Global observation of off-great-circle propagation of long-period surface waves
Global observation of off-great-circle propagation of long-period surface waves
复制标题
DOI：
10.1111/j.1365-246x.1995.tb06673.x
复制
发表时间：
1995-10
影响因子：
2.8
通讯作者：
G. Laske
中科院分区：
文献类型：
--
作者：
G. Laske
在目前这一代的全球表面波色散图中，长波结构似乎已经很好地确定了。在全球相速度异常的模式，谐波度1-6的一般协议。然而，较短的波长结构显着不同的出版地图，它似乎这部分的模型强烈依赖于反演技术和数据集的表面波色散（通常相位测量）。偏振数据取决于相速度的横向梯度，因此对较短波长的结构比相位数据更敏感;因此，包括这些数据应提高分辨率。在本文中，我证明了极化数据的长周期表面波（2 - 80秒），作为频率的函数，可以可靠地测量使用多锥度技术。由此产生的偏离大圆到达角的表面波包是相对容易解释的射线理论框架内。我们的三组分记录的数据库现在足够大，提供有用的全球色散图的约束，特别是在较短的波长部分。除了相速度模型本身之外，在每个站处的水平分量的可能的定向误差被包括在非线性反演中作为附加的独立模型参数。这在与数据的拟合方面给出了显著的改进。在调查的37个台站中，至少有4个台站的定向误差可能超过3”。
SUMMARY In the current generation of global dispersion maps of surface waves, the longwavelength structure seems to be very well determined. There is general agreement in the patterns of global phase velocity anomalies up to harmonic degree 1-6. However, the shorter-wavelength structure varies significantly between published maps, and it appears that this part of the models depends strongly on the inversion technique and on the data set of surface-wave dispersion (usually phase measurements). Polarization data depend on the lateral gradient of phase velocity and hence are more sensitive to shorter-wavelength structure than phase data; thus, including these data should enhance resolution. In this paper, I demonstrate that polarization data of long-period surface waves (2 80 s), as a function of frequency, can be reliably measured using a multitaper technique. The resulting off-great-circle arrival angles of the surface-wave packets are relatively easy to interpret within a ray-theoretical framework. Our data base of threecomponent recordings is now large enough to provide useful constraints on global dispersion maps, particularly on the shorter-wavelength parts. Apart from the phase velocity model itself, a possible misorientation of the horizontal components at each station is included in a non-linear inversion as an additional independent model parameter. This gives a significant improvement in the fit to the data. Misorientations of more than 3" are probable for at least four of the 37 stations investigated.