迭代法与FFT法位场向下延拓效果的比较

将水平观测面上的实测位场值，垂直投影至下部的延拓水平面上，作为该水平面上的位场初始值. 根据该水平面上的初始值，用快速傅里叶变换(FFT)的方法向上延拓计算观测面上的位场值. 用观测面上的实测值与计算值的差值，对延拓面上的位场值进行校正. 如此反复迭代，直至观测面上的实测值与计算值的差值小到可以忽略. 这种空间域的迭代法原理简单，不用解线性代数方程组，有较高的计算速度和良好的延拓效果. 本文用迭代法对模型数据和实际数据进行向下延拓，对比了迭代法与常规的FFT法在位场向下延拓中的效果，迭代法显著优于FFT法.     

位场延拓的积分-迭代法

本文介绍一种新的位场延拓方法——积分-迭代法.将起伏面上的实测位场值，垂直投影至起伏面下部的一个水平面上，作为该水平面上的位场初始值.根据该水平面上的初始值，用积分方法计算起伏面上的位场值.用起伏面上的实测值与计算值的差值，对水平面上的位场值进行校正.如此反复迭代，直至起伏面上的实测值与计算值的差值小到可以忽略.有了水平面上的位场值后，就可以用积分的方法或其他方法计算水平面以上的任意曲面或水平面的位场值.该方法原理简单，不用解线性代数方程组，有较高的计算速度.它特别适用于位场向下延拓，有良好的延拓效果.本文还介绍了积分迭代法的应用实例.     

位场向下延拓的波数域迭代法及其收敛性

提出了位场向下延拓的波数域迭代法. 对水平面上的位场观测值进行Fourier变换,得到其波谱. 根据第一类Fredholm积分方程的空间域迭代解法,推导出计算向下延拓水平面上位场波谱的波数域迭代公式. 在波数域中进行迭代,一直进行到相继两次迭代近似解的差值最大绝对值小于给定的精度,或迭代达到给定的最大迭代次数. 对这种迭代近似解进行Fourier逆变换,得到向下延拓的位场. 数值计算结果表明：与空间域迭代法比较,这种波数域迭代法简单、快速,并有同样好的向下延拓效果. 本文还证明了这种迭代法是收敛的,并给出了它的收敛特性和滤波特性.     

位场数据曲化平的迭代法

位场数据曲化平是位场数据处理解释中的重要运算,但是它的计算量和计算的复杂性影响了它在许多处理和解释方法技术中的应用.本文提出一种位场数据曲化平的迭代方法,即通过把位场数据曲化平视为平面位场数据向上延拓的反问题,得到曲化平的线性积分方程,再把曲面上位场数据视为曲面平均高程面上的位场数据,利用向下延拓的波数域广义逆算法把平均高程面上的位场数据向下延拓到设定平面上,再根据曲面和其平均高程面的相对起伏对设定平面上的向下延拓数据进行起伏校正,最后再把所得平面上的位场数据向上延拓得到曲面上的位场数据,并进行迭代.把这种方法用于三维理论模型数据和实际磁场数据的曲化平处理均获得了理想的结果.     

位场曲化平积分方程的迭代解

提出了位场曲化平的新方法. 给定观测曲面S上的位场、S对下方水平面P的相对高程,确定P上的位场. 利用由P向上延拓到S的积分式,建立这两个面上位场及相对高程三者所满足的方程,它是第一类Fredholm积分方程. 用Fourier逆变换式把这一空间域积分式化为波数域积分式,再由指数函数的Taylor展开进一步化为级数式. 积分方程的解采用逐次逼近法迭代计算,即用S上的位场观测值作为P上位场的初始迭代值,用导出的级数式求得S上的位场计算值、由S上的位场观测值与计算值之差校正P上的位场,多次迭代,直到满足迭代终止准则. 我们还给出该积分方程的波数域迭代计算方法. 模型算例表明,重力异常曲化平的均方差和磁异常曲化平的均方差分别为0.0008 mGal和0.0019 nT,在主频为2.26 GHz的笔记本电脑运行,2048×2048数据量,计算时间是975 s. 野外磁场实际资料处理也证实这种方法的有效性.     

瞬变电磁虚拟波场的三维曲面延拓成像

根据瞬变电磁场扩散方程与波动方程间存在的数学对应形式,将已知瞬变电磁场数据转换为虚拟波场数据,在此基础上,提出了用克希霍夫积分法将所转换的波场从地面向地下反向延拓的方法.在延拓的数值计算中,为提高计算效率,采用三维边界元技术,把边界积分分解为诸三角单元积分的积分之和, 建立了曲面延拓方程式;通过地面测点的波场值,求出地下某一点的波场值,实现了地表为曲面的向下延拓成像计算.通过对理论模型计算和实际资料处理,证明了该方法可以增强瞬变电磁法识别地下电性分界面的能力,使瞬变电磁法对地下目标体的三维精细探测成为可能.     

串联F-K加相移插值法偏移

本文介绍的偏移成像方法为递归F-K法、相移插值法及串联偏移技术的有效结合,简称CFPI法.该方法中,波场外推采用相移插值法中递归向下延拓方式,而每一延拓层间的成像则用文中给出的常速F-K加剩余速度F-K偏移,通过与横向速度变化有关的插值来实现,该过程用FFT完成.该方法可采用较大的延拓步长速归向下外推与成像.理论分析及实际计算结果比较表明,该算法效率远高于相移插值法偏移.     

一种改进的基于导数迭代的磁场向下延拓方法

高精度的地磁场数据库是地磁匹配导航的基础.在构建地磁数据库时,需要将空中的航磁测量数据向下延拓至某一水平面,而向下延拓在数学上具有不适定性,通常利用迭代方法逐次逼近"适定最优解".本文针对导数迭代法在迭代过程中的噪声累积问题,提出一种改进方法,首先利用稳定的向上延拓将观测面和向上延拓、向下延拓同等高度平面上的磁场值联系...     

地质统计学在气象要素场插值的实例研究

对两种气象要素场数据分别用距离平方反比法、三次B样条和克里格（Kriging）法插值计算.比较了三种方法结果的差异和当计算场满足不同类型克里格数学假设前提下,普通克里格法（OK）与泛克里格法（UK）插值结果的异同.结果表明：克里格法的误差普遍偏小,且在插值区域峰值处克里格法的最大绝对误差和残差方差均可能较样条的小,说明只要充分了解研究区域特点,恰当选用参数,克里格法有可能得到优于样条的结果,而距离平方反比法和克里格法用全场数据插值不如使用局部数据插值的精度高,则表明内插计算具有局域性.同时还发现,虽然插值场是否满足克里格法假设对插值结果存在影响,但这种影响有时并不重要,它依赖于插值场的性质.     

波动方程深度偏移的频率相关变步长延拓方法

发展了波动方程深度延拓的频率相关变步长深度延拓方法和表驱动的单点波场插值技术.前者通过减少深度延拓的次数减少了波动方程深度偏移的计算量,而后者用很少的计算量实现了等间距、理想采样的深度成像.就同一偏移方法,采用频率相关变步长深度延拓加单点插值,其计算量大约是常规的等间距采样延拓方法的三分之一,但两者的成像效果基本相同.文中以最优分裂Fourier方法为例,用二维理论数据（Marmousi模型）和三维实际地震资料验证了这一方法,但这一方法可适用于各类频率域波动方程深度偏移方法.     

APPLICATION OF ELASTIC MODELLING IN PROCESSING AND INTERPRETATION OF VSP DATA: A CASE HISTORY1

Data from offshore Norway is used to study applications of elastic VSP modelling in detecting shear waves and observing the effects of successive mode conversion in field-recorded VSP data. The shear-wave velocities and densities from log data are used in conjunction with compressional wave velocities determined from surface seismic and log data in the VSP modelling. The time domain non-normal incidence elastic VSP modelling technique of Aminzadeh and Mendel is used as the modelling algorithm. Two surface seismograms are computed first. One is the vertical component and the other is the horizontal component for plane waves that have specified incident angles. A downward continuation method is then applied to generate seismograms at different depth points. The collection of these seismograms constitutes non-normal incidence VSPs. Both vertical and horizontal components of VSP data can be obtained by this procedure. In this paper non-normal incidence VSPs are generated for a 12.5° incident plane wave. The modelling results of layered earth systems of thin layers and thick layers are both compared with field data, and the effect of mode conversions in thin layers is observed. Several events in the field data can be explained by this elastic VSP modelling. Comparison of the model data and field data enabled a probable tube wave or out-of-plane event to be identified, the removal of which significantly improved the final VSP section. This study also shows how the VSP data helped the interpretation of the surface 3D data.     

Depth Estimates for Slingram Electromagnetic Anomalies from Dipping Sheet-like Bodies by the Normalized Full Gradient Method

The Normalized Full Gradient (NFG) method was proposed in the mid–1960s and was generally used for the downward continuation of the potential field data. The method eliminates the side oscillations which appeared on the continuation curves when passing through anomalous body depth. In this study, the NFG method was applied to Slingram electromagnetic anomalies to obtain the depth of the anomalous body. Some experiments were performed on the theoretical Slingram model anomalies in a free space environment using a perfectly conductive thin tabular conductor with an infinite depth extent. The theoretical Slingram responses were obtained for different depths, dip angles and coil separations, and it was observed from NFG fields of the theoretical anomalies that the NFG sections yield the depth information of top of the conductor at low harmonic numbers. The NFG sections consisted of two main local maxima located at both sides of the central negative Slingram anomalies. It is concluded that these two maxima also locate the maximum anomaly gradient points, which indicates the depth of the anomaly target directly. For both theoretical and field data, the depth of the maximum value on the NFG sections corresponds to the depth of the upper edge of the anomalous conductor. The NFG method was applied to the in-phase component and correct depth estimates were obtained even for the horizontal tabular conductor. Depth values could be estimated with a relatively small error percentage when the conductive model was near-vertical and/or the conductor depth was larger.     

火山岩油气藏重磁电震综合预测方法及应用

通过准噶尔盆地陆东地区数十口钻井资料的对比分析、归纳总结,提出了火山岩油气藏重磁电震综合预测方法.将正则化下延与延拓回返垂直二次导数串联形成了一个新的滤波器,该滤波器相当于首先通过正则化下延将位场曲面延拓至地下目的层段,降低火山岩埋深对磁J异常幅值的影响,然后利用延拓回返垂直二次导数对弱信号进行增强,不仅提高了位场异常...     

重力异常视深度滤波及应用

为了解观测面以下位场场源体产生的异常,设计了一种波数域的滤波器,该滤波器可在无需岩石物性资料的前提下用于位场数据的反演．对位场资料用解析法和差分法上、下延拓,以消除浅表干扰源的影响,并最后下延至观测面以下,根据下延后场的极值确定场源的质心深度．理论模型的试算和实际资料的处理均获得了令人满意的效果．     

The iteration method for downward continuation of a potential field from a horizontal plane

This paper introduces an iteration method for the downward continuation of potential‐field data from a horizontal plane, and compares it with the conventional frequency‐domain method (Fourier transform) using 2D and 3D model tests. The paper evaluates the two methods in terms of the results, i.e. downward‐continuation distance and stability. The iteration method proves to be more stable and able to downward continue the potential for a greater distance than the Fourier transform method.     

2D and 3D potential-field upward continuation using splines

The dominant upward‐continuation technique used in the potential‐field geophysics industry is the fast Fourier transform (FFT) technique. However, the spline‐based upward‐continuation technique presented in this paper has some advantages over the FFT technique. The spline technique can be used to carry out level‐to‐uneven surface 2D and 3D potential‐field upward continuation. An example of level‐to‐uneven surface upward continuation of 3D magnetic data using the spline technique is shown, and it is evident that the continued anomalies are very close to the theoretical values. The spacing can be irregular. Synthetic examples using the spline technique to continue noise‐contaminated gravity and magnetic data upward to an altitude of 15 km on irregular grids are shown. Gaussian noise with a zero mean and a standard deviation of 1% does not cause much error and can readily be tolerated. Through comparison with the FFT technique, it is found that for low‐altitude gravity and magnetic upward continuation, both the FFT technique and the spline technique are suitable; for high‐altitude upward continuation, the FFT technique is inaccurate, whereas the spline technique works very well. Also, upward continuation by the spline technique has a smaller edge effect than upward continuation by the FFT technique. The spline‐based upward continuation technique works fairly well even when the periphery of a grid is not quiet: it is rather robust in general. A real example shows that the spline technique can be employed to perform upward continuation of total‐field magnetic data and to de‐emphasize near‐surface noise.