基于机器学习的地震动特征提取与模拟以2021年云南漾濞6.4级地震为例

随着地震动数据数量的增长和质量的提高,将基于数据驱动的机器学习方法应用到地震动模拟中有重要意义。以2021年5月21日云南漾濞M_S6.4地震为例,利用主成分析方法从前震及余震地震动记录中提取特征母波时程,将地震动三要素作为模拟误差约束,在求解母波的线性组合系数时使用多目标优化算法寻优,最终找到帕累托最优解作为模拟目标台站记录时的组合系数,得到模拟地震动时程。结果表明:主成分析法在对实际地震动记录进行特征提取后,得到的特征母波时程可以在一定程度上保留原始数据的主要信息;考虑幅值、频谱和持时这三要素的角度去控制模拟误差,可以使得模拟的地震动时程更加接近真实记录。提出的基于特征提取的地震动模拟方法可以为基于小震数据合成大震地震动提供参考。     

南昆仑造山带西段构造样式及构造序列

南昆仑造山带西段发育逆冲推覆构造、平行走滑构造、地垒一地堑构造和阿尔卑斯式褶皱、侏罗山式褶皱,前两类构造尚有韧性剪切带伴生;其构造组合在平面上具有内带一外带一前陆带明显的分带性,且形成了逆冲推覆构造一平行走滑构造一地垒一地堑构造的构造序列。     

Lithologic Hydrocarbon Deposits in Rift Lake Basins in Eastern China

The rift lake basins in the eastern China have abundant hydrocarbon resources of lithologic deposits, which resulted from excellent source rocks and multi-type sandbodies developed during strong rifting. Vertically, the lithologic deposits are mainly distributed in the lowstand, lacustrine invasion and early highstand systems of third-order sequence corresponding to a secondary tectonic episode of strong rifting, and laterally they are closely related to various fans and turbidite sandbodies controlled by syn-sedimentary faults. A variety of lithologic traps have been developed in the rift lake basins, and they generally have favorable conditions of source-reservoir-seal assemblage and hydrocarbon accumulation dynamics, indicating that there is a great exploration potential of lithologic deposits in the rift lake basins. In order to obtain satisfactory effects of lithologic deposit exploration, it is required to combine new theories with advanced technical methods.     

Predicting Terrestrial Flagstone Reservoirs in the Sha-I Member of the Qibei Depression in the Dagang Oilfield

There are few 3-D seismic profiles and drillings in the middle part of the Qibei depression in the Dagang oilfield, and more than 70% of the 2-D seismic profiles were completed before the 1980s. Meanwhile, changes in the terrestrial formations in this region have been large and complex. These factors have made it difficult to predict reservoirs in this area. The purpose of this paper is to establish a methodology for predicting potential gas and oil reservoirs. Our research combines sequence stratigraphy, well-logs, and seismic analysis to elucidate the prediction of flagstone reservoirs in the S1 (Sha-I) Member in the middle of the Qibei depression. Previous research indicates that these rocks were deposited in an environment that had a semiarid, northern subtropical, and warm, humid climate. The objective strata currently consist mainly of lake fades, deeper lake facies, and shore-shallow lake facies. The study reveals that the lower section of the S1 Member is an important objective region for explora     

利用激发极化驰豫时间谱确定渗透率的实验研究

激发极化驰豫时间是孔隙大小的函数,采用奇异值分解方法对激发极化衰减曲线进行多指数反演,得到光滑的驰豫时间谱,该谱呈现多峰特性.实验结果表明,溶质对该谱的影响主要是由于溶质离子的扩散系数不同所致.对于同一块岩心,衰减谱随扩散系数的增大而向小驰豫时间方向移动.结合岩心地层因素F,研究了激发极化驰豫时间几何平均值Tg与岩石的渗透率K的关系,结果表明,logK和log(T1/2g/F)、log(Tg/F)呈现良好的线性关系.     

二维圆柱坐标下FDTD法对多频电磁波测井的数值模拟

在介质围绕Z轴旋转对称的情况下,三维直角坐标下的麦克斯韦方程可变成两组分别对应TM波和TE波的二维柱坐标下的偏微分方程.采用线圈作为电磁波测井的发射器和接收器的时候,可采用TE波对应的方程,这时相当于磁极子的情况.而采用电偶极子作为发射和接收器的时候,则对应TM波的方程.采用时间和空间均为二阶精确的有限差分方法,将偏微分方程进行差分化.这样,空间的电磁场可由时间域有限差分法(FDTD)来求解.     

利用区域辐射特性优化获取光学遥感数据

通过分析成像过程及其影响因素,提出了对成像积分时间进行控制的优化成像方法。首先建立成像参数模型,探讨了模型参数计算、成像过程中模型参数的选择,分析了图像后处理问题,并利用北京一号小卫星进行了遥感数据优化获取实验。实验结果表明,根据地物特性进行成像参数优化后获取的图像具有更宽的动态范围和更高的辐射分辨率,信息量和信噪比也大幅增强。     

绕不开的祖父

当一个人进行时间旅行的时候,他绕不开他的祖父。因为这里面存在一个悖论：假如他回到过去,把他年轻的祖父杀了,那么他爸爸就不会出现,那他又是从哪来的呢？这是著名的“祖父悖论”,是所有研究时间旅行的学者不得不面对的一个难题。     

关于宇宙的几个误解

不准确!严格说,一天是24小时,而。地球自转一圈则是23小时56分4秒（忽略地球自转长期变慢等因素）,它们并不相等。“地球自转一圈”的时间即为地球的自转周期,等于遥远（因而在天空中的位置保持不变）的恒星两次经过上中天时的时间间隔（在同一地点观测）。而我们平常所用的公历中的一天（即24小时）是基于太阳的视运动而定义的,