Preliminary Evaluation of Crustal Medium Parameters in Western China

. The establishment of attenuation relation of ground motion is always an important research content in engineering earthquake field. In seismology-based ground motion attenuation relationship and ground motion simulation using the stochastic method, regional parameters are necessary. These parameters involve stress drop Δ σ , two segmentation points R 1 and R 2 of the geometric spreading item, Q 0 and η of the inelastic attenuation item and the near-surface high-frequency decay parameter κ 0 , which reflect the regional crustal medium and site characteristics and have little relation with the earthquake size. In this paper, we take Sichuan, Yunnan and Xinjiang regions in western China as examples and we inverse the parameters for each region by micro-genetic algorithm. The inversion ranges of crustal media parameters in the three regions were determined respectively. Finally, we get the inversion results of every parameters and find that the parameters of the three regions have a great difference.


Introduction
Generally, the characteristics of ground motion are expressed by amplitude, duration and spectrum characteristics. The attenuation relation of ground motion expresses the influence of three main factors on ground motion: source, propagation path and site condition, which is an important research direction of engineering seismology. In seismic hazard analysis, it is necessary to estimate the expected ground motion of engineering site with the help of such relations.
In essence, attenuation relationship is empirical and based on the observation data of strong ground motions. Although China is a country prone to earthquakes and has the earliest recorded earthquake history in the world, in the early 1980s, with the development of digital strong earthquake observation technology, digital strong seismometer was introduced.
With the development of strong earthquake network [1] and the earthquake activity in recent years, the data amount of strong earthquake database in China is increasing gradually. In the regions with abundant strong earthquake records, some researchers have tried to directly establish the regional attenuation relationship by statistical regression of regional earthquake data, but some results are limited by insufficient data and have low reliability. We provided the seismology-based ground motion prediction models [29]- [30] , in which the regional parameters were inversed by small earthquake data of the broadband seismological network. In this paper, we attempt to use limited strong motion data in the inversion.
The distance used in this paper is hypocentral distance. According to the longitude and latitude information of stations (X 1 , Y 1 , Z 1 ) and the epicenter (X 2 , Y 2 , Z 2 ) provided in the records, it is converted into geodetic coordinates by equation (3), and then the epicentral distance is obtained by equation (4). According to the focal depth of each earthquake, the corresponding hypocentral distance can be easily obtained.
where r is the radius of the earth, and the value is 6370km; is the longitude (°); is 90° -latitude (°).

(4)
After the above processing, all the data were screened out as M W = 3.5 -5.5, signal-to-noise ratio greater than 3, and each station had at least 3 strong earthquake records. The data were preprocessed by baseline correction, filtering, S-wave extraction and other operations to form the strong motion database of small earthquakes. The data distribution in each study area is shown in figure 7-12. . MW -focal depth distribution in Yunnan area Figure 12. MW -hypocentral depth distribution in Yunnan area

Determination of inversion range
The efficiency of inversion depends on the search range of each parameter. If the range is too large, the search speed will be slow; if the range is too small, the optimal solution may not be covered and the inversion will fail. In this paper, the search scope of six regional parameters is determined based on the numerical expansion of other seismological results. Published values of stress drop, quality factor and κ 0 for three regions from various sources [2]- [24] are collected to determine the ranges of inversion parameters for each region, as shown in Table 1.
The two geometric attenuation parameters are related to regional crustal structure and are seldom studied in China. For the western United States, a three-section expression, with segment point R 1 with 70km±5km, and R 2 with 130km±10km, is used [25]- [28] , which can be taken for reference.

Inversion Results
In this paper, three administrative regions were first used for inversion of the parameters sorted out above. The inversion results are shown in Table2. By comparing the inversion results, it is not difficult to see that the parameters of the three regions are different. Two segmentation points R 1 and R 2 of the three study areas are similar. But stress drop and quality factor of the three study areas have a great difference.

Conclusion
This paper is only a preliminary study by the limitation of data, in which site condition is absence and station location need to be corrected further. We discuss the practicability of parameter inversion by strong ground motion records of small earthquakes. Based on the research in this paper, it is not difficult to infer that the  seismic energy attenuation rate in Xinjiang region is faster, and the overall Moho depth in three regions is similar. We will carry out an overall study on western China and compare it with the research results in this paper. Based on this paper, the parameters of crustal medium obtained by us play an important role in establishing the attenuation relation of ground motion in western China. By considering the site condition of stations and corrected station location, the inversion of regional parameters need to be investigated in the future.