Anisotropy in the Earth's crust and uppermost mantle in southeastern Europe obtained from Rayleigh and Love surface waves

Digital seismograms from 25 earthquakes located in the southeastern part of Europe, recorded by three-component very broadband seismometers at the stations Vitosha (Bulgaria) and Muntele Rosu (Romania), were processed to obtain the dispersion properties of Rayleigh and Love surface waves. Rayleigh and Love group-velocity dispersion curves were obtained by frequency–time analysis (FTAN). The path-averaged shear-wave velocity models were computed from the obtained dispersion curves. The inversion of the dispersion curves was performed using an approach based on the Backus–Gilbert inversion method. Finally, 70 path-averaged velocity models (35 R-models computed from Rayleigh dispersion curves and 35 L-models computed from Love dispersion curves) were obtained for southeastern Europe. For most of the paths, the comparison between each pair of models (R-model and L-models for the same path) shows that for almost all layers the shear-wave velocities in the L-models are higher than in the R-models. The upper sedimentary layers are the only exception. The analysis of both models shows that the depth of the Moho boundary in the L-models is shallower than its depth in the R-models. The existence of an anisotropic layer associated with the Moho boundary at depths of 30–45 km may explain this phenomenon. The anisotropy coefficient was calculated as the relative velocity difference between both R- and L-models at the same depths. The value of this coefficient varies between 0% and 20%. Generally, the anisotropy of the medium caused by the polarization anisotropy is up to 10–12%, so the maximum observed discrepancies between both types of models are also due to the lateral heterogeneity of the shear-wave velocity structure of the crust and the upper mantle in the region.     

Crust and upper mantle structure of the Bohemian Massif from the dispersion of seismic surface waves

Summary The phase velocity dispersion of Rayleigh waves for the Moxa-Vienna (MOX-VIE) and Moxa-Kaperské Hory (MOX-KHC) profiles, and of both Rayleigh and Love waves for the Kaperské Hory-Ksi (KHC-KSP) profile have been measured and inverted into models of shearwave velocity vs. depth. The three paths cross, respectively, the central part of the Bohemian Massif, its western margin, and the Bohemian Pluton and Cretaceous. For the MOX-VIE profile mean and lower crustal shear wave velocities of 3.7 and 3.9 km/s, respectively, a mean Moho depth of 34 km, and no existence of a low-velocity layer in the lower crust were found. The model obtained for the MOX-KHC profile is characterized by a slightly lower velocity in the lower crust (3.8 km/s), by a slightly lower Moho depth (32 km), and by the appearance of a weak low-velocity channel between 55 and 140 km. The crustal section of the final model for the KHC-KSP profile agrees well with the KHKS82 model derived by Novotný from results of DSS along international profile VII. Our final Rayleigh-wave model has significantly lower shear-wave velocities down to 215 km in the mantle. A systematic difference of 0.18 km/s between the average velocities of Rayleigh and Love waves has been revealed for the depth range from 30 to 215 km. Since almost no contamination of the fundamental Love mode with higher modes has been observed, and since the investigated structure hardly contains an unresolved system of thin, alternately low- and high-velocity layers, the cause of the difference is evidently polarization anisotropy of the upper mantle beneath the Bohemian Massif. It is recommended that the discussed investigations should be supplemented with data from the fan of KSP-GRF (Gräfenberg Array, Germany) paths and from the KHC-BRG (Berggiesshübel, Germany) profile.     

Anelasticity of the crust and upper mantle beneath north and central India from the inversion of observed Love and Rayleigh wave attenuation data

The fundamental mode Love and Rayleigh waves generated by earthquakes occurring in Kashmir, Nepal Himalaya, northeast India and Burma and recorded at Hyderabad, New Delhi and Kodaikanal seismic stations are analysed. Love and Rayleigh wave attenuation coefficients are obtained at time periods of 15–100 seconds, using the spectral amplitude of these waves for 23 different paths along northern (across Burma to New Delhi) and central (across Kashmir, Nepal Himalaya and northeast India to Hyderabad and Kodaikanal) India. Love wave attenuation coefficients are found to vary from 0.0003 to 0.0022 km^–1 for northern India and 0.00003 km^–1 to 0.00016 km^–1 for central India. Similarly, Rayleigh wave attenuation coefficients vary from 0.0002 km^–1 to 0.0016 km^–1 for northern India and 0.00001 km^–1 to 0.0009 km^–1 for central India. Backus and Gilbert inversion theory is applied to these surface wave attenuation data to obtainQ ^–1 models for the crust and uppermost mantle beneath northern and central India. Inversion of Love and Rayleigh wave attenuation data shows a highly attenuating zone centred at a depth of 20–80 km with lowQ for northern India. Similarly, inversion of Love and Rayleigh wave attenuation data shows a high attenuation zone below a depth of 100 km. The inferred lowQ value at mid-crustal depth (high attenuating zone) in the model for northern India can be by underthrusting of the Indian plate beneath the Eurasian plate which has caused a low velocity zone at this shallow depth. The gradual increase ofQ ^–1 from shallow to deeper depth shows that the lithosphere-asthenosphere boundary is not sharply defined beneath central India, but rather it represents a gradual transformation, which starts beneath the uppermost mantle. The lithospheric thickness is 100 km beneath central India and below that the asthenosphere shows higher attenuation, a factor of about two greater than that in the lithosphere. The very lowQ can be explained by changes in the chemical constitution taking place in the uppermost mantle.     

Crustal Structure of the Korean Peninsula Using Surface Wave Dispersion and Numerical Modeling

Surface wave dispersion is studied to obtain the 1-D average velocity structure of the crust in the Korean Peninsula by inverting group- and phase-velocities jointly. Group velocities of short-period Rayleigh and Love waves are obtained from cross-correlations of seismic noise. Multiple-filter analysis is used to extract the group velocities at periods between 0.5 and 20 s. Phase velocities of Rayleigh waves in 10- and 50-s periods are obtained by applying the two-station method to teleseismic data. Dispersion curves of all group and phase velocities are jointly inverted for the 1-D average model of the Korean Peninsula. The resultant model from surface wave analysis can be used as an initial model for numerical modeling of observations of North Korean events for a velocity model appropriated to the Korean Peninsula. The iterative process is focused especially on the surface sedimentary layer in the numerical modeling. The final model, modified by numerical modeling from the initial model, indicates that the crust shear wave velocity increases with depth from 2.16 km/s for a 2-km-thick surface sedimentary layer to 3.79 km/s at a Moho depth of 33 km, and the upper mantle has a velocity of 4.70 km/s.     

Exploration of S-wave velocity profiles at strong motion stations in Eskisehir,Turkey, using microtremor phase velocity and S-wave amplification

We have explored 1D S-wave velocity profiles of shallow and deep soil layers over a basement at strong motion stations in Eskisehir Province, Turkey. Microtremor array explorations were conducted at eight strong motion stations in the area to know shallow 1D S-wave velocity models. Rayleigh wave phase velocity at a frequency range from 3 to 30 Hz was estimated with the spatial autocorrelation analysis of array records of vertical microtremors at each station. Individual phase velocity was inverted to a shallow S-wave velocity profile. Low-velocity layers were identified at the stations in the basin. Site amplification factors from S-wave parts of earthquake records that had been estimated at the strong motion stations by Yamanaka et al. (2017) were inverted to the S-wave velocities and Q-values of the sedimentary layers. The depths to the basement with an S-wave velocity of 2.2 km/s are about 1 km in the central part of the basin, while the basement becomes shallow as 0.3 km in the marginal part of the basin. We finally discussed the effects of the shallow and deep sedimentary layers on the 1D S-wave amplification characteristics using the revealed profiles. It is found that the shallow soil layers have no significant effects in the amplification at a frequency range lower than 3 Hz in the area.     

Love-Rayleigh wave incompatibility and possible deep upper mantle anisotropy in the Iberian peninsula

Love and Rayleigh wave phase velocities are analyzed with the goal of retrieving information about the anisotropic structure of the Iberian lithosphere. The cross-correlation method is used to measure the interstation phase velocities between diverse stations of the ILIHA network at periods between 20 and 120 s. Despite the 2-D structure of the network, the Love wave data are too few to enable an analysis of phase velocity azimuthal variations. Azimuthal averages of Love and Rayleigh wave phase velocities are calculated and inverted both in terms of isotropic and anisotropic structures. Realistic isotropic models explain the Rayleigh wave and short-period Love wave phase velocities. Therefore no significant anisotropy needs to be introduced in the crust and down to 100 km depth in the upper mantle to explain our data. A discrepancy is observed only at long periods, where the data are less reliable. Love wave data at periods between 80 and 120 s remain 0.15 km/s faster than predicted by isotropic models explaining the long-period Rayleigh wave data. Possibilities of biases in the measurements due to interferences with higher modes are examined but seem unlikely. A transversely isotropic model with 8% of S-wave velocity anisotropy in the upper mantle at depths larger than 100 km can explain the whole set of data. In terms of a classical model of mantle anisotropy, this corresponds to 100% of the crystals perfectly oriented in the horizontal plane in a pyrolitic mantle. This is a rather extreme model, which predicts at time delay between 0 and 2 seconds for split SKS.     

Modeling the ML4.7 mainshock of the February–July 2001 earthquake sequence in Aegion, Greece

An earthquake sequence comprising almost 2000 events occurred in February–July 2001 on the southern coast of the Corinth Gulf.Several location methods were applied to 171 events recorded by the regional network PATNET. The unavailability of S-wave readings precluded from reliable depth determination. For the mainshock of April 8, M_L= 4.7, the depth varied from 0 to 20 km. The amplitude spectra of complete waveforms at three local stations (KER,SER, DES; epicentral distances 17, 26 and 56 km) were inverted between 0.1 and 0.2 Hz for double-couple focal mechanism and also for the depth. The optimum solution (strike 220°, dip 40°, rake ‒160°, and depth of 8 km) was validated by forward waveform modeling.Additionally, the mainshock depth was further supported by the P- and S-wave arrival times from the local short-period network CRLNET (Corinth Rift Laboratory).The scalar seismic moment was 2.5e15 Nm,and the moment rate function was successfully simulated by a triangle of the 0.5 second duration. This is equivalent to a 1–1.5 km fault length, and a static stress drop 2–6 MPa. This value is important for future strong ground motion simulation of damaging earthquakes in Aegion region, whose subevents may be modeled according to the studied event. The T axis of the mainshock (azimuth 176° and plunge 67°), is consistent with the regional direction of extension N10°. However, none of the nodal planes can be associated to an active structure seen at the surface. The relationship of this earthquake sequence with deeper faults (e.g. possible detachment at about 10 km) is also unclear.     

Dispersion of surface waves recorded in Athens

Summary The records at Athens of 85 earthquakes with epicenters in several regions on the earth were used to determine group velocities along thirty five paths. The mean crustal thickness along each path has been estimated by comparing the observations withPress's standard curves. A linear relation has been found between the mean crustal thickness and mean elevation along each path. This relation is in agreement with Airy's isostatic hypothesis. Determination of Love wave dispersion along five paths and Rayleigh wave dispersion along two paths in southeastern Europe and northern Asia Minor gave values from 35 to 45 km for the crustal thickness in this region.     

利用瑞雷面波反演华北克拉通东北部边界的岩石圈结构

运用瑞雷面波相速度频散曲线分析和反演得到了华北克拉通东北部边界及其邻近区域岩石圈的精细S波速度结构.利用11个地震事件、60个台站的瑞雷面波波形资料,得到了周期从25-150 S的相速度频散曲线,并且通过线性反演方法得到了深度从40-300 km的S波速度结构.结果表明,该研究区域S波速度存在强烈的不均匀性.从东南部的...     

Shear Wave Velocity Structure of the Sinai Peninsula from Rayleigh Wave Analysis

The lithospheric structure of the Sinai Peninsula is shown by means of nine shear velocity profiles for depths ranging from zero to 50 km, determined from the Rayleigh wave analysis. The traces of 30 earthquakes, which occurred from 1992 to 1999 in and around the study area, have been used to obtain Rayleigh wave dispersion. These earthquakes were registered by a broadband station located in Egypt (KEG station). The dispersion curves were obtained for periods between 3 and 40 s, by digital filtering with a combination of MFT and TVF filtering techniques. After that, all seismic events were grouped in source zones to obtain a dispersion curve for each source-station path. These dispersion curves were inverted according to generalized inversion theory, to obtain shear wave velocity models for each source-station path, which is the main goal of this study. The shear velocity structure obtained for the Sinai Peninsula is shown through the shear velocity distributions with depth. These results agree well with the geology and other geophysical results, previously obtained from seismic and gravity data. The obtained velocity models suggest the existence of lateral and vertical heterogeneity. The shear velocity increases generally with depth for all paths analyzed in the study area. Nevertheless, in some paths a small low velocity channel in the upper or lower crust occurs. Along these profiles, it is found that the crustal structure of the Sinai Peninsula consists of three principal layers: upper crust with a sedimentary layer and lower crust. The upper crust has a sedimentary cover of 2 km thick with an average S-velocity of 2.53 km/s. This upper crust has a variable thickness ranging from 12 to 18 km, with S-wave velocity ranging from 3.24 to 3.69 km/s. The Moho discontinuity is located at a depth of 30 km, which is reflected by a sharp increase in the S-velocity values that jump from 3.70–4.12 to 4.33–4.61 km/s.     

The phase velocities of Rayleigh waves and the lateral variation of lithospheric structure in Tibetan Plateau

According to a Sino-U. S. joint project, eleven broadband digital PASSCAL seismometers had been deployed inside the Tibetan Plateau, of which 7 stations were on the profile from Lhasa to Golmud and other 4 stations situated at Maxin, Yushu, Xigatze and Linzhi. Dispersions and phase velocities of the Rayleigh surface waves (10s–120s) were obtained on five paths distributed in the different blocks of Tibetan Plateau. Inversions of the S-wave velocity structures in Songpan-Ganzi block, Qiang-Tang block, Lhasa block and the faulted rift zone were obtained from the dispersion data. The results show that significant lateral variation of the S-wave velocity structures among the different blocks exists. The path from Wenquan to Xigatze (abbreviated as Wndo-Xiga) passes through the rift-zone of Yadong-Anduo. The phase velocities of Rayleigh waves from 10s to 100s on this path are significantly higher than that on other paths. The calculated mean crustal velocity on this path is 3.8 km/s, much greater than that on other paths, where mean crustal velocities of 3.4–3.5 km/s are usually observed. Low velocity zones with different thicknesses and velocities are observed in the middle-lower crust for different paths. Songpan-Ganzi block, located in the northern part of Tibetan Plateau is characterized by a thinner crust of 65 km thick and a prominent low velocity zone in the upper mantle. The low velocity zone with a velocity of 4.2 km/s is located at a depth form 115 km to 175 km. While in other blocks, no low velocity zone in the upper mantle is observed. The value of Sn in Songpan-Ganzi is calculated to be 4.5 km/s, while those in Qiang-Tang and Lhasa blocks are about 4.6 km/s. The Chinese version of this paper appeared in the Chinese edition ofActa Seismologica Sinica,14, Supp., 566–573, 1992.     

Crustal and upper mantle structures of the brazilian coast

The Rayleigh wave phase and group velocities in the period range of 24–39 sec, obtained from two earthquakes which occurred in northeastern brazil and which were recorded by the Brazilian seismological station RDJ (Rio de Janeiro), have been used to study crustal and upper mantle structures of the Brazilian coastal region. Three crustal and upper mantle models have been tried out to explain crustal and upper mantle structures of the region. The upper crust has not been resolved, due basically to the narrow period range of the phase and group velocities data. The phase velocity inversions have exhibited good resolutions for both lower crust and upper mantle, with shear wave velocities characteristic of these regions. The group velocity data inversions for these models have showed good results only for the lower crust. The shear wave velocities of the lower crust (3.86 and 3.89 km/sec), obtained with phase velocity inversions, are similar to that (=3.89 km/sec) found byHwang (1985) to the eastern South American region, while group velocity inversions have presented shear velocity (=3.75 km/sec) similar to that (=3.78 km/sec) found byLazcano (1972) to the Brazilian shield. It was not possible to define sharply the crust-mantle transition, but an analysis of the phase and group velocity inversions results has indicated that the total thickness of the crust should be between 30 and 39 km. The crustal and upper mantle model, obtained with phase velocity inversion, can be used as a preliminary model for the Brazilian coast.     

Effect of lateral variation and model parameterization on surface wave dispersion inversion to estimate the average shallow structure in the Paraná Basin

The average layered structure of the intracratonic Paraná Basin, SE Brazil, is investigated with surface-wave group velocities from a small regional earthquake recorded by two broadband stations. Rayleigh and Love waves in the period range 1–4.2 s are used to infer average properties down to about 4 km. Genetic algorithm techniques are used to find the best fitting 1-D S-wave model. The inverted 1-D models show fair correlation with the average properties of the propagation paths as expected from geology and borehole information. However, different S-wave velocity models are obtained for the different inversion parameterizations. Since lateral heterogeneities are expected along the paths, several synthetic tests are performed with heterogeneous propagation paths. For approximately homogenous path (i.e., little lateral variation), the main features of the average synthetic model can be retrieved for different model parameterizations. For strong lateral variations, however, the average dispersion curve can produce very different 1-D inverted models depending on the parameterization. Also, the 1-D inverted models may differ significantly from the average properties of the inhomogeneous path, and wrong depths to interfaces may be inferred. For real data inversions, it is then suggested that various different parameterizations should be tested. If the resulting models show consistent features, this probably indicates homogeneity in the propagation path. But, if very different and unstable features are obtained in the 1-D inversions, then strong lateral variation may be present in the propagation path, and the average 1-D model may not represent average properties along the path.     

高频面波方法的若干新进展

面波多道分析方法(MASW)通过分析高频瑞雷波确定浅地表剪切波速度.在过去的20年中,由于该方法具有非侵入性、无损、高效及价格低的特点,越来越受到浅地表地球物理和地质工程学界的重视,视为未来最有希望的技术之一.这篇综述论文将介绍中国地质大学(武汉)浅地表地球物理团队近年来在研究高频面波的传播理论和应用中取得的部分成果.非几何波是一种仅存在于浅地表介质,尤其是未固结的沉积物中的独特的地震波.它的存在对快速而准确地获得表层S波速度有一定价值.我们的研究表明非几何波是一种具有频散特性的泄漏波.泄漏波的存在可能导致将其误认为瑞雷波的基阶或高阶能量,从而造成模式误判.这种模式误判会导致错误的反演结果.我们通过求取高基阶分离后的瑞雷波格林函数证明虚震源法瑞雷波勘探的可行性.这个结果将极大地降低野外瑞雷波勘探成本.勒夫波多道分析方法(MALW)中未知参数比瑞雷波的少,这使得勒夫波的频散曲线比瑞雷波的简单.因此,勒夫波反演更稳定,非唯一性更低.勒夫波数据生成的能量图像通常比瑞雷波的清晰,并具有更高的分辨率,从而可以更容易地拾取精确的勒夫波的相速度.利用雅克比矩阵分析波长与探测深度的关系表明对相同波长的基阶模式而言,瑞雷波的探测深度是勒夫波的1.3~1.4倍;而两种波的相同波长的高阶模式波的探测深度相同.我们也尝试了时间域勒夫波反演.按照勒夫波分辨率将地球模型剖分成了不同尺寸的块体,利用反卷积消除了地震子波对勒夫波波形的影响,通过更新每个块体的S波速度来拟合勒夫波波形,从而获得地下S波速度模型.该方法不基于水平层状模型假设,适用于任意二维介质模型.     

Tomography and velocity structure of the crust and uppermost mantle in southeastern Europe obtained from surface wave analysis

A set of two hundred shear-wave velocity models of the crust and uppermost mantle in southeast Europe is determined by application of a sequence of methods for surface-waves analysis. Group velocities for about 350 paths have been obtained after analysis of more than 600 broadband waveform records. Two-dimensional surface-wave tomography is applied to the group-velocity measurements at selected periods and after regionalisation, two sets of local dispersion curves (for Rayleigh and Love waves) are constructed in the period range 8–40 s. The shear-wave velocity models are derived by applying non-linear iterative inversion of local dispersion curves for grid cells predetermined by the resolving power of data. The period range of observations limits the velocity models to depths of 70 km in accordance to the penetration of the surface waves with a maximum period of 40 s. Maps of the Moho boundary depth, velocity distribution above and below Moho boundary, as well as velocity distribution at different depths are constructed. Well-known geomorphologic units (e.g. the Pannonian basin, southeastern Carpathians, Dinarides, Hellenides, Rodophean massif, Aegean Sea, western Turkey) are delineated in the obtained models. Specific patterns in the velocity models characterise the southeast Carpathians and adjacent areas, coast of Albania, Adriatic coast of southern Italy and the southern coast of the Black Sea. The models obtained in this study for the western Black Sea basin shows the presence of layers with shear-wave velocities of 3.5 km/s–3.7 km/s in the crust and thus do not support the hypothesis of existence of oceanic structure in this region.     

The lithosphere in the central-eastern Mediterranean area

The lithosphere beneath the central-eastern Mediterranean area has been investigated by the inversion of the regional dispersion relations derived from analysis of surface waves. It is possible to distinguish several types of crust with average S-wave velocities in the range 3.0–3.8 km/sec, and thicknesses varying from a minimum of about 30 km, which corresponds to the Apennines, Crete and Otranto Channel regions, to a maximum of about 51 km beneath the Ionian Sea, which can be considered as a submerged continent. Associated with these crustal features, large lateral variations have been detected in the lithosphere thickness, which varies from a minimum of about 30 km corresponding to the Tyrrhenian Sea and south of Crete to a maximum of about 130 km corresponding to south-eastern Alps and north-central Greece, while the sub-Moho S-wave velocity varies in the range 4.2–4.8 km/sec. The constraint furnished by our results to the geological-tectonic setting of the investigated area, characterized by the continent continent collision between Africa and Europe, is pointed out.Publication No. 405, P. F. Geodinamica, CNR, Roma, Italy.     

Apparent phase velocities and fundamental-mode phase velocities of Rayleigh waves

Rayleigh wave dispersion data usually appear in the form of multimodal spectra for a layered model structure. The number of dispersion modal curves depends on the number of layers in the model. The measured dispersion velocities from the frequency–wavenumber (F–K) space, however, may not represent the true phase velocities of the fundamental-mode dispersion curve, but apparent phase velocities. The present study discusses how multimode curves are generated in the multichannel analysis of surface waves (MASW) method and the cause of the apparent velocity produced by the F–K method. Results from a field trial demonstrate that apparent phase velocities fail to reveal thin layers with low velocities. A better agreement of the inverted model with the geotechnical record is obtained by using the data points extracted from the fundamental-mode curve of the MASW spectral image.     

Investigation of upper mantle structure under young regions of the southeast Pacific using long-period Rayleigh waves

Long-period Rayleigh wave group velocity dispersion curves are presented for paths across the east Pacific Ocean. The records are from the IPG site at Pamatai (Tahiti) and the IDA station at Nana. For the first time, direct-path observations of group velocities up to 300 s have been obtained. This study shows that for young oceanic regions group velocities are low even at long periods. The observations are interpreted in terms of an S-wave velocity model by a generalized inversion scheme. In the models for young ages, the low-velocity zone under the asthenosphere lid is well developed, with a strong velocity gradient at the bottom of this zone, followed at larger depths by a plateau representing a lower-velocity zone, and a marked gradient at 400 km.     

2-D Crustal Velocity Structure Along Hirapur-Mandla Profile in Central India: An Update

The 2-D crustal velocity model along the Hirapur-Mandla DSS profile across the Narmada-Son lineament in central India (Murty et al., 1998) has been updated based on the analysis of some short and discontinuous seismic wide-angle reflection phases. Three layers, with seismic velocities of 6.5–6.7, 6.35–6.40 and 6.8 km s^–1, and upper boundaries located approximately at 8, 17 and 22 km depth respectively, have been identified between the basement (velocity 5.9 km s^–1) and the uppermost mantle (velocity 7.8 km s^–1). The layer with 6.5–6.7 km s^–1 velocity is thin and is less than 2-km deep between the Narmada north (at Katangi) and south (at Jabalpur) faults. The upper crust shows a horst feature between these faults, which indicates that the Narmada zone acts as a ridge between two pockets of mafic intrusion in the upper crust. The Moho boundary, at 40–44 km depth and the intra-crustal layers exhibit an upwarp suggesting that the Narmada faults have deep origins, involving deep-seated tectonics. A smaller intrusive thickness between the Narmada faults, as compared to those beyond these faults, suggests that the intrusive activities on the two sides are independent. This further suggests that the two Narmada faults may have been active at different geological times. The seismic model is constrained by 2-D gravity modeling. The gravity highs on either side of the Narmada zone are due to the effect of the high velocity/high density mafic intrusion at upper crustal level.     

3-D Velocity Structure of the Middle and Upper Crust in the Yunnan Region,China

Using short-period (1–18 s) surface wave data recorded by 23 stations of the Yunnan Digital Seismic Network of China we determined phase velocities of the fundamental Rayleigh wave along 209 paths by the two-station narrowband filtering and cross-correlation method, followed by an inversion for phase velocity distributions at various periods using the Ditmar-Yanovskaya method. We then obtained a 3-D S-wave velocity structure of the middle and upper crust in the Yunnan region using the genetic algorithm. The results show strong lateral variation of phase velocity in the region. The short-period phase velocity variation is closely related to thickness variation of sedimentary layer in the shallow crust. Within the depth range of 26–30 km, the S-wave velocity in the Sichuan-Yunnan rhombic block is lower than in the surrounding areas. Most large earthquakes of M > 6.0 in Yunnan occurred in the transition zones between low and high velocities.