Non-linear methods for event location in a global context

Directed search methods in four-dimensional hypocentre space do not require the differentiation of travel times and are particularly suitable for the use of information from multiple seismic phases including joint use of primary and depth phases (e.g. pP) for a reference model such as a k135. These approaches can be used with a wide range of measures of misfit between observed and estimated times. A very effective method of this class is the use of the neighbourhood algorithm that allows all previous hypocentre estimates to be employed to estimate the shape of the misfit surface. When used with a robust measure of misfit such as the sum of the absolute values of the residuals (L1 norm), the neighbourhood algorithm provides rapid convergence, and a measure of the uncertainty in the hypocentral estimate from the concentration of the points with low misfit in four-space. Such a procedure is well suited to the first pass analysis of observations since it is not distracted by occasional readings associated with other events. Phase association can then be improved and a more conventional procedure used to provide error ellipsoids around a well-defined depth.     

邻近算法在微地震震源位置和一维速度模型同时反演中的应用

微地震事件的空间分布可以用来监测水力压裂过程中裂缝的发育情况.因此,震源定位是微震监测中重要的环节.震源定位依赖准确的速度模型,而震源位置和速度模型的耦合易导致线性迭代的同时反演方法陷入局部极小值.邻近算法作为一种非线性全局优化算法,能够最大程度地避免陷入局部最优解.本文将邻近算法应用于单井监测的微震定位和一维速度模型...     

应力张量反演的遗传算法及其在青藏高原东北缘的应用

为准确而快速地求解区域构造应力张量,构建了基于震源机制解反演应力张量的遗传算法策略,详细介绍了计算原理,分析了以断层面上剪应力方向和滑动方向的偏差为约束反演应力张量而忽略剪应力大小偏差对反演结果的影响,表明仅考虑剪应力方向和滑动方向的偏差就可以得到正确结果.利用不同应力状态下人工合成的包含不同噪声水平的震源机制数据对该方法进行检验,并与网格搜索法得到的结果比较.表明本文方法所得结果的拟合差均小于网格搜索法结果的拟合差,并且二者相差较小,体现了本文方法的稳健性.将该方法应用于青藏高原东北缘(103-106°E,34.5-37.5°N)应力张量的估计,结果显示,该区域主要受控于青藏高原近东西向的挤压,从而导致阿拉善块体以及华南块体方向的拉张,并且向华南块体方向的拉张作用强于向阿拉善块体的拉张作用.     

Soil production and hillslope transport in mid‐latitudes during the last glacial–interglacial cycle: a combined data and modelling approach in northern Ardennes

The relative efficiency of various hillslope processes through Quaternary glacial–interglacial cycles in the mid‐latitudes is not yet well constrained. Based on a unique set of topographic and soil thickness data in the Ardennes (Belgium), we combine the new CLICHE model of climate‐dependent hillslope evolution with an inversion algorithm in order to get deeper insight into the ways and timing of hillslope dynamics under one such climatic cycle. We simulate the evolution of a synthetic hill reproducing the slope, curvature, and contributing area distributions of the hillslopes of a ~ 2500 km² real area under a simple two‐stage 120‐kyr‐long climatic scenario with linear transitions between cold and warm stages. The inversion method samples a misfit function in the model parameter space, based on estimates of the fit of topographic derivative distributions in classes of soil thickness and of the relative frequencies of the predicted soil thickness classes. Though the inversion results show remarkable convergence patterns for most parameters, no unique solution emerges. We obtain five clusters of good fits, whose centroids are taken as acceptable model solutions. Based on the predicted time series of average denudation rate and soil thickness, plus snapshots of the soil distribution at characteristic times, we discuss these solutions and, comparing them with independent data not involved in the misfit function, we identify the most realistic scenario. Beyond providing first‐order estimates of several parameters that compare well with published data, our results show that denudation rates increase dramatically for a short time at both warm–cold and cold–warm transitions, when the mean annual temperature passes through the [0, ?5 °C] range. We also point to the overwhelming importance of solifluction in shaping hillslopes and transporting soil, and the role of depth‐dependent creep (including frost creep) throughout the climatic cycle, whereas the contributions of simple creep and overland flow are minor. Copyright © 2016 John Wiley & Sons, Ltd.     

Determination of Stress State from Focal Mechanisms of Microseismic Events Induced During Hydraulic Injection at the Hijiori Hot Dry Rock Site

—?The stress state at the Hijiori hot dry rock site was estimated based on the inversion from focal mechanisms of microseismic events induced during hydraulic injection experiments. The best fit stress model obtained by inverting 58 focal mechanisms of seismic events simultaneously indicates that the maximum principal stress σ₁ is vertical, while the minimum principal stress σ₃ is horizontal and trends north-south. The average misfit between the stress model and all the data is 6.8°. The inversion results show that the average misfit is small enough to satisfy the assumption of homogeneity in the focal mechanism data and that the 95% confidence regions of σ₁ and σ₃ are well constrained, i.e., they do not overlap, suggesting that the inversion results are acceptable. The stress estimates obtained by the focal mechanism inversion essentially agree with other stress estimates previously obtained. It is therefore concluded that the focal mechanism inversion method provides a useful tool for estimating the stress state. The hypocentral distributions of microseismic events associated with the hydraulic fracturing experiments are distributed around the plane that spreads to almost east–west from the injection wells and declines to the north at a high angle. The vertical orientation and east–west strike of the seismic events are essentially coplanar with the caldera ring-fault structure in the southern portion of the Hijiori Caldera. This indicates that tensile fractures of intact rock were not being created, but pre-existing fractures were being re-opened and developed in the direction of the maximum horizontal principal stress, although microseismic events were caused by shear failures.     

基于新的惩罚因子算法的波场重构反演

全波形反演是一种高精度的反演方法,其目标函数是一个强非线性函数,易受局部极值影响,而且反演过程计算量较大.波场重构反演是近几年提出的一种改进的全波形反演理论.该反演方法通过将波动方程作为惩罚项引入到目标函数中,通过拓宽解的寻找空间减弱了局部极小值的影响,而且反演过程不需要计算伴随波场,提高了计算效率.但该反演方法一直缺少准确的惩罚因子算法,直接影响到该方法的准确度.本文将波场重构反演拓展到时间域并利用梯度法进行波场重构.频率域的惩罚因子用来加强波动方程的约束,而时间域惩罚因子表现为调节模拟波场和实际波场的权重因子.为此,我们根据约束优化理论,在波动方程准确以及重构波场与反演参数解耦的假设下,提出以波动方程为目标函数的新的惩罚因子算法.根据波形反演在应用时普遍存在的噪音干扰、子波错误和低频信息缺失的情况下,应用部分Sigsbee2A模型合成数据对本文提出的算法进行实验.数值实验结果表明：基于新的惩罚因子算法,在其他信息不准确的情况下,波场重构反演可以给出高精度的反演结果.     

Hybrid genetic algorithms in view of the evolution theories with application for the electrical sounding method

The classical genetic algorithm is a stochastic process which operates by natural selection. Although the algorithm may localize a point around the global minimum of the misfit function, it is not efficient at finding the precise solution. This paper suggests some hybrid genetic algorithms, derived from evolution theories, to overcome this problem. Firstly, sexual selection has been incorporated in the classical genetic algorithm to obtain a full representation of the Darwinist evolution concept. The simulation of sexual selection is performed by assigning a higher probability of surviving to some parameters that satisfy some algebraic relationships. This method is called the ‘marked constraints’ algorithm since it permits us to insert geological and geophysical constraints into the problem. The algorithm implementation is realized by progressively shrinking the parameter search space through successive generations. In this way, the genetic algorithm gains some degree of determinism. Secondly, since the evolution theory of Lamarck postulates that the acquired traits are passed on to the next generation, a hybrid use of the damped least‐squares method and the genetic algorithm is called Lamarckian inversion. Lamarckian inversion involves some improvement procedures that simulate the reduction of the misfit with the help of a derivative‐based method between two generations. Finally, although there is no correspondence in nature, Lamarckian and Darwinist evolution concepts are combined to strengthen the deterministic part of the solution algorithm. This is called the Lamarckian‐marked‐constraint algorithm. The merits and behaviours of the suggested algorithms are discussed using two examples. The first is a hypothetical example affected by a multiminima problem. The second examines the equivalence problem using vertical electrical sounding data.     

多尺度逐次逼近遗传算法反演大地电磁资料

遗传算法是一种随机全局搜索算法,与常规的基于局部线性化的最优化方法相比对初始模型的依赖性大为减弱,但是存在着有效基因丢失和早熟收敛问题．采用多尺度逐次逼近反演思想而建立的多尺度逐次逼近遗传算法,能有效地解决上述问题．用该算法对大地电磁资料进行反演,理论曲线和实测资料的试算结果表明多尺度逐次逼近遗传算法能够自动反演地电参数．     

Seismic tomography in the Ostrava-Karviná mining region

The effective algorithm of seismic tomography was applied to the simultaneous determination of the velocity field in the medium and the location of hypocentres of events in the area of the underground seismological network in the Ostrava-Karviná mining region (OKR). Although the configuration of the network is unsuitable and the used data are relatively unreliable, after classifying the data it is possible to obtain representative information that demonstrates the existence of significant inhomogeneities of the medium in the area under study. In addition, it is possible to obtain more precise hypocentre parameters of local seismic events.     

Joint Inversion of Body-Wave Arrival Times and Surface-Wave Dispersion for Three-Dimensional Seismic Structure Around SAFOD

We incorporate body-wave arrival time and surface-wave dispersion data into a joint inversion for three-dimensional P-wave and S-wave velocity structure of the crust surrounding the site of the San Andreas Fault Observatory at Depth. The contributions of the two data types to the inversion are controlled by the relative weighting of the respective equations. We find that the trade-off between fitting the two data types, controlled by the weighting, defines a clear optimal solution. Varying the weighting away from the optimal point leads to sharp increases in misfit for one data type with only modest reduction in misfit for the other data type. All the acceptable solutions yield structures with similar primary features, but the smaller-scale features change substantially. When there is a lower relative weight on the surface-wave data, it appears that the solution over-fits the body-wave data, leading to a relatively rough V _s model, whereas for the optimal weighting, we obtain a relatively smooth model that is able to fit both the body-wave and surface-wave observations adequately.     

Full moment tensor inversion constrained by double-couple focal mechanism for induced seismicity*

In this study, we propose a new method to determine full moment tensor solution for induced seismicity. This method generalizes the full waveform matching algorithm we have developed to determine the double-couple (DC) focal mechanism based on the neighbourhood algorithm. One major difference between the new method and the former one is that we adopt a new misfit function to constrain the candidate moment tensor solutions with respect to a reference DC solution in addition to other misfit terms characterizing the waveform matching. Through synthetic tests using a real passive seismic survey geometry, the results show the new constraint can help better recover the DC components of inverted moment tensors. We further investigate how errors in the velocity model and source location affect the moment tensor solution. The synthetic test results indicate that the constrained inversion is robust in recovering both the DC and non-DC components. We also test the proposed method on several real induced events in an oil/gas field in Oman using the same observation system as synthetic tests. While it is found that the full moment tensor solutions without using the DC constraints have much larger non-DC components than solutions with the DC constraints, both solutions are able to fit the observed waveforms at similar levels. The synthetic and real test results suggest the proposed DC constrained inversion method can reliably retrieve full moment tensor solutions for the induced seismicity.     

3-D multi-parameter type traveltime tomography in a spherical coordinate frame: comparison of double and triple class simultaneous inversions

It is now common practice to perform simultaneous traveltime inversion for the velocity field and the reflector geometry in reflection/refraction tomography, or the velocity field and the hypocenter locations in regional earthquake tomography, but seldom are all three classes of model parameters updated simultaneously. This is mainly due to the trade-off between the different types of model parameters and the lack of different seismic phases to constrain the model parameters. Using a spherical-coordinate ray tracing algorithm for first and later(primary reflected) arrival tracing algorithm in combination with a popular linearized inversion solver, it is possible to simultaneously recover the three classes of model parameters in regional or global tomographic studies. In this paper we incorporate the multistage irregular shortest-path ray tracing algorithm(in a spherical coordinate system) with a subspace inversion solver to formulate a simultaneous inversion algorithm for triple model parameters updating using direct and later arrival time information.Comparison tests for two sets of data(noise free and added noise) indicate that the new triple-class parameter inversion algorithm is capable of obtaining nearly the same results as the double-class parameter inversion scheme. Furthermore,the proposed multi-parameter type inversion method is not sensitive to a modest level of picking error in the traveltime data, and also performs well with a relatively large uncertainty in earthquake hypocentral locations. This shows it to be a feasible and promising approach in regional or global tomographic applications.     

AN ALTERNATIVE STRATEGY FOR NON-LINEAR INVERSION OF SEISMIC WAVEFORMS1

A common example of a large-scale non-linear inverse problem is the inversion of seismic waveforms. Techniques used to solve this type of problem usually involve finding the minimum of some misfit function between observations and theoretical predictions. As the size of the problem increases, techniques requiring the inversion of large matrices become very cumbersome. Considerable storage and computational effort are required to perform the inversion and to avoid stability problems. Consequently methods which do not require any large-scale matrix inversion have proved to be very popular. Currently, descent type algorithms are in widespread use. Usually at each iteration a descent direction is derived from the gradient of the misfit function and an improvement is made to an existing model based on this, and perhaps previous descent directions. A common feature in nearly all geophysically relevant problems is the existence of separate parameter types in the inversion, i.e. unknowns of different dimension and character. However, this fundamental difference in parameter types is not reflected in the inversion algorithms used. Usually gradient methods either mix parameter types together and take little notice of the individual character or assume some knowledge of their relative importance within the inversion process. We propose a new strategy for the non-linear inversion of multi-offset reflection data. The paper is entirely theoretical and its aim is to show how a technique which has been applied in reflection tomography and to the inversion of arrival times for 3D structure, may be used in the waveform case. Specifically we show how to extend the algorithm presented by Tarantola to incorporate the subspace scheme. The proposed strategy involves no large-scale matrix inversion but pays particular attention to different parameter types in the inversion. We use the formulae of Tarantola to state the problem as one of optimization and derive the same descent vectors. The new technique splits the descent vector so that each part depends on a different parameter type, and proceeds to minimize the misfit function within the sub-space defined by these individual descent vectors. In this way, optimal use is made of the descent vector components, i.e. one finds the combination which produces the greatest reduction in the misfit function based on a local linearization of the problem within the subspace. This is not the case with other gradient methods. By solving a linearized problem in the chosen subspace, at each iteration one need only invert a small well-conditioned matrix (the projection of the full Hessian on to the subspace). The method is a hybrid between gradient and matrix inversion methods. The proposed algorithm requires the same gradient vectors to be determined as in the algorithm of Tarantola, although its primary aim is to make better use of those calculations in minimizing the objective function.     

Iteratively re‐weighted and refined least squares algorithm for robust inversion of geophysical data

A robust metric of data misfit such as the ?₁‐norm is required for geophysical parameter estimation when the data are contaminated by erratic noise. Recently, the iteratively re‐weighted and refined least‐squares algorithm was introduced for efficient solution of geophysical inverse problems in the presence of additive Gaussian noise in the data. We extend the algorithm in two practically important directions to make it applicable to data with non‐Gaussian noise and to make its regularisation parameter tuning more efficient and automatic. The regularisation parameter in iteratively reweighted and refined least‐squares algorithm varies with iteration, allowing the efficient solution of constrained problems. A technique is proposed based on the secant method for root finding to concentrate on finding a solution that satisfies the constraint, either fitting to a target misfit (if a bound on the noise is available) or having a target size (if a bound on the solution is available). This technique leads to an automatic update of the regularisation parameter at each and every iteration. We further propose a simple and efficient scheme that tunes the regularisation parameter without requiring target bounds. This is of great importance for the field data inversion where there is no information about the size of the noise and the solution. Numerical examples from non‐stationary seismic deconvolution and velocity‐stack inversion show that the proposed algorithm is efficient, stable, and robust and outperforms the conventional and state‐of‐the‐art methods.     

Seismic inversion through Tabu Search1

Highly non-linear seismic trace inversion problems can be solved efficiently by an implementation of Tabu Search, a meta-heuristic method related to artificial intelligence. The implementation under consideration is a deterministic, global search that combines the advantages ofa local search, giving a quick descent to local misfit minima, with an ability to cross misfit barriers in the model space. Once Tabu Search has found an area of low misfit, it performs an extensive exploration of its deepest points. This property makes it possible to use Tabu Search for a semiquantitative resolution and uncertainty analysis of the inverse problem.     

大地电磁数据的Occam反演改进

Occam 反演以其稳定的收敛性,在大地电磁（MT）数据解释中有广泛的应用。但是其每次迭代均用一维搜索方法求拉格朗日因子μ值,需要许多次正演计算,速度非常慢。在讨论了 Occam 反演中数据拟合差随μ变化的基础上,本文采用了μ值在一定步长下逐次递减的求取方法,每次迭代只需一次正演,极大地提高了计算速度。另外,反演求得光滑模型而非最光滑模型,分辨率更高。理论及实际数据的反演试算均表明,和 Occam 反演相比,反演依然稳定,但速度更快,结果也更真实。     

Exploring seismic inversion methodologies for non‐stationary geological environments: a benchmark study between deterministic and geostatistical seismic inversion

This paper presents a comparison between subsurface impedance models derived from different deterministic and geostatistical seismic inversion methodologies applied to a challenging synthetic dataset. Geostatistical seismic inversion methodologies nowadays are common place in both industry and academia, contrasting with traditional deterministic seismic inversion methodologies that are becoming less used as part of the geo‐modelling workflow. While the first set of techniques allows the simultaneous inference of the best‐fit inverse model along with the spatial uncertainty of the subsurface elastic property of interest, the second family of inverse methodology has proven results in correctly predicting the subsurface elastic properties of interest with comparatively less computational cost. We present herein the results of a benchmark study performed over a realistic three‐dimensional non‐stationary synthetic dataset in order to assess the performance and convergence of different deterministic and geostatistical seismic inverse methodologies. We also compare and discuss the impact of the inversion parameterisation over the exploration of the model parameter space. The results show that the chosen seismic inversion methodology should always be dependent on the type and quantity of the available data, both seismic and well‐log, and the complexity of the geological environment versus the assumptions behind each inversion technique. The assessment of the model parameter space shows that the initial guess of traditional deterministic seismic inversion methodologies is of high importance since it will determine the location of the best‐fit inverse solution.     

三维大地电磁自适应正则化有限内存拟牛顿反演

有别于传统基于梯度信息的反演方法在正则化约束中用总梯度逼近海塞逆矩阵的技术,本文将正则化约束问题的数据拟合项和模型光滑项分开考虑,只利用数据拟合函数的梯度信息对数据拟合项的海塞矩阵进行逼近,通过求解类高斯牛顿下降方向方程得到不依赖前几次迭代正则化因子的更精确下降方向,在求解当前迭代下降方向的过程中,通过保证右端项中两个向量的二范数在同一数量级的原则,实现了正则化因子的自动更新.对理论模型的试算表明这种自适应正则化反演方案可以在拟牛顿反演框架下基本达到OCCAM的算法稳定性,反演结果对初始模型依赖性较小,同时又无需在一次迭代中多次搜索最佳正则化因子.本文还基于此算法讨论了大地电磁各参数对于反演结果的影响,由于本文的反演结果能得到充分的正则化约束,因而在此框架下讨论阻抗和倾子在反演中的作用相对更为客观.     

基于单台加速度记录的混合全局优化HVSR反演场地浅层速度结构

利用水平与竖向谱比（HVSR）方法反演场地速度结构是国际上迅速发展的研究领域.HVSR反演计算实质是一个土层场地模型空间搜索的全局优化问题,当模型搜索空间的复杂程度增大时,目前常用的搜索算法收敛速度慢,计算效率较低.本文实现了一种结合遗传和模拟退火方法优点的混合全局优化HVSR反演算法,通过理论模型和竖向台阵实测数据的检验,表明该算法能获得很好的反演效果,较好地解决了蒙特卡罗方法收敛速度慢,遗传算法收敛早熟和模拟退火算法搜索效率低的问题.本文在此基础上讨论了单台加速度S波记录用于场地速度结构HVSR反演的适用性,为基于单个地震台的地震观测记录反演浅层速度结构提供了一种高效且较为准确的反演方法.     

烈度分布的几何特征和震源参数的估计

本文在考虑环境刚度效应和非匀阻效应的情况下,采用椭圆形震源简化模型,研究了走滑断层的地震烈度几何要素与震源参数之间的关系,初步建立了估计断层有效发震段长度、错动角、应力降和错动距离等参数的等震线法,并结合唐山、通海、炉霍和永善等震例进行了试算