℃.地壳浅层较热，花岗岩中放射性元素衰变放热是地壳浅层地下水热活动的重要热源，但地壳总体温度不高，为"冷壳热幔"型热结构.地幔中，90 km深度，温度950~1250℃；120 km深度，温度1050~1400℃；150 km深度，温度1200~1450℃；220 km深度，温度1500~1700℃."热"岩石圈底界深度在110~150 km之间，西深东浅.岩石圈内，地幔应力场为挤压-伸展相间格局；岩石圈之下，地幔应力场为一个以南昌为中心、长轴NE-SW向的椭圆.分析认为，晚中生代以来，太平洋板块的西向俯冲，导致华南陆缘在区域性SE向地幔对流背景上叠加局域性不稳定热扰动，在175~85Ma期间，上地幔物质向上流动，形成不同的岩浆活动高峰期.同时，岩石圈地幔受俯冲洋壳流体的影响，含水量高，黏度小，在地幔流切向应力场作用下，岩石圈底界由西向东"波浪"状减薄.现今岩石圈之下仍具备地幔小尺度热对流温度条件，但除地表浅层外，地壳整体温度不高，岩石圈构造稳定.

Attenuation of seismic random noise is considered an important processing step to enhance the signal-to-noise ratio of seismic data. A new approach is proposed to attenuate random noise based on a deep-denoising autoencoder (DDAE). In this approach, the time-series seismic data are used as an input for the DDAE. The DDAE encodes the input seismic data to multiple levels of abstraction, and then it decodes those levels to reconstruct the seismic signal without noise. The DDAE is pretrained in a supervised way using synthetic data; following this, the pretrained model is used to denoise the field data set in an unsupervised scheme using a new customized loss function. We have assessed the proposed algorithm based on four synthetic data sets and two field examples, and we compare the results with several benchmark algorithms, such as f- x deconvolution ( f- x deconv) and the f- x singular spectrum analysis ( f- x SSA). As a result, our algorithm succeeds in attenuating the random noise in an effective manner.

塔中地区主要目的层是深层奥陶系碳酸盐岩储层,由于地表沙丘起伏大、沙层松散,地震波吸收衰减严重,干扰波复杂,导致目的层地震资料信噪比非常低。沙丘所产生的噪声不仅使有效波特征受到影响,而且使面波等相干噪声的规律性遭到破坏,从而增加了去噪处理的难度。为了在保真的前提下最大限度地压制干扰波、提高塔中奥陶系目的层资料信噪比,基于多年沙漠区地震资料处理所取得的经验与认识,通过对塔中地区地震资料干扰波成因、类型及其特点进行分析,提出了根据噪声类型、能量强弱、频带范围及其在不同域中所表现的特征,分类、分步、分频、分域、分时窗、分区的六分法多域组合去噪技术思路与处理流程。实际资料应用效果表明,多域组合去噪技术能明显提高奥陶系内幕反射的信噪比以及小断点、碳酸盐岩内部异常或“串珠”状反射的清晰度。

Deep Ordovician carbonate reservoir is the main exploration target in Tazhong area.The undulate dunes,the loose surface sand bed and the serious absorption & attenuation effect leads to low SNR seismic data of the target layers.The noise caused by sand dunes affects the characteristics of the effective wave,and also breaks the law of the coherent noise,which increases the difficulty for seismic data denoising processing.In order to suppress interference wave and improve the SNR of Ordovician target layers with the premise of fidelity,based on the experience and recognition obtained from the desert seismic data processing in the last few years,in terms of noise type,energy strength,frequency bandwidth and their characteristics in different domain,we proposed multi-domain composite denoising technique and processing workflow,and the multi-domain includes scale,step,frequency,domain,time window,zone etc.Actual data application results indicate that mlti-domain composite denoising technique obviously improve the SNR of Ordovician insider reflections,the clearness of small fault points,inside abnormal of carbonate or moniliform reflection.

复杂地质条件下的超深层碳酸盐岩断控油气藏是近年来油气勘探的热点之一,其所发育的超深层断控缝洞型储集体受走滑断裂控制。沙漠起伏地表造成地震波能量吸收衰减严重,储层的上覆地层的火成岩严重影响目的层成像,同时超深层储集体地震波有效信号弱、反射特征复杂,影响了预测精度。以塔里木盆地顺北油气田为例,对该地区储集体高精度成像及储层预测难点进行剖析,建立了针对性技术策略:采用以弱信号恢复为代表的地震资料预处理技术提升叠前道集品质,为速度建模及成像提供基础数据;将火成岩高斯束层析速度建模与宽频RTM成像技术相结合,有效提升火成岩及高陡构造的建模成像精度;建立了面向超深层断裂及储集体的多属性综合预测技术流程,通过模拟断控缝洞体的地震波场响应,分析不同类型储集体的地震响应特征,对断裂带边界及内幕的叠后多属性进行降维处理,实现对超深层碳酸盐岩断控缝洞体的有效预测,为塔里木盆地超深层碳酸盐岩油气勘探开发提供了技术支撑。

Oil and gas reservoirs in fault-karst,ultra-deep carbonate rocks under complex geological conditions havebecome a popular exploration target in recent years.The fault-karst reservoirs are controlled by strike-slip faults.The undulating surface of the desert causes significant energy absorption and attenuation of seismic waves,the effectivesignal of seismic waves in ultra-deep reservoirs is weak and the reflection characteristics are complex,which affects the prediction accuracy.Simultaneously,the igneous rocks in the strata overlying the reservoirs affect the imaging of the target layer.The difficulties of high-precision reservoir imaging and prediction are analyzed using the Shunbei oil and gas field in the Tarim Basin as an example.The seismic data pre-processing technology represented by weak signal recovery is adopted as a targeted technical countermeasure to improve the quality of pre-stack gathers,thus providing basic data for velocity modeling and imaging.The modeling and imaging accuracy of igneous rocks and high and steep structures can be effectively improved using the combination of igneous rock Gaussian beam tomography velocity modeling and broadband reverse-time migration (RTM) imaging technology.A multi-attribute comprehensive prediction process which analyzes the seismic response characteristics of different types of reservoirs is established for ultra-deep fractures and reservoirs,through seismic wave field response simulation.This uses post-stack multi-attribute dimensionality reduction to process the fault zone and its boundaries and predict ultra-deep carbonate fault-karst reservoirs.This method provides improved technical support for the exploration of ultra-deep carbonate rocks in the Tarim Basin and the development of fault-karst reservoirs in the area.

Ray-based tomography in the imaging domain, implemented with seismic migration, is currently widely used in industrial applications. However, conventional ray-based tomography has some inherent problems, such as shadow area, multi-path problem and so on, which limit the inversion accuracy. To alleviate these problems, we proposed Gaussian beam velocity tomography (GBT) based on azimuth-opening angle domain common imaging gathers (ADCIGs). According to the first-order Born and Rytov approximations, we derived a linear relationship between travel-time perturbation and velocity perturbation in the imaging domain, by which we construct the explicit expression of the sensitivity kernel function and use a Gaussian beam operator to compute the kernel. Furthermore, by introducing the preconditioned model regularization, a method of GBT under the constraint of a structure-guided filter is derived. Iterative applications of migration and tomography, both based on a Gaussian beam propagator, embody the idea of integrating velocity inversion and imaging. Numerical tests on both synthetic data and field data demonstrate that Gaussian beam propagator-based travel-time tomography in the imaging domain is effective.

地震勘探的关键是速度分析,常用的叠前深度偏移速度建模技术为网格层析反演,但当研究区断层较为发育时,网格层析跨越断层进行,速度更新导致断层两侧速度描述不准确,从而产生断层阴影构造假象,对解释及井位部署产生影响。为此,从井震标定和精细构造解释入手,首先建立测井速度与地震层速度的关系,根据层位标定结果在时间域利用VSP速度及测井速度校正沿层层速度;然后将校正后的时间域层速度变换到深度域,将得到的结果作为叠前深度偏移的初始速度;再采用层控、断控高精度网格层析技术进行速度建模,利用深度域解释的层位和断层控制断层两侧速度的准确性;最后采用小网格层析技术提高垂向速度分析的精度,对断层两侧速度进行准确刻画。实际资料处理结果表明利用井控与构造约束条件下的网格层析速度建模技术建立的速度模型符合实际情况,叠前深度偏移结果可以较好地消除断层引起的下部构造假象。

<p>Seismic velocity analysis is important for geophysical exploration.Grid tomography inversion is a common technique to estimate the velocity of seismic waves in geological formations.If faults are abundant in the domain under investigation,and the tomography is carried out in a direction transverse to the faults,the seismic wave velocity on the two sides of the faults cannot be described with sufficient accuracy.This phenomenon is known as &ldquo;fault shadowing&rdquo; and can have a great impact on the interpretation,and ultimately on the locations chosen for well deployment.In this work,on the basis of seismic calibration of well logs and fine-structure interpretation,a relationship between well logging velocity and seismic interval velocity was established.Accordingly,VSP and logging velocities were used to correct the estimated velocities along the horizon in the time domain.Then,these velocities in the time domain were transformed into depth domain information,so as to serve as the initial velocity field in the pre-stack depth migration.Subsequently,high-precision grid tomography&mdash;entailing formation-controlled and fault-controlled techniques&mdash;was carried out to obtain the velocity.The information on the faults and that on the horizons obtained in the depth domain were used to invert the velocities on both sides of the faults.Meanwhile,the small-grid tomography technique was also used to improve the vertical accuracy of the velocity field.By this technique,the velocity on both sides of the faults could be accurately described.Results obtained by processing actual data showed that the velocity model obtained by the aforementioned technique was accurate,and the pre-stack depth migration can eliminate the fault shadowing phenomenon effectively.</p>

断裂发育地区地震速度建模困难,但其精度决定着成像结果的精度和可信度。将断裂属性作为先验信息应用于正则化约束中,针对人工解释断裂信息精度不高、重复工作量大的难题,以高斯束层析反演理论为基础,研究了基于断裂属性约束的高斯束层析速度建模技术。该方法通过引入基于最优路径寻优的断层特征提取技术,利用数据驱动得到高精度断裂信息,结合预条件正则化技术将断裂属性应用于层析反演迭代过程中,建立深度域断裂属性约束高斯束层析速度建模的地质框架,得到更为精确的速度模型,所得结果能够更加准确地刻画断裂两侧的地层速度,从而提高断裂的成像精度。理论模型数据及实际地震资料测试结果表明该技术反演得到的速度模型更加精细,相比于常规网格层析结果,断裂成像效果好,断面干脆,断点清晰,验证了方法的有效性与实用性。

Fault-controlled reservoirs have attracted increasing attention as the main exploration targets.Fault-constrained velocity model building and seismic imaging are of great significance for exploration.To improve the accuracy of the velocity model in fault developed areas,fault information has been introduced into a velocity building model in the depth domain based on the Gaussian beam theory.Furthermore,fault-controlled Gaussian beam tomography has been deduced to improve model accuracy,which depends on the accuracy of the information on fault attributes.By introducing a fault detection algorithm based on optimal surface voting,high-accuracy fault information was obtained in a data-driven manner.By combining this algorithm with pre-condition regularization to constrain the Gaussian beam tomography,a fault-constrained velocity building model in the depth domain was established.The proposed model could improve the accuracy of velocity modeling in fault developed regions.Testing on synthetic and actual field data confirmed the validity and practicability of the proposed method.

谱元法综合了有限元法边界适应性强和谱方法精度高、收敛快的优点。研发了一种基于谱元法的起伏地表逆时偏移方法,采用切比雪夫谱元法并结合隐式Newmark时间积分法求取波场传播算子,推导出波动方程切比雪夫谱元逆时偏移算法(CSE-RTM),该算法在空间上具有谱精度,在时间上达到二阶精度。同时,为了充分利用计算机集群资源,提高该方法的并行效率,基于自由度凝聚和局部松弛的思想,推导了切比雪夫谱元逆时偏移方法的多级并行算法(HEP-CSE-RTM),该算法具有并行效率不随处理器数目增多而下降的优点。二维加拿大起伏地表逆掩断层模型及三维SEAM起伏地表复杂构造模型测试结果表明:该方法在起伏地表复杂构造区成像精度明显高于常规有限差分法逆时偏移,由于该方法本身具有谱方法的稀疏网格快速收敛特性,结合自由度凝聚的多级并行计算,运行效率大幅度提高,达到了与常规有限差分法逆时偏移相当的计算效率。