3.1 Data acquisition
The geological structure around the Cuonadong dome is complex and includes multiple structural areas (STDS, Dongga syncline, Naji anticline, Cuonadong dome, etc.) and mining areas (Zhaxikang mining area, Mingsai mining area, Suoyue mining area, etc.). To study the deep geophysical characteristics and metallogenic background of the Cuonadong dome, the Institute of Geology, Chinese Academy of Geosciences set a deep reflection profile across the Cuonadong dome in 2019 (Fig. 1). The profile, which is approximately 40 km, crosses the core and the sides of the dome and passes through an area of mainly Quaternary sediments, Jurassic sandstones, slates, leucogranite, and Cambrian gneisses.
A total of 183 single-shot points were collected from deep reflection profiles, with specific acquisition parameters, as shown in Table 1. In the raw data (Fig. 2), surface waves are relatively developed, especially the near-offset surface wave signal (red circle), showing regular linear characteristics indicating strong energy. The far offset (green circle) can also track the surface wave signal. By extracting regular linear surface wave signals, the mapping quality of surface wave dispersion energy can be improved, the reliability of picking up dispersion curves can be enhanced, and the inversion results can be more accurate.
Table 1
Field data acquisition parameters
Acquisition parameters |
Shot interval | Small shot 250 m; Medium shot 3000 m; Large shot 50000 m |
Trace interval | 50 m |
Sampling rate | 2 ms |
Record duration | Medium, small 30 s; Large shot 60 s |
Offset | Small shot: 14975 m; Medium shot: 22475 m; Big shot: full array |
Receiving mode | Medium, small: 720 channel reception; Big shot; full array |
Instrument types | 428 Digital seismograph |
Source mode | Explosive source |
3.2 Spectrum analysis
Traditional multichannel surface wave data processing tends to suppress other forms of interference waves to achieve the desired filtering effect while also losing some of the effective signals. Spectrum analysis reveals that the surface wave energy is mainly concentrated between 6 and 20 Hz (Fig. 3a). We conducted different forms of bandpass filtering according to the frequency band range and obtained different dispersion energy maps. As shown in Fig. 3b, in the original single-shot data without filtering, the continuity of the fundamental-order dispersion energy is relatively excellent, rich in high- and low-frequency information, and it is easier to pick up the fundamental-order dispersion curve (Fig. 3b); the bandpass is 1-2-9-10-filtered, the continuity of the fundamental-order dispersion curve is still excellent, but some of the low frequency data is missed (Fig. 3c); the bandpass is 7-8-18-19, the low-frequency information in the dispersion energy is missed, and there is no continuity in the low-frequency signal before 7 Hz. The above tests show that preserving the surface wave signals in the main frequency range while losing more effective signals is not conducive to extracting low-frequency signals and studying deeper geological structures (Fig. 3d). Therefore, we use the original data to preserve the effective signals in different frequency ranges of the surface waves by intercepting the data and extracting the dispersion curves.
3.3 Channel number test
To obtain the reflection signal of the crust-mantle structure in deep reflection seismic exploration, the reflection data are collected by a large offset and explosive source. In multichannel surface wave imaging, a single-shot can reflect only a certain point of shear wave velocity information. Therefore, it is difficult to accurately assess the underground medium information by directly using the 720-channel reception data with the deep reflection single-shot data. However, when the number of receiving channels is small, it affects the mapping quality of dispersion energy, which is not conducive to improving the accuracy of dispersion curve extraction. Therefore, we conducted tests to select the most suitable multichannel collection method for data interception in our region. As shown in Fig. 4, (1) the tests of bilateral 60-channel reception and bilateral 40-channel reception are as follows: in the dispersion diagram with bilateral 60-channel reception, the continuity of the fundamental-order dispersion curve is better, especially for the low-frequency information, which can be continuously traced, even to 1.5 Hz, and the high-frequency signal is also more developed, which can be traced to 9.5 Hz; for bilateral 40-channel reception, the continuity of the fundamental-order dispersion is significantly reduced, and the low-frequency signal can be picked only to 2 Hz, but the continuity of the high-frequency signal is still better, and higher-order dispersion appears. (2) The single-sided 60-channel reception and single-sided 40-channel reception tests are as follows: for the single-sided 60-channel reception, the high- and low-frequency parts of the fundamental-order dispersion curve do not have continuity, the imaging quality is poor, and the continuity of the high-order dispersion is also poor; for the single-sided 40-channel reception, the fundamental-order dispersion curve between the 2.5-4 Hz part of the high-order dispersion dominates, although the continuity has improved, the imaging quality is still poor, and the high-frequency part of the dispersion curve is difficult to track. Through the comparison of different reception methods and different channels, the 60-channel bilateral reception method was finally adopted to extract the dispersion curves based on the consideration of improving the accuracy of the dispersion curve pickup and increasing the reliability of the inversion.
3.4 Forward simulation
The surface wave inversion is inseparable in a good initial model. Establishing a suitable initial model can provide better constraints on the inversion of the fundamental surface wave. Because the deep reflection profile across the Cuonadong dome is relatively long and passes through multiple tectonic units and ore-concentration areas, the regional velocity changes drastically, and the initial model is difficult to set. For data processing, we performed forward simulations by calculating the curvature of the original data dispersion curve and the sparsity of the dispersion points, given an initial model by the Parkseis surface wave software. The one-dimensional velocity structure is inverted according to the initial model, and the velocity structure obtained from the inversion is used as the forward model. The single-shot forward simulation is performed by setting the same acquisition parameters (Fig. 5), and the dispersion curve of the simulated single gun is inverted to obtain the one-dimensional transverse velocity structure (Fig. 5f). By comparison, for a detection point, the initial model is set by the dispersion curve curvature and sparsity, the obtained results are similar to the forward results, and the velocity structure has the same morphology; however, the actual data are influenced by topography, horizontal anisotropy, and multiple other convenient factors, which make it impossible to fit the forward results with certain inaccuracies. Therefore, in the process of data processing, we set the initial model of each point according to the curvature and sparsity of different dispersion curves to inversion one-dimensional velocity structure.
3.5 Dispersion curve extraction and reliability analysis
The extraction of the dispersion curve is the most important step in surface wave data processing. Reasonable acquisition of the dispersion curve is conducive to improving the inversion accuracy. In the process of picking the dispersion, the high-quality fundamental dispersion curve is picked by combining automatic picking with manual picking to remove the dispersion curve with poor quality. The total length of the deep reflection data line is approximately 40 km, and the line is mainly through the core and edge of the Cuonadong dome. The frequency band range of the dispersion curve varies greatly, but in the range of 1.5-8 Hz, the continuity of the dispersion curve is better, and the phase velocity is basically within 500–3500 m/s.
According to different structural positions, we selected three representative single shots for reliability analysis (Point 21433, Point 21695, and Point 21837). (1) Point 21433 (Fig. 6a) is located in the Quaternary sedimentary layer close to the Cuona Rift; the sedimentary layer is thicker, and the S-wave velocity is lower. The one-dimensional S-wave velocity indicates that at a position above 200 m underground, the velocity changes greatly, showing a trapezoidal increase, which may be related to the surface sediment and groundwater, and the velocity changes are small below 230 m. (2) Point 21695 (Fig. 6b) is located in the core of the Cuonadong dome, and it is mainly composed of leucogranite and metamorphic rocks. The dispersion curve shows that the phase velocity corresponding to a low frequency is significantly higher than that of other points. The S-wave velocity is generally high, and the minimum velocity is close to 1000 m/s. (3) Point 21837 (Fig. 6c) is located on a rare metal mineralization belt on the edge of the Cuonadong dome; around this point is mainly sandstone, slate, and schist. The dispersion curve indicates that the phase velocity increases significantly in the high-frequency and low-frequency parts, and the maximum S-wave velocity is close to 4000 m/s. This trend is mainly due to the existence of a metal metallogenic belt, which makes the consolidation and compactness of the rock increase, and the S-wave velocity increases significantly.
Through the above analysis, the different dispersion curves can reflect the characteristics of different tectonic and metallogenic zones, which are consistent with the actual stratigraphic and geological conditions and have higher accuracy. Finally, we used the selected fundamental dispersion curves and set different initial models for different measuring points according to the sparsity and curvature of each curve so that the final two-dimensional initial model is more representative and can reflect the real situation of the subsurface. To make the inversion results more accurate, the inversion fitting error was controlled to within 5% as much as possible. By inversion, the S-velocity structural characteristics of the east-west survey line were finally obtained (Fig. 7). The profile indicates that the Cuonadong dome shows obvious high-speed features underneath, and there are features of high- and low-velocity undulations in the core that may be related to the existence of hidden faults. The metamorphic rocks in the eastern part of the dome show more high-speed features, mainly because the rocks at this location are denser and have higher velocities.