SDM - A geodetic inversion code incorporating with layered crust structure and curved fault geometry
SDM - A geodetic inversion code incorporating with layered crust structure and curved fault geometry
复制标题
DOI：
--
复制
发表时间：
2013-04
期刊：
影响因子：
--
通讯作者：
Rongjiang Wang;F. Diao;A. Hoechner
中科院分区：
文献类型：
--
作者：
Rongjiang Wang;F. Diao;A. Hoechner
目前，地震断层破裂大地测量数据的反演大多基于统一的半空间地球模型，因为它是封闭形式的绿色函数。然而，地壳的层状结构可以显着影响反演结果。另一个影响，这是经常被忽视的，是有关弯曲的断层几何形状。特别是，大多数大型逆冲地震的断层面的倾角随深度变化，从几度到几十度不等。此外，许多大地震的走向是可变的。为了简单起见，这种弯曲的断层几何形状通常被近似为几个相连的矩形段，从而导致滑动分辨率和数据拟合的人为损失。在这篇演讲中，我们介绍了一个免费的FORTRAN代码，结合分层地壳结构和弯曲断层几何形状在一个用户友好的方式。SDM代表最速下降法，一种用于约束最小二乘优化的迭代算法。新代码可用于不同数据集的联合反演，其中可能包括系统偏移，因为大多数大地测量数据是从相对测量中获得的。这些偏移量被视为未知量，与滑移未知量同时确定。此外，先验和物理约束被认为是。先验约束包括由用户定义的滑移幅度的上限和滑移方向（前角）的变化范围。该模型需要物理约束以获得光滑滑动模型，通过使与数据的失配最小化的光滑项来实现。与大多数以前的反演代码不同，平滑可以可选地应用于滑移或应力降。该代码与一个输入文件一起工作，源代码中提供了一个文档齐全的示例。给出了应用实例。
Currently, inversion of geodetic data for earthquake fault ruptures is most based on a uniform half-space earth model because of its closed-form Green’s functions. However, the layered structure of the crust can significantly affect the inversion results. The other effect, which is often neglected, is related to the curved fault geometry. Especially, fault planes of most mega thrust earthquakes vary their dip angle with depth from a few to several tens of degrees. Also the strike directions of many large earthquakes are variable. For simplicity, such curved fault geometry is usually approximated to several connected rectangular segments, leading to an artificial loss of the slip resolution and data fit. In this presentation, we introduce a free FORTRAN code incorporating with the layered crust structure and curved fault geometry in a user-friendly way. The name SDM stands for Steepest Descent Method, an iterative algorithm used for the constrained least-squares optimization. The new code can be used for joint inversion of different datasets, which may include systematic offsets, as most geodetic data are obtained from relative measurements. These offsets are treated as unknowns to be determined simultaneously with the slip unknowns. In addition, a-priori and physical constraints are considered. The a-priori constraint includes the upper limit of the slip amplitude and the variation range of the slip direction (rake angle) defined by the user. The physical constraint is needed to obtain a smooth slip model, which is realized through a smoothing term to be minimized with the misfit to data. In difference to most previous inversion codes, the smoothing can be optionally applied to slip or stress-drop. The code works with an input file, a well-documented example of which is provided with the source code. Application examples are demonstrated.