Crustal Structure And Analysis Of Volcanic Activity In The Changbaishan Volcanic Area

The Changbaishan volcanic region is composed of several, mostly inactive, volcanoes, and is located on the border between NE China and North Korea. The basaltic base of the Changbaishan region was formed approximately 4.5 to 1.5 Ma, while the rylotitic and dacitic volcanic edifices are younger than about 3-1 Ma. The youngest volcano in the Changbaishan is the 2744m high Tianchi volcano. The Tianchi is capped by a summit lake (the Tianchi lake) and last erupted in 1215CE and 1702CE. There have been at least five large eruptions in the Changbaishan area during the common era. The eruption at about 1000CE was one of the largest eruptions in the last 2000 years, with ash deposits found up to 1200 km away in Japan. The total volume of rock erupted from the Changbaishan volcanic region is estimated to be about 9,000 km³.The Changbaishan region is seismically active, with an average of 18 earthquakes per month, although during the summer of 2003, more than 30 were detected per month. The majority of the hypocenters of these events cluster near the Tianchi caldera lake in the upper 5 km of the crust, and while the Changbaishan region is much more seismically active than the surrounding area, it is far less active than other volcanic regions. There are several hot-springs surrounding the Tianchi volcano, mostly concentrated on the northern slope, and they primarily discharge meteoric water that has circulated in regions of hot rock at upper crustal depths. Tang et al. (1998) resolved a low-resistivity body below about 10 km under the Tianchi, possibly reaching depths up to 30 km, and extending about 20 km to the north of the Tianchi summit.In this study, both broadband seismic data and seismic refraction/wide-angle reflection data were analyzed by Receiver Function method and Ray Tracing Inversion method, respectively. The velocity structures of Changbaishan region are presented and the possible relationship with magma chamber activity is analyzed.Crustal structure determined by modeling receiver functionsIn the summer of 1998, the State University of New York at Binghamton, in collaboration with the Research Center of Geophysical Exploration (RCGE) of the Chinese Seismological Bureau, installed 19 portable, broadband seismic stations in NE China, in the region of the Changbaishan volcanic area. The data from this seismic network are analyzed in this study.Out of 41 earthquakes considered, 29 events below 48 km depth (one event was shallow, with a depth listed as 33 km) and at a range of 30-908 were used in this study. Two criterions were employed for event selection. The first criterion was used to avoid complex near-source structure effects, while the second was used to ensure that the direct P arrived well before secondary P phases. To ensure that the signal was strong only large events were selected. The minimum magnitude event found to equalize well was 5.3 Mb, although this depended on the signal-to-noise ratio of a given seismogram. Most of the accepted events detected by the temporary broadband seismometers were located south of the network, since the highest concentration of seismicity meeting the above criterion was in the seismic belt from Indonesia through Papua New Guinea to the Fiji islands. One event was from the Mariana Trench, three events from the north (the Aleutian Trench region) and one from the west Oocated in Iran). Due to equipment failures, power outages, periods of unusually high anthropogenic noise levels, and differing durations of deployment not all of the events were recorded at all of the stations. In this study, total 109 seismograms recorded at the 19 stations were used for modeling.The first 20 or so seconds following direct P in a teleseismic seismogram principally contains rupture effects, source-side structure, and receiver-side structure. The source-side signal results from near source reflected P waves and S to P conversions. Receiverside signal consists of P to S conversions and P and S reflections near the receiver, principally in the crust. Direct P is incident on the Mono nearly vertically for a teleseismic distance source. Therefore, the amplitudes of S conversions are larger on the horizontal components, while the P signals are larger on the vertical component. Since the horizontal components of the seismogram principally contain receiver-side effects, it is possible to equalize the source effects and retain the receiver-side effects by deconvolving the vertical component from the radial component leaving a radial receiver function. In order to avoid division by small numbers, the equalization procedure of uses a water-level deconvolution where signal with spectral power below some threshold, called the water level, is discarded. This equalization method includes a Gaussian filter in the deconvolution, which acts as a low-pass filter removing noise in the seismograms. Because the receiver function method uses crustal converted and reflected phases to determine crustal structure, the region of crust sampled depends on the offset of these crustal reverberations. The offsets of the reverberations depend on the ray parameter (p) and the crustal thickness (or more generally the depth to the deepest interface). The p was estimated from the event depth and range using the IASP91 model of P global seismic travel times. The lateral extent from the station that the crust is sampled by a receiver function is at least the distance to the penetration point of the first bounce P arrival. In this region where the crust is expected to be about 30-35 km thick, with an average P wave velocity of 6.0-6.5 km/s, and for the data set used, the receiver functions sample the Moho about 40-50 km from the stations. In this area, the lateral resolution of P on the Moho is about 10-15 km.After identifying all usable seismograms, the receiver functions were determined using a constant Gaussian width parameter of 2.4 Hz and varying water levels (from 5×10^-5 to 10^-2). Water level was determined largely by the quality of the original seismogram and stability of the equalization. The radial receiver functions were interpreted using three analysis techniques which estimate differing sets of model parameters: a slant-stacking procedure, a direct search forward modeling scheme, and a least-squares inversion (LS). In all of these methods, that the crust is locally homogeneous, isotropic, and that the Moho is not dipping was assumed. While using the above three techniques does not provide independent measurements, due to the differing methods and model parameters estimated, the three methods indicate robustness of the final results.Receiver functions recorded at MANG do not suggest a low velocity anomaly and we concluded that the western extent of the low velocity anomaly is located under the western flank of the Changbaishan. One velocity profile determined at WUSU station also did not contain a midcrust low velocity zone. With these data we were unable to discern the eastern extent of this low velocity region; however, receiver functions sampling the crust to the south of DRAG are not indicative of a midcrust low velocity anomaly.P-Wave Velocity Structure from Wide-Angle Reflection and Refraction DataIn 1998, Geophysics Exploration Center of China Earthquake Administration conducted seismic refraction/wide-angle reflection experiment in Changbai volcano area. This study also analyzes seismic arrivals collected by receivers placed along two primary lines in this experiment, the first trending from south to north, called line SN, and the second trending west to east, called line WE. Lines SN and WE are 270 and 203 km long and are composed of 66 and 57 recorders, respectively. There were three explosions along line WE, and four along line SN; however, the northernmost explosion on line SN was not used, since the signals from this source were not detected by enough recorders. The explosions were composed of 1.2-1.5 tons of TNT distributed among several shallow wells separated by 3-15 m.Based on apparent velocity analysis, following phases were identified: the upper crust refracted phase (P_g), three mid-crust refractions/reflections (P₁, P₂ and P₃), the Moho reflection (PmP), and the upper mantle refracted phase (P_n). In total, 397 and 315 travel-time picks along lines SN and WE are used, respectively.Throughout the inversion process, a graphical user interface of the inversion program of Zelt and Smith (1992), RayGUI, that was developed by J. Song during this study was used based on the modeling approach outlined by Zelt (1999). The 2D velocity models consist of several layers containing smooth velocity gradientsTravel-times predicted by the final P-wave velocity (v_p) models for lines SN and WE fit the observed travel-times with an RMSE of 0.1654 and 0.1636 seconds, and predict 93.7% and 94.9% of the travel-time picks, respectively. In general, the observed travel-times are fit better away from the Tianchi volcano. This is most likely due to an increased complexity of velocities near the Tianchi.Let's refer to the crust away from the LVZ as unperturbed crust, and the velocities in the unperturbed crust increase with depth rapidly to 6.0 km/sec in the upper several kilometers. The velocities then increase less rapidly to about 6.4 km/sec at depths of about 20 km, increasing slightly faster to about 6.8-7.8 km/sec near the Moho transition. The Moho is deepest under the Tianchi and shallowest under the LVZ. The depth of the Moho is at about 40 km under the Tianchi, 30-35 km under the LVZ, and about 35 km elsewhere.The LVZ is located roughly between the Tianchi summit and the intersection of the two lines. The velocities in the LVZ are as low as 5.4 km/sec, roughly 0.6 km/sec slower than the velocities immediately adjacent to the LVZ. Taking the 6.0 km/sec velocity contour as a proxy for the edge of the LVZ, the LVZ is located between about 75-150 km along line SN and 205-240 km along line WE, with the top of the LVZ about 10-15 km below the surface. The LVZ is at most about 10-15 km thick, and it is thickest to the north of the Tianchi volcano.The reliability of the final P velocity models depends on a number of factors, including the accuracy of the individual travel time picks, the uncertainty of the apparent velocities of the phases, the type of phases, and the ray coverage. However, the resolving power of data can be appraised by attempting to recover a particular model feature by inverting synthetic travel-times. The most prominent result of this study is the presence of anomalous low velocities in the crust to the north of the Tianchi volcano, and to assess whether this result is robust, the source/receiver geometry of line SN was used to attempt to recover a similar anomaly in a trial model.Crustal Structure and Magama ChamberReceiver Function results also inferred two regions of low velocities in the upper to middle crust, one to the south and one to the north of the Tianchi. In the velocity models of Receiver Function, the crust is unperturbed south of about 20 km and north of about 175 km along line SN, and that west of about 160 km and east of about 280 km along line WE the crust is unperturbed, similar to in these 2D velocity models. The northern LVZ of Receiver Function is located in the depth range of about 8-16 km and between about 75-125 km along line SN, and v_p is as low as 5.0-5.4 km/sec. The southern LVZ of Receiver Function is located approximately between 20-60 km along line SN, with v_p as low as about 5.0 km/sec, and extends from depths of about 15 km to the Moho. The northern LVZ of Receiver Function is roughly consistent with these results, although the velocities and the depth of the LVZ do not overlap; however, a low velocity anomaly in the velocity model of line SN that would correspond to the southern LVZ of Receiver Function is not resolved. The northern LVZ of Receiver Function was better constrained than the southern LVZ of Receiver Function, the latter was only constrained with two receiver functions from the broadband station CANY (slightly to the south of the Tianchi summit. Even though there are not as many crossing rays-paths under the Tianchi compared to the more northern regions of line SN, resolutions tests indicate that the velocities are relatively well constrained in this region of the line. Hence, the differences between refraction/wide-angle reflection results and those of Receiver Function are most likely due to the relatively unconstrained velocity-depth trade-off in the receiver function method, which is less sensitive to absolute velocities and more sensitive to velocity impedance.Dike ModelingThis study also investigated the relationship between ground deformation from dike emplacement and geometries using analytical models. A free surface half space model of Coulomb 2.5 was used in the investigation. Basic dike parameters investigated are: dip, depth of the top, depth of the bottom, and thickness of dilation, and the cross-sections are presented to illustrate temporal vertical deformation. In vertical dike (dip 90°) models, vertical ground deformation increases while the dike develops upward, and the deformation pattern is symmetric about the dike, forming two peaks at both sides of the dike where the lateral extent of surface vertical deformation increases as the dike develops. For a non-vertical dike, vertical deformation is asymmetric about the dike and extends further laterally when compared to the vertical case. Deformation amplitude is greater as well. Surface expression is negligible when the depth of dike bottom is greater than 6 km. Vertical deformation reaches 0 and is downward at the location where the dike will open on the surface. For models with dikes dipping greater than 45°, along a perpendicular bisector starting from center of the dike vertical deformation decreases on the dipping side of the dike to negative values and then back to positive values, forming a second peak. In models with dikes with dip less than 45°, vertical deformation is negative after the initial decrease near the dike location. Finally, vertical deformation increases when the thickness of dike dilation increases. These models can help guide the initial interpretation of leveling observations made in volcanic areas.