利用人工爆破资料研究福建一维P波速度结构和地震定位

通过人工爆破资料研究地球结构的独特优点是震源时间和位置精确知道.2010—2012年间福建省进行了一系列的爆破实验.本文利用手工拾取来自省地震台网记录的爆破地震初至Pg、Pn以及续至Pg波到时数据,采用联合反演方法构建了新的一维P波速度模型,即福建爆破模型(FJEM).与华南模型相比,FJEM模型对走时的拟合程度提高了45%,有明显改善.利用不同爆破地震数据组合得到稳定类似的福建地区一维速度模型,显示福建地区存在较简单的一维速度结构.对爆破地震的重定位显示传统使用的华南模型在福建地区具有较小的水平定位误差(平均0.52±0.45km),但存在较大深度误差(平均4.7±1.2km).FJEM模型表现出与华南模型相似的水平定位能力,但是震源深度误差更小(1.3±1.1km).对基于FJEM模型的合成天然地震目录的重定位,华南模型显示出相似的定位结果:(1)台站方位覆盖较好的福建中部地区的水平定位误差小;(2)台站方位覆盖差的福建海岸及海峡区域水平定位误差大;(3)震源深度误差则跟台站数目及方位分布没有明显的关系,而是与发震时间误差有互易关系.从中可以看出,地震水平定位误差基本上受台站方位覆盖影响,而受参考速度模型影响不大;而在深度方面,本文改进的FJEM模型不仅更加接近真实的速度结构(拟合走时更好)而且也减小了深度误差.因此建议在福建及其邻近区域的日常定位中用FJEM模型替代华南模型.

A 1D P-wave velocity model of Fujian Province,China and earthquake locations from controlled explosions

Data from controlled explosions provide a unique advantage to study velocity models as the source locations are precisely known. A series of explosion were conducted in Fujian Province in 2010-2012. In this study, we hand-picked first Pg and Pn arrivals and secondary Pg arrivals from the explosion seismograms recorded at provincial stations. Using a joint inversion method, we inverted for a new 1-D P-wave velocity model, Fujian explosion model (FJEM). The FJEM model shows a 45% improvement in fitting the travel-time data over the current routine model. Similar velocity models inverted from subsets of explosion data suggest a relative stable and simple velocity structure in Fujian province. Relocations of the explosions suggest that the traditional South China model has small horizontal location errors (average 0.52±0.45 km), but errors in depth are significant (average 4.7±1.2 km), while the FJEM model has similar horizontal location capability but much smaller errors in depth (average 1.3±1.1 km). Synthetic tests using real earthquake bulletins show similar results for the South China model:(1) for earthquakes in the interior of Fujian with good azimuthal coverage, the horizontal mislocation is small; (2) the horizontal mislocations for earthquakes along the coast and in Taiwan Strait with poor azimuthal coverage can be very large; and (3) mislocations in depth can be large regardless station distribution, which have strong trade-off with origin time errors. Essentially, the horizontal location is largely affected by station azimuthal coverage and much less by the reference model. But the improved FJEM model not only is closer to the real Earth (fit data better) but also can reduce depth mislocation. We propose that South China model be replaced and the FJEM model be used in routine earthquake locations in Fujian and perhaps neighboring regions as well.