【作者】 王万银；

【导师】 朱光明；

【作者基本信息】 长安大学 ， 地质工程， 2009， 博士

【摘要】 重、磁勘探方法是应用最早的地球物理勘探方法之一,使用的前提条件是被探测对象与围岩在横向上具有一定的密度或磁性差异,通过观测重、磁力异常可以研究地质体的几何参数或物性参数。在利用重、磁力异常研究地质体的几何参数时有三个阶段,第一阶段是研究地质体边缘的平面位置,第二阶段是研究地质体边缘的垂向位置(深度),第三阶段是研究地质体上、下界面的起伏变化。第一阶段称为边缘位置识别,第二阶段称为边缘参数反演,第三阶段称为形体界面反演。上述三个阶段相互独立、又相互依存,边缘位置识别是基础、边缘参数反演是过渡、形体界面反演是结果。因此对边缘识别方法的理论研究不但可以解决地质体边缘位置识别问题,而且能为边缘参数反演和形体界面反演的理论研究奠定基础。另外,边缘识别技术措施的研究不但可以更好地解决地质体边缘位置识别问题,还可以促进位场理论研究的进步。因此,边缘识别方法和识别技术的研究具有重要的理论意义。在我国的区域重、磁力勘探,矿产、能源等重、磁力勘探中,均需要识别研究区的断裂和岩体边界的分布位置,故位场边缘识别的研究具有重要的实际应用价值。本文依托“十五”国家科技攻关计划项目——大型紧缺金属矿产资源基地综合勘查与高效开发技术研究之专题“复杂地形区中大比例尺高精度航磁方法技术研究”(专题编号：2001BA609A—5—3)、国家科技重大专项——大型油气田及煤层气开发专项“海洋深水区油气勘探关键技术”项目(项目编号：2008ZX05025)和国土资源部全国油气资源战略选区调查与评价专项第一批项目(项目编号：XQ-2007-03和XQ-2007-05)及第二批项目(项目编号：2009GYXQ03、2009GYXQ05、2009GYXQ06和2009GYXQ09),以位场边缘识别系统化、自动化为研究目标,以解析分析为研究手段,对位场边缘识别方法和技术进行深入的理论研究和实际应用研究。主要研究成果如下：(1)利用Green函数(位函数)的导数形式表示重、磁力异常及其导数,并推导了铅垂台阶、倾斜台阶、直立平行四边形、梯形台阶、直立六面体、水平圆柱体和球体的Green函数一阶、二阶、三阶、四阶混合导数表达式,并由此得到以上简单规则形体的重、磁力异常及其导数表达式,为边缘识别方法的理论研究奠定了基础。(2)对数值计算类位场边缘识别方法的函数性质进行了研究,并对边缘识别方法进行了分类,指出“解析奇点”是造成数值不稳定的根源。研究结果表明：边缘识别方法可分为一阶导数、二阶导数和高阶导数三类,最基本的边缘识别方法为垂向导数、总水平导数和解析信号振幅,其他方法均在这三种方法的基础上发展而来。在所有的边缘识别方法中,只有垂向导数方法依然保持位函数的性质,而其他方法均不能保持位函数的性质。以位函数为原函数的边缘识别方法无“解析奇点”,其计算结果稳定；而以非位函数为原函数的边缘识别方法存在“解析奇点”,其计算结果稳定性较差。因此,位场边缘识别方法的研究重点是以位函数为原函数的各种边缘识别方法,即基于重力异常和基于重力异常垂向导数的各种边缘识别方法的研究；对于磁力异常,应将其转换成磁源重力异常或化极磁力异常。磁源重力异常与重力异常的边缘识别相同,化极磁力异常与重力异常垂向导数的边缘识别相同。(3)研究了重力异常垂向导数零值位置、总水平导数极大值位置和解析信号振幅极大值位置的空间变化规律。研究结果表明：对单一边界模型,所有边缘识别方法的零值位置或极大值位置均能准确识别直立边界的边缘位置,但不能准确识别倾斜边界的边缘位置；在倾斜边界上,解析信号振幅极大值位置距上顶面边缘位置的偏移量最小,其次是总水平导数极大值位置,而垂向导数零值位置的偏移量最大；不论是零值位置或极大值位置均随着地质体埋深的增加自上顶面边缘位置向下底面边缘位置移动,偏移量不超过倾斜界面宽度的一半。对双边界模型,边缘识别方法零值位置或极大值位置的空间变化规律与单一边界模型的空间变化规律截然不同：当地质体埋深较浅时,垂向导数零值位置、总水平导数极大值位置和解析信号振幅极大值位置的空间变化规律与单一边界模型的空间变化规律相同；但当地质体埋深达到某一深度时,解析信号振幅极大值位置消失,出现“极大值位置盲区”,无法识别地质体的边缘位置；随着地质体埋深的进一步增加,垂向导数零值位置和总水平导数极大值位置均自地质体上顶面边缘位置向外侧移动,垂向导数零值位置的偏移量等于地质体的平均埋深,而总水平导数极大值位置的偏移量为地质体平均埋深的1/(?)倍；而随着地质体埋深的进一步增加,解析信号振幅极大值位置自地质体上顶面边缘位置向内侧收敛,并很快收敛到地质体中心位置,无法识别地质体的边缘位置。在多个边界模型上,垂向导数零值位置、总水平导数极大值位置和解析信号振幅极大值位置的空间变化规律与双边界模型上的空间变化规律相同。因此,从边缘识别结果的偏移量大小来比较,总水平导数的偏移量最小,其次为垂向导数,而解析信号振幅的偏移量最大。(4)通过组合模型重力异常垂向导数零值位置、总水平导数极大值位置和解析信号振幅极大值位置的空间变化规律研究这三种方法的分辨能力。研究结果表明：组合模型重力异常垂向导数零值位置、总水平导数极大值位置和解析信号振幅极大值位置均受单一模型零值位置或极大值位置的影响,分辨能力受地质体埋深、水平尺寸和边界形状等因素的影响。总水平导数的分辨能力最强,其次为垂向导数,而解析信号振幅的分辨能力最差。(5)研究了基于重力异常垂向导数边缘识别方法——垂向二阶导数零值位置、垂向导数总水平导数极大值位置和垂向导数解析信号振幅极大值位置的空间变化规律以及上述方法的分辨能力。通过理论研究表明,基于重力异常垂向导数比基于重力异常的边缘识别结果偏移量小、分辨能力强,并发现垂向导数总水平导数存在“次级极大值”的影响。由于该幅值较小,故对边缘识别结果的影响也较小。(6)提出了最小曲率网格数据扩边和补空技术,频率域偶层位曲面位场处理和转换技术以及边缘识别技术(峰值锐化技术、阈值技术、归一化增强技术),上述技术使边缘识别逐步向系统化、自动化和高精度目标迈进。理论模型试验和实际资料处理研究结果表明,本文提出的最小曲率扩边和补空技术,降低了频率域处理中Gibbs效应的影响,提高了频率域处理和转换结果的精度；提出的频率域偶层位曲面位场处理和转换技术,不但可以将曲面数据换算到平面(“曲化平”),而且在“曲化平”的过程中可以直接计算位场的各种导数用来进行边缘识别,并能同时完成磁源重力异常或化极磁力异常的转换；提出的边缘识别峰值锐化技术和阈值技术,不但提高了边缘识别的分辨能力,而且突出了地质体边缘的影响,使得图面简单、清晰,便于识别；提出的归一化增强技术,突出了规模小、物性差异小、埋深大的地质体边缘影响。(7)利用卫星测高重力异常,并结合航空磁力异常对珠江口盆地韩江凹陷及邻区的断裂分布位置进行了推断解释,并进行了构造单元划分。研究结果表明,卫星测高重力异常精度高于1：100万比例尺重力勘探工作的要求,接近1：50万比例尺的精度。通过断裂识别发现在韩江凹陷中部存在一个明显的北西西向凸起,该凸起及两侧的断裂将韩江凹陷分成南、北两个部分。南部的断裂方向为北东向,北部的断裂方向为北北东向：南部的沉积以新生代为主,北部的沉积以中生代为主,是一个中、新生代叠合沉积凹陷。这一研究成果为韩江凹陷结构、构造和油气资源分布等研究提供了新的依据。更多还原

【Abstract】 The gravity and magnetic exploration is one of the earliest geophysical exploration methods. We can study the geometric or physical parameters of geological body by the measured gravity and magnetic potential field data, and the prerequisite for the use is there have certain differences in the density or magnetic properties in horizontal between the detected objects and the country rock. The research on the geometric parameters consists of three stages:firstly, to determine the edge plane position of the geological body, known as edge position recognition; secondly, to study the vertical position (depth) of geological body edge, known as edge parameters inversion; and the last is to research the geological interface fluctuation, known as geological interface inversion. The above three steps are mutually independent as well as mutually related. The edge position recognition is the basic problem, the edge parameter inversion is the transitional period and the geological interface inversion is the final aim. Therefore, the theoretical study on the edge recognition methods can not only solve the edge recognition problems of geological bodies, but also lay a foundation for the further theoretical study on the edge inversion and interface inversion. In addition, the research on the technology and measures of edge recognition can not only better solve the edge recognition problems, but also promote the progress of potential field theory. Therefore, the researches on the edge recogniton methods and recogniton techniques have important theoretical significance. In China, the regional gravity & magnetic exploration, mining and energy resource gravity & magnetic exploration, all need to identify the fractures distribution and the rock boundary in the measured zone. So the potential field research on edge detection has important actual significance and practical application value.The research relies on the 10th five years programs for science and technology development of China—Large Scarce Mineral Resource Base Comprehensive Exploration and Efficient Development Research Topics of "Research on middle-and large-scale high-resolution aeromagnetic methods and techniques in complex terrain area" (Item No. 2001BA609A-5-3). Relies on National major special science and technology project—Large Oil & Gas Fields and Coal-bed Methane Development of "Key technology of deepwater oil and gas exploration" (Item No.2008ZX05025). This work is also supported by the project of national oil and gas resources strategic constituency survey and evaluation of the Ministry of Land and Resources (Item No.XQ-2007-03 and XQ-2007-05) and the second-stage project (Item No.2009GYXQ03,2009GYXQ05,2009GYXQ06 and 2009GYXQ09). The main aim of this paper is to research the automatization and systematization of edge recognition technology for potential field data. By means of analytic analysis, a profound theoretical and practical application research on edge recognition methods and techniques for the potential field data has been carried out. The main research results and conclusions are as follows:(1) Expressed the computational formula of gravity & magnetic anomalies and their derivative in the derivative of Green function (potential function). And Derived the mixed derivative expressions of Green function of first-, second-, third-or forth-order for the vertical step, tilted step, vertical parallelogram, ladder-step, vertical hexahedron, horizontal cylinder and sphere. Finally, we obtained the gravity & magnetic anomalies and their derivative expressions of the above simple and regular geological bodies. These results provide theoretical basis for further research on the potential field edge recognition.(2) Studied the function properties of the potential field edge recognition methods belong to numerical compution object, classified the edge recognition methods, and at last, and pointed out the "analytic singularities" is the main reason causing poor numerical stability. The research results show that the edge recognition methods are classified into the first-order derivatives, second-and higher-order derivatives by the orders of derivative. The most basic edge recognition methods are vertical derivative, the total horizontal derivative and amplitude of analytic signal, and the other methods are developed on the above three methods. In all of the edge recognition methods, only the vertical derivative method remains the nature of the potential functions, the others can not. For the edge detection method with potential function as the original function, there hasn’t the "analytic singularities", and the results of edge recognition methods are stable. Otherwise, there has the "analytic singularities" and the results are unstable. Therefore, the research focus on the various edge detection methods with potential function as the original function, in other words, on the various edge recognition methods based on the gravity anomalies and the gravity anomalies vertical derivative. For the magnetic anomaly, the data should be converted into pseudo-gravity anomaly or the RTP magnetic anomalies. The edge recognition methods used for pseudo-gravity anomalies similar to that used for gravity anomaly, while the edge recognition methods used for the RTP magnetic anomalies similar to that used for gravity anomaly vertical derivative.(3) Studied the spatial variation rule of the zero value position of gravity anomalies vertical derivative, the maximum value position of total horizontal derivative and the maximum value position of analytic signal amplitude. Research results show that, For the models with a single boundary, all of the edge recognition methods can accurately recognize edge location of the vertical border by the means of the zero value or the maximum value position, but can not to the single-border tilted body. For the models with a tilted boundary, the maximum value position of analytic signal amplitude has the least deviation from the top border location, followed by the maximum value position of total horizontal derivative, while the zero value position of the vertical derivative has the maximum deviation. Either the zero value position or the maximum value position, move from the top border position to bottom with the geological body depth increasing, and the deviation is no more than half width of inclined interface. For the models with double boundaries, the spatial variation rule of zero value position and maximum value position takes a huge difference with that of the models with single boundary. When the depth of geological body is shallow, the zero value position of vertical derivative, the maximum value position of total horizontal derivative and the maximum value position of analytic signal amplitude behave the same spatial variation rule as that of the models with single boundary. When the depth of geological body increases to a certain depth, the maximum value position of analytic signal amplitude disappears, which is called " the blind zone of maximum value position " and lead to the failure of edge recognition. And with the depth increases, the zero value position of vertical derivative and the maximum value position of total horizontal derivative all move from the top border position of the body to the outside, and the deviation of the zero value position of vertical derivative is the average buried depth of the geological body, the deviation of the maximum value position of total horizontal derivative is of 1/√3 times of the average buried depth. And with the depth further increases, the maximum value position of analytic signal amplitude converges from the top border to the inside, and quickly converges to the center of the body, which also leads to the failure of edge recogniton. For the models with multi boundaries, the zero value position of vertical derivative, the maximum value position of total horizontal derivative and the maximum value position of analytic signal amplitude behaves the same spatial variation rule as that of the models with double boundaries. Contrasting from the deviation, total horizontal derivative has the least deviation, followed by the vertical derivative, while analytic signal amplitude has the most.(4) Studied the recognition ability of the above-mentioned three kinds of basic edge recognition methods by the spatial variation rule of the zero value position of gravity anomalies vertical derivative, the maximum value position of total horizontal derivative and the maximum value position of analytic signal amplitude for the combined-models. The research results show that, for the gravity anomalies of combined-models, the zero value position of vertical derivative, the maximum position of total horizontal derivative and the maximum position of analytic signal amplitude were influenced by the zero value position or the maximum value position of the single model. And the recognition ability is subject to the following factors, such as buried depth, the size in horizontal and boundary geometry, and so on. In a word, total horizontal derivative shows the best recognition ability, followed by vertical derivative, and analytic signal amplitude has the worst ability.(5) Studied the spatial variation rule and the recognition ability of the edge recognition methods based on the gravity anomaly vertical derivative, such as the zero value position of the second-order vertical derivative, the maximum value position of total horizontal derivative of the vertical derivative, the maximum value position of the vertical derivative of analytical signal amplitude. The theoretical study results show that the edge recognition based on the gravity anomaly vertical derivative has smaller derivation and more power resolution than that based on the gravity anomaly. And the results of total horizontal derivative of vertical derivative are affected by the "secondary maximum value". The "secondary maximum value" is small and can be neglected.(6) Put forward the extending edge and interpolation technology based on the minimum curvature method, the frequency domain dipole layer method for the processing and transformation of potential field on curved surface and the edge recognition technologies (the peak sharpening technology, the threshold technology and the normalized enhancement technology). The above technologies enable edge detection step by step to the systematic, automated and high-precision development. From the results of the models and field data, the extending edge and interpolation technology, presented in this paper, overcomes the Gibbs effect and improves the results accuracy of the processing and transformation in the frequency domain. The proposed frequency domain dipole layer method for the processing and transformation can not only reduce the potential field data to a horizontal plane, and at the same time, directly calculate variety kinds of derivatives used to recognite the edges of geological bodies, but also make the processing and transformation of pseudo-gravity anomaly or the RTP magnetic anomalies. The proposed peak sharpening technology and threshold technology can not only enhance the ability of distinguish the edge, but also highlight the effect of geological body boundaries, making the map more simple and clear, and easier to identify. The adopted normalized enhancement technology highlights the geological body with small-scale, delicate-difference in physical properties and deeper-depth.(7) Interpreted the fault distribution and divided the tectonic units of Hanjiang depression and it’s adjacent regions by use of the gravity anomaly from satellite altimetry and the aeromagnetic anomaly. The research results show that the accuracy of the gravity anomaly from satellite altimetry is higher than 1:100 million scale gravity exploration work requirements, and is close to 1:50. Through the fault identification and division of tectonic units, the results indicate that there is a clear NWW uplift in the middle of Hanjiang depression. And the faults, on both sides of the uplift, divided the Hanjiang depression into north and south parts. The NE orientation faults are in the southern, and the NNE orientation faults are in the northern. The Hanjiang depression, with the southern sediment dominated by Cenozoic and the northern dominated by Mesozoic, is an overlaid sedimentary depression of Mesozoic-Cenozoic. Our research results provide a new basis for the study of the Hanjiang depression structure, construction and the distribution of oil-gas resources.更多还原