地震动模拟之GMPE & PBM

关键是理解“经验公式”和“物理模型”两大类方法在地震动模拟中的优缺点，以及随机过程理论在地震动模拟中的位置。

Specia Issue edited by Professor Jacobo Bielak in EESD, Regional seismic ground-motion simulation and observation with engineering applications

The scarcity of strong ground-motion records due to large-magnitude earthquakes—especially in near-fault regions—presents a significant challenge. This challenge has traditionally been addressed by the development of empirical ground-motion prediction equations (GMPEs) within a probabilistic seismic hazard analysis (PSHA) framework,1,2,3 which are based primarily on recorded motions. More recently, the advent of high-performance computing has motivated the development of alternative three-dimensional (3D) numerical simulation methods derived from physics-based models (PBMs).

GMPE（经验公式）

• Advantages - adapted to different tectonic environments and site conditions; （简答，适用性强）
• Disadvantages - the number of existing records is insufficient for constraining important conditions, such as near-source and complex geological environments. Currently, they provide only peak values of various metrics of motion and must be recalibrated when new data become available. In most cases, GMPEs are derived without due explicit attention to the spatial correlation of ground-motion intensities between multiple sites and different spectral periods（实测地震动如果不足，公式误差较大，且每次得到新的数据，公式需要更新；不考虑空间相关性）

PBM（物理模型）

• Advantages - produce spatially distributed time histories of simulated ground motions within entire regions under source-specific conditions; naturally for spatial correlation of simulated ground motions, and they serve as a numerical laboratory for gaining insight into the physics of earthquakes and seismic wave-propagation phenomena. （自然空间相关，考虑震源信息，且可以分析地震的本质）
• Disadvantages - the corresponding computational effort is costly, and extensive information is required to construct realistic seismic-source models and a reliable 3D velocity model of the desired region. Another drawback is that the accuracy of the results of deterministic simulations decreases at higher frequencies due to modeling and computational limitations; thus, stochastic methods are required for simulations at higher frequencies.（计算成本高，所需震源信息繁多，高频精度低）