【作者】 韩兆红；

【导师】 吴燕冈；

【作者基本信息】 吉林大学 ， 地球探测与信息技术， 2011， 博士

【摘要】 本文的研究内容主要包括利用重磁数据进行边界识别和寻找区域弱异常等项研究,同时通过理论模型试验加以证实,最后对具有复杂区域地质构造格局的齐齐哈尔地区重磁资料进行了相关处理,给出了初步地质解释,主要工作如下：1.提出了利用双方向区块扫描Lyapunov指数极小值确定物质区块分界线的方法,进行了有益的尝试,具体过程如下：(1)重力和航磁的网格化数据都满足行或列在同一直线上,以每条线为一个窗口,即以行或列作为单变量时间序列直接计算其Lyapunov指数。(2)把m×n数据矩阵按照行来划分,得到n行数列,可以计算出每一行对应的一个Lyapunov指数值；相似地按照列来划分,可以得到每一列所对应的一个Lyapunov指数值。这样实质上就是得到两个Lyapunov指数序列。(3)在所得到的两个Lyapunov指数序列中,行和列两个方向交汇处的每一个点都存在唯一的一对Lyapunov指数值,当满足双方向的Lyapunov指数值同时达到极小时,这个点的坐标就是所计算曲面拐点或极值的平面位置。2.通常重磁异常曲面的拐点或极值的连线往往对应地下地质体物性的分界面或轴线在地表的投影。为了检验上述Lyapunov指数方法的有效性,进行了台阶、水平圆柱体两个理论模型试验,并对齐齐哈尔地区实测重力数据进行了处理。理论模型试验结果与设计位置吻合,实际处理所得极小Lyapunov指数值点的连线与大兴安岭梯度带边界吻合。3.为了更有效地提取弱异常,文中总结了自相关滤波法在重磁数据处理中的优缺点。通过建立单球体和多球体模型试验发现,秦葆瑚1991年提出的自相关滤波因子具有基本消除了随机干扰的影响、能够在一定程度上反映区域异常的变化趋势、能显示异常有无的优点；但也存在滤波结果仍受区域场影响、提取的弱异常幅值太小且伴随有假异常、其滤波公式部分物理意义不明确等缺点。针对自相关滤波方法在提取局部弱异常中的缺陷,对一维和二维自相关滤波因子进行了适当改造,具体方法如下：(1)一维自相关滤波因子改造：在每条测线内自动寻找剩余局部异常最大点处的观测值做为参考模型,并沿整条测线进行滑动自相关处理。数据处理时,只需通过分析异常规模给出滤波窗口K,并让K等于最大异常范围半径,模型试验结果位置准确,减小了区域场的影响,异常相对强度明显变大,有效地突出了弱异常。当对含有随机干扰的数据进行处理时,不能完全消除随机干扰,虚假异常分布不对称,区域场大的一侧虚假异常强度较大,处理结果有窗口滑动痕迹。(2)二维自相关滤波因子改造：改变自相关滤波函数ρ的系数,以及四周测点的平均值Z(l)的表达式。二维自相关滤波只适用于线距等于点距的情况,对于其它类型数据必须先网格化成正方形网格；自相关半径选取在3～6个网格节点距为佳,异常零值线能较好反映异常体范围；对随机干扰压制效果有所改善,不仅消除了背景影响,有效地提取了弱异常,还可以通过零值线大体划出异常边界；高值突变点对本方法有较大影响,所以要求对原始资料先作剔除畸变点的预处理。4.为了检验改造后自相关滤波因子的效果,对齐齐哈尔地区实测航磁数据进行了处理,结果能很好地确定出各局部异常位置,边界划分清晰,基本提出了强区域场内的所有弱异常。但由于自相关滤波因子对随机干扰压制能力有限,不能完全消除它们,结果中可能还存在一定虚假异常。选择用匹配滤波和自相关滤波进行对比分析,发现处理结果有区域场影响小、局部异常独立性强的特点,更好的反映了该区局部异常的分布状况。按局部弱异常密集程度将该区大致分为两个区：第一区属大兴安岭华力西褶皱带,局部弱异常受区域地质构造控制,构造线多数沿NE～NNE向延伸；第二区为松嫩平原地带,地磁场相对平稳,无明显磁异常。其中A处具有较大研究价值,这里磁异常反应强烈,相对集中,对此地可做为找矿远景区进一步勘探。最后绘制了构造特征线图。5.介绍了分形几何的理论知识、分形与混沌成矿的关系,总结了分形理论在重磁数据处理中的应用。由于分形维数和Lyapunov指数都是描述非线性系统特征的量,所以把双方向Lyapunov指数极小值识别物质区块分界线的思想应用到分形理论中对地质体进行边界识别,这是将来本人要做的后续工作之一更多还原

【Abstract】 In this paper, we mainly use the gravity and magnetic data to identify the border and extract the weak anomalies and then prove it with theoretical models. Finally, we consider the gravity and magnetic data in Qiqihar as the research object, which has complex geological constitution, and then give some geological interpretation. The main parts of my work are as follows:1.We propose a method to determine the boundary by scanning the smallest Lyapunov index on both directions and made a useful attempt. The specific process is as follows:(1) The gravity and aeromagnetic grid data meets in the same line along rows or columns. We select each line as a window, that is, row or column is selected as a single variable time list to calculate the Lyapunov index directly.(2) Divide the data matrix m x n by line to get n rows, and each row can be calculated, that is, there is only one Lyapunov index corresponding to every latitude,which is the same with every longitude according to the column. Two Lyapunov index sequences are attained essentially.(3) We can see from the two Lyapunov index sequences that, every point has only one couple Lyapunov indexes. As we known for the definition of lyapunov indexes, when the two indexes are both the smallest of each sequence, the point should be inflexion of surface or plane location of extremum.2. Usually, link line of inflexions or extremums corresponds with the boundary surface or axes projection on ground of geological properties. To test the validity of the method above, the two theoretical models of step and Horizontal cylinder are established, and the Qiqihar gravity and magnetic data are processed. The result of theoretical model tallied with the location of design, and link line of lyapunov index minimums tallied with the boundary of Great Xing’an Mountains gradient zone.3. In order to extract weak anomalies more Effectively, advantages and disadvantages of the self-autocorrelation filters in gravity and magnetic data processing are summarized. Through testing the established single sphere and multi sphere gravity models, it is found that Baohu Qin’s self-autocorrelatioin filter factor could basically eliminate the random noise and could reflect the regional tendency at a certain extent in portrait as well as showing up the existence of anomaly in landscape in 1991.However, it has some disadvantages. For example, the result of the filter is still affected by the regional filed,the weak anomaly extracted is too small and accompanied false anomaly, some of the filter functions are ambiguous and so forth.1D and 2D self-autocorrelatioin filter factors are researched and modified to remedy these drawbacks that extracting local weak anomalies. The specific methods are as follows:(1) Modify 1D self-autocorrelatioin filter factor:by automaticly finding the point whose local residual anomaly was maximum in each line, considering it’s value as our reference model and making use of sliding autorelation processing in the whole line. In data processing, we only need to give the filter window K, which should be equal to the radius of the scale of the anomaly maximum, by analyzing the anomaly scale. The result of models is correct, and it does not only reduce the regional effect, but also highlight the weak anomaly. When the data is combined with random noise, it can not be completely healed, and falsehood anomalies distributed inhomogeneous, and falsehood anomalies is bigger on the side where has bigger regional field, and the result is scored by window sliding.(2) Modify 2D self-autocorrelatioin filter factor:we change the coefficient of self-autocorrelatioin function p and the expression of the mean value of the point all around Z(l).2D self-autocorrelatioin filter could only be applied to the case that line space equals to spot space, and the data has to be transformed to a network of square if not; Choosing 3～6 plus the grid distance as the self-autocorrelatioin radius is preferred, in the time the zero isoline could reflect the range of the geologic body well; The improved method has many advantages such as suppressing the random noise at a certain extent, eliminating the impact of background, extracting the weak anomaly, and outlining the boundary of geologic body by the zero isoline; As high saltation points will strongly influence the result, eliminating preprocessing is required.4.To test the effect of modified filter factor, we apply this method to the aeromagnetic data in Qiqihar. The result can well determine the location of every local anomaly, partition the boundary clearly and extract the weak anomaly on the field of strong anomaly basically. However, the random noise could not be eliminated absolutely as the self-autocorrelatioin filter factor does not suppress the random noise well, which will lead to the existence of false anomaly.By comparing matched filtering with self-autocorrelatioin filtering, it shows that the effect of regional filed is small and the local anomaly is independent, and therefore, the processing result could better reflect the distribution of local anomaly. According to the denseness of the weak anomalies, the area is divided into two parts. The first, whose local anomaly is controlled by the regional geological structure and most of the tectonic lines extend along between NE and NNE directions, belongs to the Variscan fold belt of Daxinganling:while the other one belongs to Songnen Plain, where the magnetic filed is stable without obvious anomaly. Because of it’s strong and concentrated magnetic anomaly, the aero marked A has some important scientific interests and can be future explored as a prospect area. Finally, the map of structural characteristic lines is drawn.5. Introduce academic knowledge of fractal geometry, the connection of fractal and chaos ore formation, summarize the application of fractal in gravity and magnetic data. Both the fractal dimension and lyapunov index characterize the non-linear system, so the mind that identifying the boundary by Lyapunov index on both directions can be used into fractal, that is, window sliding in fractal can be used for identifying the boundary of geological body, and this is one of my works forward.更多还原