Nonlinear inversion of electrical resistivity imaging using pruning Bayesian neural networks

Conventional artificial neural networks used to solve electrical resistivity imaging (ERI) inversion problem suffer from overfitting and local minima. To solve these problems, we propose to use a pruning Bayesian neural network (PBNN) nonlinear inversion method and a sample design method based on the K-medoids clustering algorithm. In the sample design method, the training samples of the neural network are designed according to the prior information provided by the K-medoids clustering results; thus, the training process of the neural network is well guided. The proposed PBNN, based on Bayesian regularization, is used to select the hidden layer structure by assessing the effect of each hidden neuron to the inversion results. Then, the hyperparameter α _k , which is based on the generalized mean, is chosen to guide the pruning process according to the prior distribution of the training samples under the small-sample condition. The proposed algorithm is more efficient than other common adaptive regularization methods in geophysics. The inversion of synthetic data and field data suggests that the proposed method suppresses the noise in the neural network training stage and enhances the generalization. The inversion results with the proposed method are better than those of the BPNN, RBFNN, and RRBFNN inversion methods as well as the conventional least squares inversion.     

印尼大地震震前次声异常信号的分析

2004年12月26日发生在印尼苏门答腊岛附近印度洋海斌的地震震级高达9．0（Mw9．0）,引起了巨大海啸．造成了几十万人的伤亡,至今全世界的人们还记忆犹新。美国地质调查局把此次地震列为百年以来全球第五大地震。时隔3个月,2005年3月29日在印尼又发生了Ms8．7地震,5月17日再次发生了7．1级地震。北京工业大学地震研究所在这几次印尼地震前都收到次声波异常信号,本文对这些异常信号进行了初步的分析与探讨。     

辽宁西部地区地震前后重力异常反映

采用高精度拉科斯特重力仪在辽宁西部地区进行多期重力重复测量，获得了较为丰富的基础资料，通过对资料的分析与研究，进一步加深了辽宁西部地区重力场变化的认识。特别是１９９４年１０月和１９９６年７月，在测区内及其附近相继发生ＭＬ４．９级和ＭＬ５．０级地震。而这两次地震为１９８８年以来在辽宁西部地区发生的最大地震。通过对该区流动重力资料空间上和时间上的研究分析，从中获得了震前重力变化的某些前兆信息。通过对该     

Investigation into the daily precipitation variability in the Yangtze River Delta,China

In this article, by using the daily precipitation data measured at 58 meteorological stations, spatial and temporal variability of daily precipitation including zero rainfall values (called “precipitation”) and without zero rainfall values (called “rain”) and four precipitation extrema (P₀, P₂₀, P₅₀, and P₁₀₀ representing the daily precipitation with the magnitude smaller than 0.1 mm, bigger than 20 mm, 50 mm, and 100 mm per day, respectively) in the Yangtze River Delta (YRD) during 1958–2007 were analyzed, and the effects of urbanization were further investigated. Results indicate that (i) differing from the downward trends in 1958–1985, daily precipitation and rain in 1986–2007 show slowly downward trends in the mid YRD but show upward trends in the northern and southern YRD. (ii) Spatial and temporal variability of the rain is more complex than daily precipitation. Both of them become smaller but show more obvious fluctuations in 1986–2007. (iii) Urbanizations cause not only the urban rainfall island problem but also more obvious fluctuations of rain intensity in the mid YRD, reflecting more uncertainty of daily precipitation variability. (iv) Urbanizations have little effects on the variability of P₀ and P₁₀₀ but cause notable increases of P₂₀ and P₅₀. (v) The spatial variability of daily precipitation and precipitation extrema in 1958–1985 clearly shows a breakpoint at 30°20′N latitude, but the breakpoint disappears afterward because of the effects of urbanization. Copyright © 2012 John Wiley & Sons, Ltd.     

CHARACTERISTICS OF TURBULENCE SPECTRA IN DIFFERENT WEATHER PROCESSES UNDER UNSTABLE CONDITIONS*

In this paper,the data chosen from measurements by supersonic anemometers in Tongxian County of Beijing in 1990 are used to study characteristics of turbulence spectra in precipitation weather.Some turbulence parameters such as turbulence intensity,heat and momentum fluxes,friction velocity,M-O length and stability parameter are calculated and their dependence on turbulence spectra in different stability conditions are analyzed.Some encouraging results are obtained and compared with others' results.     

两次强降水过程雨滴谱特征及差异

利用寿县观测站内的Parsivel激光雨滴谱仪结合观测站雨量数据及雷达基数据,分析了发生在2015年6月26—30日梅雨期间和2015年8月7—10日超强台风"苏迪罗"影响期间2次强降水过程的雨滴谱结构特征及其差异,拟合了雨强与雷达反射率因子之间的关系。结果表明:雨强的大小直接影响雨滴谱的特征参量,且随着雨强增大而增大;梅雨锋暴雨中1.0mm直径≤1.5mm的粒子所占比例最多,雨强贡献率最大;台风雨中0.75mm直径≤1.0mm的粒子所占比例最多,但1.0mm直径≤1.25mm的粒子对雨强的贡献最大,说明较大粒子对强降水的贡献较大。     

含有硬包裹体分布的非均质岩石水力压裂特性研究

基于完全流-固耦合的弹塑性理论给出了水力压裂数值计算的弥散裂缝模型,其中材料的弹性部分采用线弹性本构关系,塑性部分采用摩尔-库仑破坏准则及强化准则。依据当前的有效应力状态修正渗透系数来模拟压裂液在裂缝中的流动。渗透系数的修改使用双曲正切函数,并采用平均有效应力作为水力裂缝的起裂判据。在ABAQUS软件中通过用户自定义程序添加了该模型。根据岩石的切面照片建立了含有硬包裹体分布的非均质岩石的有限元计算模型,模拟了中心点注水条件下的水力压裂传播过程,讨论了在常应力状态下非均质岩石中开裂区域、典型位置的应力路径变化和裂缝传播范围随时间变化的特点。进行了多种条件下含有硬包裹体分布的岩石材料的数值试验,得出了基岩材料的弹性模量、凝聚力和渗透系数以及注水速率对峰值注水压力、平均注水压力和裂缝开度的影响规律。     

四川牦牛坪稀土矿床萤石REE配分模式的影响因素

四川牦牛坪稀土矿床萤石REE配分模式表现出明显变化的特征，本文从稀土离子半径和温度两个方面分析了影响矿区萤石配分模式的原因。结果表明：矿区萤石中LREE的配分模式主要受温度影响，而MREE、HREE主要受离子半径控制。     

殷墟文化发生的环境背景及人类活动的影响

不同农业单元122个表土样品花粉分析结果表明,安阳地区表土花粉组合中草本植物花粉含量最高,以禾本科、蒿属、藜科和十字花科为主,乔木植物花粉百分含量较低,以松属、桦属、胡桃属、黄栌属、栎属为主,应为西部太行山区传播而来。汪家店和茶店陂两个全新世地层自然剖面孢粉分析表明,约 10000cal.aB.P.以前,安阳地区分布着以松属、栎属、桦属等为主的落叶阔叶林;  约 10000～8200cal.aB.P.,气候变暖,降水量增加,森林植被中夹杂着常绿栎等亚热带树种;  约 8200～3400cal.aB.P.,安阳地区进入全新世大暖期,森林植被演变为亚热带落叶阔叶和常绿阔叶混交林;  商王朝在安阳地区建都后（3400cal.aB.P.后）,乔木植物花粉急剧减少,草本植物花粉和中华卷柏百分含量升高,表明森林被大量砍伐,砍伐后的林地开垦为农田,主要种植禾本科等农作物;  两条剖面上部的冲积层中出现大量中华卷柏孢子,表明森林植被破坏后水土流失加剧。古地貌面孢粉分析也证明殷商时期安阳地区生长有亚热带落叶阔叶和常绿阔叶混交林,气候和植被条件适于野象生存;  殷墟附近为禾本科农作物花粉高含量区,表明殷商时期农业已具一定规模。     

青藏高原东部基于噪声的面波群速度分布特征

通过收集青海、甘肃、四川三省的76个地震台记录的2008年1—12月三分量的连续噪声数据,利用噪声面波层析成像的方法获得了青藏高原东部的面波群速度分布特征。首先采用多重滤波方法提取了1 000多条台站对5～50 s的三分量面波群速度频散曲线,然后将研究区域划分为0.2°×0.2°的网格,利用O ccam方法反演了瑞利波(R-R)和勒夫波(T-T)的群速度分布。反演得到的群速度分布特征与地表地质和构造特征表现出较好的相关性,清晰地揭示了地壳内部的横向速度变化。层析成像的结果显示在短周期(8～20 s)内,拥有较厚的沉积层的四川盆地表现为明显的低速特征,而青藏高原东部则表现为较高的群速度分布特征;随着周期的增加(>20 s),群速度的分布特征呈现出与短周期相反的特性,青藏高原东部下方的速度远远低于四川盆地,这可能与青藏高原东部中、下地壳低速层相关联,同时也意味着研究区域的地壳结构具有明显的横向不均匀性。在群速度分布图上,龙门山不仅是四川盆地与青藏高原的地形和构造分界带,同时也对应着高群速度与低群速度的过渡带。