Journal of Earth Science  2018, Vol. 29 Issue (6): 1409-1418   PDF    
Distribution of Intra-Crustal Low Velocity Zones beneath Yunnan from Seismic Ambient Noise Tomography
Weibing Qin1,2, Shuangxi Zhang1,3,4, Mengkui Li1, Tengfei Wu1, Chaoyu Zhang1,3    
1. Department of Geophysics, School of Geodesy and Geomatics, Wuhan University, Wuhan 430079, China;
2. China Three Gorges Corporation, Yichang 443000, China;
3. Key Laboratory of Geospace Environment and Geodesy of Ministry of Education, Wuhan University, Wuhan 430079, China;
4. Collaborative Innovation Center of Geospace Information Science, Wuhan University, Wuhan 430079, China
ABSTRACT: Previous studies have reached consensus that low velocity zones are widespread in the crust beneath Yunnan region. However, the relationships between the low velocity zones and large faults, earthquake distribution are less investigated by available studies. By analyzing the seismic ambient noise recorded by Yunnan Seismic Networks and Tengchong volcano array, we construct a 3D crustal shear wave velocity model for the Yunnan region, which provides more details of the distribution of intra-crustal low velocity zones all over Yunnan. The distribution of low velocity zones shows different features at different depths. At shallow depths, the results are well correlated with near surface geological features. With increasing depth, the low velocity zones are gradually concentrated on the northern part of our study area, most likely reflecting variations in crustal thickness beneath the Yunnan region. The low velocity zones are truncated at depth by several large faults in Yunnan. It is interesting that most strong earthquakes (Ms ≥ 5.0) occurred in Yunnan are distributed in low velocity zones or the transition zones between low and high velocity anomalies within the upper-to-middle crust. The crustal structure is composed of a brittle, seismically active upper-to-middle crust and a warm, aseismic lower crust.
KEY WORDS: ambient noise tomography    crustal structure    low velocity zone    fault    strong earthquake    


Yunnan is located in the southeastern margin of the Tibetan Plateau (TP), suffering from the squeezing and subduction of the Indian Plate against the Tibetan Plateau Plate and Yangtze Platform (Fig. 1) (Zhang and Wang, 2009). This region has complex tectonic background with criss-cross faults, high level of seismicity and volcanic activities (Li et al., 2008; Hu et al., 2005a; Wu and Ming, 2001). It is one of the most active and complicated regions of tectonic movement in China mainland (Zhang and Wang, 2009; Huang et al., 2002), attracting a lot of investigations in recent years. Previous studies suggested that the resistance from the Sichuan Basin makes Yunnan region an important passage way for crustal materials flowing from internal part of the TP (Bao et al., 2015; Bai et al., 2010; Yao et al., 2010; Royden et al., 2008, 1997). However, the spatial distribution of low velocity zones (LVZs) beneath Yunnan is still not completely understood.

Figure 1. Topographic relief and tectonic map of Yunnan. The red triangles denote Yunnan network stations, the blue triangles denote Tengchong volcano array stations, the black lines denote the major faults, the enclosed area by white lines is the Yunnan Province. NJF. Nujiang fault; LCJF. Lancangjiang fault; RRF. Red River fault; ZDF. Zhongdian fault; LNF. Lijiang-Ninglang fault; CHF. Chenghai fault; YMF. Yimen fault; PDHF. Puduhe fault; XJF. Xiaojiang fault; MSF. Maitreya-Shizong fault; CTF. Chuxiong-Tonghai fault; WLSF. Wuliangshan fault; NTRF. Nanting Rive fault; LLF. Longling fault; TCF. Tengchong fault; BS. Baoshan; CX. Chuxiong; JG. Jinggu; KM. Kunming; SM. Simao.

The main faults in our study area are illustrated in Fig. 1. Some studies suggest that the LVZs may be truncated by faults at depth (Yao et al., 2010, 2008; Wang et al., 2003; Huang et al., 2002) and the existence of LVZs may facilitate the motion of the faults (Bao et al., 2015). Unfortunately, the relationships between LVZs and faults remain unclear due to the resolution limitation of previous lithospheric structure models. On the other hand, the seismicity in Yunnan is of high level. According to the earthquake catalog of China Seismological Network, there are eight earthquakes with magnitude larger than 7.0 occurred in this region since the 1970s. Generally, the LVZs may represent weak sections of the seismogenic crust (Huang et al., 2002; Zhao et al., 2000). It is worthwhile to thoroughly understand the crustal structure particularly the spatial distribution of LVZs in Yunnan, which can improve the understanding of seismic activity in this area.

Geological and geophysical surveys have been conducted in recent years to investigate the crustal structure in Yunnan. Results from deep seismic sounding have shown that the average crustal shear wave velocity is low and the LVZs exist in the middle-to-lower crust in some parts of Yunnan (Zhang et al., 2005a, b; Wang and Huangfu, 2004; Kan et al., 1986). Zhang and Wang (2009) reinterpreted four seismic transects and found that the velocity anomalies, earthquakes and faults seem to be coupled. Regional travel time tomographic studies (Xu et al., 2013; Wang et al., 2003; Huang et al., 2002) have found that prominent low velocity anomalies are widespread in the middle-to-lower crust in Yunnan. These results also show that some large fault zones and most strong earthquakes are located in the transition zone between positive and negative anomalies. He et al. (2005) obtained a 3D S-wave velocity structure of the upper-to-middle crust in Yunnan from two-station phase velocity tomography. They found that the Sichuan-Yunnan Diamond Block (SYDB) bounded by Xiao- jiang fault (XJF) and Red River fault (RRF) is a large LVZ with the boundary basically coinciding with the faults (He et al., 2005). And most large earthquakes of M > 6.0 in Yunnan occurred in the transition zones between low and high velocities (He et al., 2005). Receiver function analysis (Xu et al., 2007; Hu et al., 2005a) have shown considerable variation in spatial distribution of the low-velocity layers beneath Yunnan seismic stations, though the geometries have not been well resolved. Based on the joint inversion of Rayleigh wave dispersion and receiver functions, Hu et al. (2005b) have found that RRF is an obvious boundary and the structures of the crust and mantle on either of its sides are notably different. Besides, there are two low velocity layers of the mantle associated with the distributions of major earthquakes (Hu et al., 2005b). Li et al. (2008) have concluded that markable LVZs exist in middle-to-lower crust beneath the Yunnan region, especially in western Yunnan. The observation of LVZs distribution also indicates RRF is a major boundary fault and may have cut through the entire crust reaching down to the uppermost mantle (Li et al., 2008). It is possible that the occurrence of earthquakes is associated with criss-cross faults in this region (Li et al., 2008). Due to the limited numbers of seismic stations, the results may not be very convincing. Bao et al. (2015) have used more data and discovered two low-velocity channels that bound major strike- slip faults in SE Tibet, parts of the channels (LVZs) extend to Yunnan. Moreover, most large earthquakes in this study area occurred in the boundaries of the LVZs (Bao et al., 2015).

In this study, we investigate the detailed distribution of intra-crustal LVZs beneath the whole Yunnan region through seismic ambient noise tomography, which has been widely used to illuminate the velocity structures of the Earth (Wang and Gao, 2014; Cheng et al., 2013; Fang et al., 2010; Yao et al., 2010, 2008; Li H Y et al., 2009; Yang et al., 2008). By using the seismic data observed in Yunnan Seismic Networks and Tengchong volcano array, we seek to obtain a 3D model of crustal shear wave structure. Our research targets are to: (1) determine the spatial distribution of the intra-crustal LVZs in detail; (2) investigate the geometric relationships between LVZs and faults; and (3) investigate the relationship between LVZs and earthquake distribution.

1 DATA PROCESSING 1.1 Data and Cross-Correlation

The raw seismic ambient noise data we used in this study were recorded by 55 broadband seismic stations operated by Yunnan Seismic Networks (red triangles in Fig. 1) and Tengchong volcano array (blue triangles in Fig. 1) from January 1, 2012 to December 31, 2013. We recovered Rayleigh wave signals from the vertical component of ambient noise.

The data processing procedure applied here is very similar to that introduced by Bensen et al. (2007). First, we cut the continuous data into one-day long segments, decimated them to one sample per second, removed the instrument response, the mean and the trend, and band-pass filtered the seismograms in the period range of 5–50 s. Next, we performed temporal normalization using running-absolute-mean method to suppress the influence of earthquake signals and other irregularities, and spectral whitening to produce a flat-amplitude spectrum. Finally, the processed one-day long traces were cross- correlated between all station-pairs and stacked over the two- year time window. Figure 2 displays the cross-correlation functions (CCFs) between KMI (Kunming Station) and other 54 stations. Prominent surface wave signals are observed on both positive and negative correlation lags with an average velocity of ~3 km/s.

Figure 2. Cross-correlation functions from seismic ambient noise recorded at KMI station paired with other 54 stations in the study area, prominent surface wave signals are observed on both positive and negative correlation lags with an average move-out velocity of ~3 km/s.

As seen in Fig. 2, the CCFs are usually time asymmetrical due to the inhomogeneous distribution of noise sources and attenuation (Paul et al., 2005; Sabra et al., 2005a). By assuming a homogeneous noise field, we averaged positive and negative cross-correlation lags of each station pair to create a symmetric CCFs, which can simplify data analysis and enhance the signal-to-noise ratio (SNR) of the surface waves (Yang et al., 2008). The CCFs has π/2 phase difference from that of empirical Green's functions (EGFs). The latter makes it possible to enhance the higher frequencies (Sabra et al., 2005a, b). Previous theoretical studies (Roux et al., 2005; Snieder, 2004; Lobkis and Weaver, 2001) have demonstrated that one can get EGFs from CCFs by taking the negative of the time derivatives except for a frequency-dependent amplitude correction. The following analysis is performed on the symmetric EGFs exclusively.

1.2 Phase Velocity Dispersion Measurement

Phase velocity dispersion curves of Rayleigh wave for all appropriate station pairs are obtained from the symmetric EGFs by far-field approximation and image transformation technique (Yao et al., 2006, 2005). We summarize briefly the procedure here. First, we adopted a narrow frequency band of 0.4 s to filter the symmetric EGFs at central periods ranging from 5 to 35 s with 1 s interval. Next, we constructed a time-period (t-T) image for the Rayleigh waves (see Fig. 6a in Yao et al., 2006). The frequency dependence is readily observed in the image. Then, this t-T image was transformed into a velocity-period (c-T) image (e.g., Fig. 3). There are multiple phase-velocity dispersion branches on the image (red regions) due to the 2π ambiguity, but only one is correct. Based on the values of each branch, we can easily identify and automatically pick the desired dispersion curve in the image as shown in Fig. 3.

Figure 3. Phase velocity dispersion measurement for an example station pair based on the image transformation technique (Yao et al., 2006, 2005). Red line represents the cutoff period at which the interstation distance reaches three wavelengths. Cyan circles mark the velocity-period coordinates where the SNRs are larger than 5. Red dots superposed on the black line are the phase velocities finally determined from the EGFs.

Data quality control is vital so as to identify and reject bad measurements. We selected raw data based on two criteria: (1) the interstation distance must be at least three wavelengths to satisfy the far-field approximation; (2) SNR should be larger than 5. Here SNR is period dependent, and defined as the ratio of the peak amplitude of the signal window and the mean envelope amplitude of the 150 s long noise window right after the signal window at around the considered period. Theoretically, 55 stations could produce up to 1 485 paths. However, the unreasonable dispersion curves that are obviously different from the majority should be removed by careful visual inspection (Cheng et al., 2013). The final number of paths used in the inversion at each period are shown in Fig. 4.

Figure 4. The path numbers used in the inversion at different periods.
2 INVERSION METHODS 2.1 Phase Velocity Tomography

The dispersion characteristics provide information about the interstation wave propagation and hence about seismic velocities on the path the wave pass through (Moschetti et al., 2007). We constructed the phase velocity maps from 5 to 35 s by using a 2D linear inversion procedure developed by Yanovskaya and Ditmar (1990). This method is a 2D generalization of the classical 1D method of Backus and Gilbert (1968). The tomographic method estimates the phase velocity Ce(θ, φ) at each period by the minimizing misfit function Φ show as follows

$ \mathit{\Phi }{\text{ = (}}\mathit{\boldsymbol{d}} - \mathit{\boldsymbol{Gm}}{{\text{)}}^T}(\mathit{\boldsymbol{d}} - \mathit{\boldsymbol{Gm}}) + \alpha \iint {{{\left| {\nabla \mathit{\boldsymbol{m}}(\mathit{\boldsymbol{r}})} \right|}^2}{\text{d}}\mathit{\boldsymbol{r}} = \min } $ (1)


$ \mathit{\boldsymbol{m}}(\mathit{\boldsymbol{r}}) = (C_e^{ - 1}(\mathit{\boldsymbol{r}}) - C_0^{ - 1}){C_0} $ (2)
$ {d_i} = {t_i} - {t_i}_0 $ (3)
$ {(\mathit{\boldsymbol{Gm}})_i} = \iint {{G_i}(\mathit{\boldsymbol{r}})\mathit{\boldsymbol{m}}(\mathit{\boldsymbol{r}}){\text{d}}\mathit{\boldsymbol{r}} = \int_{{l_{0i}}} {\frac{{\mathit{\boldsymbol{m}}(\mathit{\boldsymbol{r}})}}{{{C_0}}}{\text{d}}s} } $ (4)
$ \iint {{G_i}(\mathit{\boldsymbol{r}}){\text{d}}\mathit{\boldsymbol{r}} = \int_{{l_{0i}}} {\frac{{{\text{d}}s}}{{{C_0}}}} } = {t_{i0}} $ (5)

In Eqs. (1)–(5), r=r(θ, φ) is the position vector, C0 is the initial phase velocity for each period and often given by the average along all paths. ti is the observed travel time along the ith path. ti0 is the travel time calculated from the starting model. l0i is the length of the ith path, and s is the segment along which the forward calculation is performed. The regularization parameter α controls the trade-off between the data misfit and the smoothness of the resulting phase velocity maps. The larger α is, the smoother the inversion result, the lower the resolution, and vice versa. After several experiments, we finally determined that the parameter α=0.2 would yield relatively smooth maps with small solution errors.

2.2 Resolution Analysis

Tomographic results are not always unique. Sometimes, the initial data cannot constrain the seismic velocities uniquely for all points of the medium (Fang et al., 2010). It needs to estimate the resolution. Regardless of theoretical errors, the resolution of the tomographic results is mainly determined by coverage and azimuthal distribution of ray paths (Ritzwoller and Levshin, 1998). As suggested by Backus and Gilbert (1968), a reliable method to evaluate the resolution is to estimate the spatial averaging kernels for each point in the study area.

For 2D tomography problems, a function R(θ, φ) for different orientations of the coordinate system is used to evaluate the size of the averaging area along different directions (Yanovskaya et al., 1998). The spatial averaging kernel can be approximated by an ellipse centered at a grid point, with axes equal to the largest Rmax(θ, φ) and the smallest Rmin(θ, φ) values of Rmax(θ, φ). Once the largest and the smallest axes of the ellipse are calculated, the resolution in each point is given by a single number, which can be expressed as the mean size of the averaging area L=[Rmax(θ, φ)+ Rmin(θ, φ)]/2. Figures 5a, 5b, 5c show the path coverage for three selected periods (5, 15 and 35 s). The corresponding spatial resolution maps are shown in Figs. 5d, 5e, 5f. The resolution radius was estimated to be ~50 km for most of the study area. Therefore, we finally parameterized the study area into grids with cell size of 0.5º×0.5º. In general, the resolving power is generally good in the center of the study area but it deteriorates toward the edge especially at long period. So the velocity anomalies in the edge of the study area may not be accurate enough because of resolution limitations.

Figure 5. Path coverage of phase velocity measurements at three selected periods (a), (b), (c). The red triangles denote stations. Map views of average spatial resolution of estimated Rayleigh wave phase velocity (d), (e), (f). The distribution maps of estimated Rayleigh wave phase velocity at three selected periods (g), (h), (i). The white lines denote faults. Period is indicated in the lower left corner of each map.

Figures 5g, 5h, 5i present the results of Rayleigh wave phase velocity maps at periods 5, 15 and 35 s. The maps show obviously lateral inhomogeneity and vertical variation in the crust. At short periods, the Rayleigh wave phase velocity is dominantly sensitive to the shallow crustal structure. And the distribution of phase velocities shows good consistency with surface geological features, with low-velocity anomalies for Tengchong volcanic region and the central Yunnan Basin. The longer periods mainly reflect the deeper crustal structure. As the period increases, the LVZs are gradually concentrated to the northern part of our study area.

2.3 Shear Wave Velocity Structure Inversion

Surface waves at different periods are sensitive to the Earth structure at different depths. In general, waves at short periods tend to sample materials close to the surface, whereas long periods are sensitive to deep structure. The maximum sampling depth of a Rayleigh wave is approximately one-third of its wavelength (Li H Y et al., 2011, 2009). More detailed interpretation of the observed phase velocity variations at different periods requires inversions for the shear wave velocity structure at different depths.

In this study, we constructed the shear wave velocity structure beneath Yunnan using a damped least-square inversion scheme (Herrmann and Ammon, 2004). We first extracted the pure path dispersion curves at each grid node and then inverted it for the 1D velocity profile (Xu et al., 2011). These 1D profiles were then interpolated to construct a 3D velocity structure model. As seen in Fig. 6, the sensitivity functions for Rayleigh phase velocities at periods of 5, 15, 25 and 35 s were calculated. The results indicated that the maximum sampling depth is down to ~50 km. During the inversion, we adopted an isotropic layered Earth model composed of 25 layers with 2-km layer-thickness overlying a half-space based on AK135 model (Kennett et al., 1995). It is well known that the surface waves are mainly sensitive to S-wave velocity comparing to P-wave velocity and density. Therefore the S-wave velocity in each layer is constant and taken as the inversion parameter. P-wave velocity and density are calculated from S-wave velocity using empirical formulas (Dziewonski and Anderson, 1981).

Figure 6. The sensitivity kernels for the fundamental mode Rayleigh phase velocity calculated from AK135 for four periods (5, 15, 25, 35 s). The sensitivity kernel can give a general indication of the depth sensitivity of a given period of surface wave.

From the results obtained in the previous section, the final S-wave velocity structures at depths 5, 10, 15, 20, 35 and 50 km are depicted in Fig. 7. In this study area, the results show that the LVZs within the crust are distributed at various depths and in different regions (Xu et al., 2013). Two representative velocity profiles from the surface to 50 km depth are shown in Fig. 8, which reveals the complex distribution of LVZs and earthquake activity in crustal depth.

Figure 7. Shear wave velocity variations at the depths of 5, 10, 15, 20, 30, and 45 km. The earthquakes with depth less than 15 km are projected in (a), (b) and (c); that in (d) and (e) are depth between 15 and 30 km; that in (f) are depth more than 30 km. Magnitude is distinguished by circle radius. The white lines denote fault. The blue dashed lines (AA', BB') in (a) indicate the vertical profiles shown in Fig. 8.
Figure 8. Vertical profiles of shear wave velocity variation along the lines AA', BB' in Fig. 7a. The black crosses represent earthquakes (Ms≥5.0) that occurred within a 30-km width from each profile. The wave velocity color scale is shown in the right. Tomography is depicted above each profile (black area) and the marks are the major location (black) and fault (red) along each profile.

Compared with previous studies, our model comprehensively displays the distribution of intra-crustal LVZs in the whole Yunnan region. The distribution pattern shows different features at different depths. At shallow depth (Fig. 7a), the results show strong lateral variation in LVZs distribution. They are mostly correlated with subsurface geological features and the previous results (He et al., 2005). For instance, sedimentary basins (e.g., Baoshan Basin and Chuxiong Basin) and Tengchong volcanic zone appear as prominent low-velocity anomalies, whereas Sanjiang fold region displays high-velocity anomaly corresponding to the mountainous topography.

From Figs. 7b, 7c, almost all prominent LVZs are observed in the southwest of the southern section of RRF and west of Chenghai fault (CHF) in the upper crust. In southern Yunnan region, the southern sections of Lancangjiang fault (LCJF) and RRF are two apparent transition zones of high-velocity and low-velocity anomalies, which proves the southeastward extrusion of Indochina Block (Xu et al., 2013). As an exception, prominent LVZs appear in the northeast of Chuxiong-Tonghai fault (CTF), which may be related to faulting and thickening of the basement layer (He et al., 2005). At the same time, notable LVZs liking a passageway appear between Lijiang-Ninglang fault (LNF) and CHF, showing significant differences from the features at shallower depth (Fig. 7a).

From depth of 20 to 30 km (Figs. 7d, 7e), the outline of the southern SYDB is emerging (Wei et al., 2010; Wang et al., 2003). The distribution pattern of LVZs matches the change in crustal thickness over this region. The LVZs is gradually concentrated on the north of RRF, but this feature seems to disappear at larger depths (Fig. 7f). In the lower crust, the distribution of LVZs shrinks northward further. In general, the distribution of LVZs is gradually narrowed northwards from the shallow depth to the deep depth (Figs. 7d, 7e, 7f). It is supported by the facts that the crustal thickness gradually increases from ~32 km in the southeast to ~56 km in the northwest of Yunnan (Li Y H et al., 2009).

Figure 8 displays shear wave velocity structure along two representative profiles. Along profile AA', the LVZs are at about 15 km depth beneath Tengchong volcanic area. And there are no obvious LVZs from middle-to-lower crust, which is identical with the results of Bao et al. (2015). Chuxiong Basin also shows visible LVZs in the upper crust. As shown in profile BB', the LCJF and the RRF is evident for its discontinuity in upper and middle crusts. It is clear that prominent LVZs exist not only in depth of 10–20 km (upper crust), but also in the depth of 30–40 km (lower crust) beneath Yunnan, which is consistent with the results of receiver function analysis by Hu et al. (2005a) and Xu et al. (2007).

4 DISCUSSION 4.1 Relationship between LVZs and Faults

The overlap of the main faults and the boundaries of the LVZs may indicate a close relationship between them. As shown in Fig. 7, some faults, such as the LCJF, RRF, XJF and Maitreya- Shizong fault (MSF), are situated in the boundary areas between low-velocity and high-velocity anomalies. In the upper crust, a majority of LVZs exist in the southwest of RRF (Figs. 7b, 7c). On the contrary, almost all LVZs are located at the northeast of RRF in middle and lower crust (Figs. 7d, 7e, 7f). The major boundary roughly outlines the distribution of crustal LVZs in Yunnan and is regarded as a major tectonic boundary between the South China and Indochina blocks (Xu et al., 2013; Lei et al., 2009; Schärer et al., 1994).

The LVZs is truncated at depth by aforementioned large fault zones, velocity changes are visible across them (Fig. 7). The material property on either side of the large faults may be different. For example, the southern part of LCJF shows the conversion from low-velocity anomaly to normal-velocity (Figs. 7d, 7e), indicating that it may not reach up to lower crust. Figure 7e shows that RRF is the southern border of SYDB, which has been inferred by Wang (2001) and Li Y H et al. (2009). RRF may have cut through the entire crust and reach up to the uppermost mantle because of the existence of adjacent LVZs in the lower crust (Li et al., 2008; Wang et al., 2003). The deep faults might play the role of pathways for energy transference from deep levels (Zhang and Wang, 2009). Huang et al. (2002) also found that some main fault zones are located in the boundary areas between high-velocity and low-velocity anomalies in the crust. The southeast of Yunnan region has obviously high-velocity anomaly, indicating that the local area is relatively stable tectonic with few faults.

Velocity anomalies may reflect fault shearing at depths. The observation suggests that major faults influence (or be influenced by) the distribution of LVZs. The frictional heating is released by the relative shear movements between the two contact blocks. The increasing temperature may reduce mid-lower crustal velocity along the main faults, and thus form the intra-crustal LVZs (Leloup et al., 1999). In other words, LVZs with lower viscosity would facilitate the relative motions of rigid upper crustal blocks, making the moving of the strike-slip faults less resistant. Furthermore, the mantle upwelling through the fault systems would lead to a penetration of heat flows into the crust, causing LVZs within the crust. In general, the influence of fault is stronger in the upper- to-middle crust than the lower crust. Both crustal LVZs and fault are important to the deformation of Yunnan.

4.2 LVZs and Strong Earthquakes Distribution

In order to investigate the relationship between crustal LVZs and strong earthquakes distribution in Yunnan, we use the earthquake catalog from China Earthquake Networks and count more than 200 events with Ms≥5.0 since the 1970s. We project the earthquakes with depth less than 15 km in Figs. 7a, 7b, 7c and earthquakes with depth between 15 and 30 km in Figs. 7d, 7e. The earthquakes with depth more than 30 km are illustrated in Fig. 7f. The spatial distribution of strong earthquake is extremely uneven in Yunnan but closely related to the shear wave velocity structure.

Zhao et al. (2000) found that large crustal earthquakes (magnitude 5.7–8.0, depth 0–20 km) from 1885 to 1999 in Japan occurred in or around zones of low seismic velocity revealed by seismic tomography. Their results indicate that large crustal earthquakes do not strike anywhere but closely linked to LVZs. Huang et al. (2002) have determined detailed P-wave tomographic images for the Sichuan-Yunnan region and found that most of the large crustal earthquakes (M > 5.0) were located in the boundary areas between slow and fast velocity anomalies or above low-velocity zones in the lower crust and uppermost mantle. Based on P-wave tomographic images beneath southeastern Tibet, Wei et al. (2010) believed that most of the large crustal earthquakes (M > 4.0) in this region are located in transitional areas between low-velocity and high- velocity zones. Although the tectonic background in Yunnan region is not exactly the same with above study areas, the strong earthquakes distribution feature is quite consistent. LVZs represent weak sections of the seismogenic crust, strong earthquakes likely occurred in LVZs or the transition zones between low-velocity and high-velocity anomaly. In our study area, most of the strong crustal earthquakes are shallow earthquakes and only very few occur in the lower crust. This fact implies that the crust is composed of a brittle, seismically active upper-to-middle crust and a warm, a seismic lower crust (Li et al., 2008; Wang et al., 2003).

LVZs always exhibit low electric resistivity, negative gravity anomaly, low-Q value and high heat flow, which may cause the weakening of the seismogenic layer in the crust (Huang et al., 2002). In the upper crust, Tengchong volcano and Simao Basin is clearly visible as LVZs. The former represents high temperature anomalies or magma chambers (Lei et al., 2009; Wang and Huangfu, 2004). Because the low-velocity anomaly is relatively weak, it is prone to strong crustal earthquakes under the tectonic stress and lateral extrusion (Figs. 7b, 7c). With increasing depth (Figs. 7d, 7e), the seismic activity in Tengchong-Longling area is stronger, but the earthquake cluster is moving southward. Qin et al. (1996) found the fact that the seismic activity and high heat flow in the area south of Tengchong volcanic center are more significant than that in the north. As shown in Figs. 7a, 7b, 7c, the Simao Basin shows low-velocity anomaly in the upper crust. It is caused by the existence of thick sedimentary layer (Wang and Gao, 2014). Seismic sounding data reveals that the upwelling of basaltic magmatic rocks intrude into the bottom of the basal layer with well-developed faults, which leads to crustal stress accumulation and an imbalance. Therefore, strong earthquakes occur frequently in this area (Hu and Gao, 1993). In the middle- to-lower crust, the LVZs anomaly in Simao Basin disappears with almost no earthquake. In the lower crust (Fig. 7f), the distribution of LVZs is moving northward with few earthquakes in whole Yunnan region.

As shown in the profiles (Fig. 8), it is also evident that most strong earthquakes occurred in or around LVZs in the middle and upper crust. In the edge portions of LVZs, the mechanical strength of materials is stronger than interior but still weaker than the normal sections of the seismogenic layer. Hence the edge portion of the LVZs becomes the ideal location to generate strong earthquakes that may produce active faults reaching to the Earth's surface (Zhao et al., 2000). Of course, the seismotectonics of Yunnan are very complex, and each seismic belt has its own special environment. There are few exceptional areas that are seismically active and yet occur in high-velocity regions. In order to determine the possible strong seismogenic pattern to explain these observations in Yunnan, much future work is needed by using and combining different geophysical imaging methods such as gravity, magnetotelluric, geoelectric, geothermal and geology methods.


In this study, we used ambient noise data from 55 broadband seismic stations to construct Rayleigh phase velocity maps from 5–35 s, and then inverted the 3D crustal structure in Yunnan. The results show clear variation of the crustal shear wave velocity. The detailed distribution of intra-crustal LVZs gives significant insights into the deformation and seismicity in the whole Yunnan region. Some conclusions can be summarized.

The distribution of LVZs shows different features at different depths. At shallow depths, the LVZs are well correlated with known subsurface geological features; from depth of 20 to 30 km, the LVZs are gradually concentrated to the north of RRF; in the lower crust, the distribution of LVZs shrinks northward further.

The LVZs are truncated at depth by several large faults in Yunnan. In general, the influence of fault is stronger in upper- to-middle crust than in the lower crust. Especially, RRF roughly outlines the distribution of crustal LVZs in Yunnan. The major faults influence (or be influenced by) the distribution of crustal LVZs, both of them are important to the deformation of Yunnan.

Since 1970s, the seismic activity reveals that, the spatial distribution of strong earthquake in Yunnan is uneven. The Yunnan strong earthquakes (Ms≥5.0) are more likely to occur in LVZs or the transition zones between low-velocity and high-velocity anomalies. Most of them are shallow earthquakes, occurred in the upper-to-middle crust. The crustal structure is composed of a brittle, seismically active upper-to-middle crust and a warm, aseismic lower crust in Yunnan.


We thank the editors and two anonymous reviewers for their useful comments. We thank T. B. Yanovskaya of Leningrad State University for providing the 2D tomography software and Robert Herrmann from Saint Louis University for his Computer Programs in Seismology (CPS) software package. We are also grateful to Prof. Huajian Yao, University of Science and Technology of China, Hefei, China, for providing the software package to extract the phase dispersion curves. Figures 1 and 58 are constructed using the GMT software (Wessel and Smith, 1998). Seismic data are provided by Yunnan Seismic Networks. This study was financially supported by the National 973-Project (No. 2013CB733303), the National Natural Science Foundation of China (No. 41474093), and the Key Natural Science Foundation of Hubei Province (No. 2014CFA110). The final publication is available at Springer via

Backus, G., Gilbert, F., 1968. The Resolving Power of Gross Earth Data. Geophysical Journal International, 16(2): 169-205. DOI:10.1111/j.1365-246x.1968.tb00216.x
Bai, D. H., Unsworth, M. J., Meju, M. A., et al., 2010. Crustal Deformation of the Eastern Tibetan Plateau Revealed by Magnetotelluric Imaging. Nature Geoscience, 3(5): 358-362. DOI:10.1038/ngeo830
Bao, X. W., Sun, X. X., Xu, M. J., et al., 2015. Two Crustal Low-Velocity Channels beneath SE Tibet Revealed by Joint Inversion of Rayleigh Wave Dispersion and Receiver Functions. Earth and Planetary Science Letters, 415(5): 16-24. DOI:10.1016/j.epsl.2015.01.020
Bensen, G. D., Ritzwoller, M. H., Barmin, M. P., et al., 2007. Processing Seismic Ambient Noise Data to Obtain Reliable Broad-Band Surface Wave Dispersion Measurements. Geophysical Journal International, 169(3): 1239-1260. DOI:10.1111/j.1365-246x.2007.03374.x
Cheng, C., Chen, L., Yao, H. J., et al., 2013. Distinct Variations of Crustal Shear Wave Velocity Structure and Radial Anisotropy beneath the North China Craton and Tectonic Implications. Gondwana Research, 23(1): 25-38. DOI:10.1016/
Dziewonski, A. M., Anderson, D. L., 1981. Preliminary Reference Earth Model. Physics of the Earth and Planetary Interiors, 25(4): 297-356. DOI:10.1016/0031-9201(81)90046-7
Fang, L. H., Wu, J. P., Ding, Z. F., et al., 2010. High Resolution Rayleigh Wave Group Velocity Tomography in North China from Ambient Seismic Noise. Geophysical Journal International, 16(2): 1171-1182. DOI:10.1111/j.1365-246x.2010.04571.x
Herrmann, R. B., Ammon, C. J., 2004. Surface Waves, Receiver Functions and Crustal Structure. Computer Programs in Seismology, Version, 3.30.[2018-4-9].
He, Z. Q., Ye, T. L., Su, W., 2005. 3-D Velocity Structure of the Middle and Upper Crust in the Yunnan Region, China. Pure and Applied Geophysics, 162(12): 2355-2368. DOI:10.1007/s00024-005-2780-x
Hu, H. X., Gao, S. Y., 1993. The Investigation of Fine Velocity Structure of the Basement Layer of Earth's Crust in Western Yunnan Region. Earthquake Research in China, 9(4): 356-363.
Hu, J. F., Su, Y. J., Zhu, X. G., et al., 2005a. S-Wave Velocity and Poisson's Ratio Structure of Crust in Yunnan and Its Implication. Science in China Series D:Earth Sciences, 48(2): 210-218. DOI:10.1360/03yd0062
Hu, J. F., Zhu, X. G., Xia, J. Y., et al., 2005b. Using Surface Wave and Receiver Function to Jointly Inverse the Crust-Mantle Velocity Structure in the West Yunnan Area. Chinese Journal of Geophysics, 48(5): 1148-1155. DOI:10.1002/cjg2.758
Huang, J. L., Zhao, D. P., Zheng, S. H., 2002. Lithospheric Structure and Its Relationship to Seismic and Volcanic Activity in Southwest China. Journal of Geophysical Research:Solid Earth, 107(B10): ESE 13-1-ESE 13-14. DOI:10.1029/2000jb000137
Kan, R. J., Hu, H. X., Zeng, R. S., et al., 1986. Crustal Structure of Yunnan Province, People's Republic of China, from Seismic Refraction Profiles. Science, 234(4775): 433-437. DOI:10.1126/science.234.4775.433
Kennett, B. L. N., Engdahl, E. R., Buland, R., 1995. Constraints on Seismic Velocities in the Earth from Traveltimes. Geophysical Journal International, 122(1): 108-124. DOI:10.1111/j.1365-246x.1995.tb03540.x
Lei, J. S., Zhao, D. P., Su, Y. J., 2009. Insight into the Origin of the Tengchong Intraplate Volcano and Seismotectonics in Southwest China from Local and Teleseismic Data. Journal of Geophysical Research, 114(B5): B05302. DOI:10.1029/2008jb005881
Leloup, P. H., Ricard, Y., Battaglia, J., et al., 1999. Shear Heating in Continental Strike-Slip Shear Zones:Model and Field Examples. Geophysical Journal International, 136(1): 19-40. DOI:10.1046/j.1365-246x.1999.00683.x
Li, H. Y., Liu, X., Li, X. F., et al., 2011. Rayleigh Wave Group Velocity Distribution in Ningxia. Journal of Earth Science, 22(1): 117-123. DOI:10.1007/s12583-011-0162-0
Li, H. Y., Su, W., Wang, C. Y., et al., 2009. Ambient Noise Rayleigh Wave Tomography in Western Sichuan and Eastern Tibet. Earth and Planetary Science Letters, 282(1/2/3/4): 201-211. DOI:10.1016/j.epsl.2009.03.021
Li, Y. H., Wu, Q. J., Zhang, R. Q., et al., 2008. The Crust and Upper Mantle Structure beneath Yunnan from Joint Inversion of Receiver Functions and Rayleigh Wave Dispersion Data. Physics of the Earth and Planetary Interiors, 170(1/2): 134-146. DOI:10.1016/j.pepi.2008.08.006
Li, Y. H., Wu, Q. J., Tian, X. B., et al., 2009. Crustal Structure in the Yunnan Region Determined by Modeling Receiver Functures. Chinese Journal of Geophysics, 52(1): 67-80.
Lobkis, O. I., Weaver, R. L., 2001. On the Emergence of the Green's Function in the Correlations of a Diffuse Field. The Journal of the Acoustical Society of America, 110(6): 3011-3017. DOI:10.1121/1.1417528
Moschetti, M. P., Ritzwoller, M. H., Shapiro, N. M., 2007. Surface Wave Tomography of the Western United States from Ambient Seismic Noise:Rayleigh Wave Group Velocity Maps. Geochemistry, Geophysics, Geosystems, 8(8). DOI:10.1029/2007gc001655
Paul, A., Campillo, M., Margerin, L., et al., 2005. Empirical Synthesis of Time-Asymmetrical Green Functions from the Correlation of Coda Waves. Journal of Geophysical Research, 110(B8): B08302. DOI:10.1029/2004jb003521
Qin, J. Z., Qian, X. D., Huangfu, G., 1996. The Seismicity Feature of the Volcanic Area in Tengchong. Seismol. Geomagn. Obs. Res., 17: 19-27.
Ritzwoller, M. H., Levshin, A. L., 1998. Eurasian Surface Wave Tomography:Group Velocities. Journal of Geophysical Research:Solid Earth, 103(B3): 4839-4878. DOI:10.1029/97jb02622
Roux, P., Sabra, K. G., Kuperman, W. A., et al., 2005. Ambient Noise cross Correlation in Free Space:Theoretical Approach. The Journal of the Acoustical Society of America, 117(1): 79-84. DOI:10.1121/1.1830673
Royden, L. H., Burchfiel, B. C., King, R. W., et al., 1997. Surface Deformation and Lower Crustal Flow in Eastern Tibet. Science, 276(5313): 788-790. DOI:10.1126/science.276.5313.788
Royden, L. H., Burchfiel, B. C., van der Hilst, R. D., 2008. The Geological Evolution of the Tibetan Plateau. Science, 321(5892): 1054-1058. DOI:10.1126/science.1155371
Sabra, K. G., Gerstoft, P., Roux, P., et al., 2005a. Extracting Time-Domain Green's Function Estimates from Ambient Seismic Noise. Geophysical Research Letters, 32(3): L03310. DOI:10.1029/2004gl021862
Sabra, K. G., Gerstoft, P., Roux, P., et al., 2005b. Surface Wave Tomography from Microseisms in Southern California. Geophysical Research Letters, 32(14): L14311. DOI:10.1029/2005gl023155
Schärer, U., Zhang, L. S., Tapponnier, P., 1994. Duration of Strike-Slip Movements in Large Shear Zones:The Red River Belt, China. Earth and Planetary Science Letters, 126(4): 379-397. DOI:10.1016/0012-821x(94)90119-8
Snieder, R., 2004. Extracting the Green's Function from the Correlation of Coda Waves:A Derivation Based on Stationary Phase. Physical Review E, 69(4): 046610. DOI:10.1103/physreve.69.046610
Wang, C. Y., Chan, W. W., Mooney, W. D., 2003. Three-Dimensional Velocity Structure of Crust and Upper Mantle in Southwestern China and Its Tectonic Implications. Journal of Geophysical Research:Solid Earth, 108(B9): 2442. DOI:10.1029/2002jb001973
Wang, C. Y., Huangfu, G., 2004. Crustal Structure in Tengchong Volca-no-Geothermal Area, Western Yunnan, China. Tectonophysics, 380(1/2): 69-87. DOI:10.1016/j.tecto.2003.12.001
Wang, Q., Gao, Y., 2014. Rayleigh Wave Phase Velocity Tomography and Strong Earthquake Activity on the Southeastern Front of the Tibetan Plateau. Science China Earth Sciences, 57(10): 2532-2542. DOI:10.1007/s11430-014-4908-2
Wang, Y., 2001. Heat Flow Pattern and Lateral Variations of Lithosphere Strength in China Mainland:Constraints on Active Deformation. Physics of the Earth and Planetary Interiors, 126(3/4): 121-146. DOI:10.1016/s0031-9201(01)00251-5
Wei, W., Sun, R. M., Shi, Y. L., 2010. P-Wave Tomographic Images beneath Southeastern Tibet:Investigating the Mechanism of the 2008 Wenchuan Earthquake. Science China Earth Sciences, 53(9): 1252-1259. DOI:10.1007/s11430-010-4037-5
Wessel, P., Smith, W. H. F., 1998. New, Improved Version of Generic Mapping Tools Released. EOS, Transactions American Geophysical Union, 79(47): 579-579. DOI:10.1029/98eo00426
Wu, J. P., Ming, Y. H., 2001. The S Wave Velocity Structure beneath Digital Seismic Stations of Yunnan Province Inferred from Teleseismic Receiver Function Modeling. Chinese Journal of Geophysics, 44(2): 228-237.
Xu, L. L., Rondenay, S., van der Hilst, R. D., 2007. Structure of the Crust beneath the Southeastern Tibetan Plateau from Teleseismic Receiver Functions. Physics of the Earth and Planetary Interiors, 165(3/4): 176-193. DOI:10.1016/j.pepi.2007.09.002
Xu, X. M., Li, H. Y., Gong, M., et al., 2011. Three-Dimensional S-Wave Velocity Structure in Eastern Tibet from Ambient Noise Rayleigh and Love Wave Tomography. Journal of Earth Science, 22(2): 195-204. DOI:10.1007/s12583-011-0172-y
Xu, Y., Yang, X. T., Liu, J. H., 2013. Tomographic Study of Crustal Velocity Structures in the Yunnan Region Southwest China. Chinese Journal of Geophysics, 56(6): 1904-1914. DOI:10.6038/cjg20130613
Yang, Y. J., Li, A. B., Ritzwoller, M. H., 2008. Crustal and Uppermost Mantle Structure in Southern Africa Revealed from Ambient Noise and Teleseismic Tomography. Geophysical Journal International, 174(1): 235-248. DOI:10.1111/j.1365-246x.2008.03779.x
Yanovskaya, T. B., Ditmar, P. G., 1990. Smoothness Criteria in Surface Wave Tomography. Geophysical Journal International, 102(1): 63-72. DOI:10.1111/j.1365-246x.1990.tb00530.x
Yanovskaya, T. B., Kizima, E. S., Antonova, L. M., 1998. Structure of the Crust in the Black Sea and Adjoining Regions from Surface Wave Data. Journal of Seismology, 2(4): 303-316. DOI:10.1023/1009716017960
Yao, H. J., Beghein, C., van der Hilst, R. D., 2008. Surface Wave Array Tomography in SE Tibet from Ambient Seismic Noise and Two-Station Analysis-Ⅱ. Crustal and Upper-Mantle Structure. Geophysical Journal International, 173(1): 205-219. DOI:10.1111/j.1365-246x.2007.03696.x
Yao, H. J., van der Hilst, R. D., de Hoop, M. V., 2006. Surface-Wave Array Tomography in SE Tibet from Ambient Seismic Noise and Two-Station Analysis-I. Phase Velocity Maps. Geophysical Journal International, 166(2): 732-744. DOI:10.1111/j.1365-246x.2006.03028.x
Yao, H. J., van der Hilst, R. D., Montagner, J. P., 2010. Heterogeneity and Anisotropy of the Lithosphere of SE Tibet from Surface Wave Array Tomography. Journal of Geophysical Research, 115(B12): B12307. DOI:10.1029/2009jb007142
Yao, H. J., Xu, G. M., Zhu, L. B., et al., 2005. Mantle Structure from Inter-Station Rayleigh Wave Dispersion and its Tectonic Implication in Western China and Neighboring Regions. Physics of the Earth and Planetary Interiors, 148(1): 39-54. DOI:10.1016/j.pepi.2004.08.006
Zhang, X., Wang, Y. H., 2009. Crustal and Upper Mantle Velocity Structure in Yunnan, Southwest China. Tectonophysics, 471(3/4): 171-185. DOI:10.1016/j.tecto.2009.02.009
Zhang, Z. J., Bai, Z. M., Wang, C. Y., et al., 2005a. Crustal Structure of Gondwana-and Yangtze-Typed Blocks:An Example by Wide-Angle Seismic Profile from Menglian to Malong in Western Yunnan. Science in China Series D:Earth Sciences, 48(11): 1828-1836. DOI:10.1360/03yd0547
Zhang, Z. J., Bai, Z. M., Wang, C. Y., et al., 2005b. The Crustal Structure under Sanjiang and Its Dynamic Implications:Revealed by Seismic Reflection/Refraction Profile between Zhefang and Binchuan, Yunnan. Science in China Series D:Earth Sciences, 48(9): 1329-1336. DOI:10.1360/01yd0567
Zhao, D. P., Ochi, F., Hasegawa, A., et al., 2000. Evidence for the Location and Cause of Large Crustal Earthquakes in Japan. Journal of Geo-physical Research:Solid Earth, 105(B6): 13579-13594. DOI:10.1029/2000jb900026