Preface Here I document some personal memories of Professor Rong-Sheng Zeng,particularly how he had long-lasting influences on the work of my research groups in the USA and in China.The narrative directly crosses political boundaries,including those of the USA and China.Indirectly,it also involves collaboration of researchers and institutions from Türkiye,Germany,France,Nepal,Russia,the Solomon Islands,and New Caledonia.Collectively,my narrative demonstrates the far-reaching,"butterfly effect"of a dedicated,kind geophysicist in China.

A combination of 60 years of seismicity, over 45 years of fault plane solutions of moderate-sized to large earthquakes, and recent databases of high-resolution topography delineate spatial patterns of seismogenic faulting in northeastern Asia. Here I synthesize new knowledge with known features in a regional context. Strike -slip faulting characterizes active deformation along the northern and southeastern margins, as well as the interior of this vast region; while normal and reserve faulting dominate its northwestern/southwestern and eastern margins, respectively. Consistent patterns of transpressive and reverse seismogenic faulting persist along the Sakhalin-Hokkaido shear zone and the Eastern Japan Sea fold-and-thrust belt over lengths of over 1000 km each, respectively, but not as throughgoing faults. The latter transitions into strike-slip faulting farther south along the North Chugoku shear zone in southwestern Japan and continues southward into Kyushu. Meanwhile, the true intra-continental, strike-slip Tanlu fault zone reaches an uninterrupted length of about 2000 km. Fault plane solutions, including one from a recent event near the great intra-continental earthquake of 1668, provided much-needed new evidence for the dextral-slip nature of this long fault and a paleo-seismic study confirmed that the entire fault is seismogenic. Meanwhile, active segments of strike-slip faults elsewhere in the North China basin, of lengths less than about 200 km, have been responsible for devastating earthquakes. These observations are noteworthy as the current speed of the ground, based on Global Navigation Satellite System (GNSS) measure-ments, is only 3-4 mm/yr over a length-scale of 3000 km in the entire area of study. Regions of particular concern include: 1) The east coast of South Korea where recent, moderate earthquakes ruptured small sections of an active fault system that has a set of sharp fault scarps extending southward near the metropolitan area of Busan; and 2) the North China basin where intense historical seismicity contrasts with quiescence that persisted since 1976. Overall, no continental block, including the Amurian microplate, is well defined and only the oceanic Japan Sea exhibits little internal deformation.

ABSTRACTWe augment the method of virtual deep seismic sounding (VDSS) by adding the phases Sp, the SV-P conversion across the Moho, to determine the average speed of the S wave (VS) in the crust. VDSS uses the strong SV-P conversion below the free surface from teleseismic earthquakes as a virtual source for wide-angle reflections of the P wave. The large signal generated by the virtual source is the strongest aspect of VDSS in which no stacking is necessary to build up the signal. Previous work used the large moveout of the wide-angle reflection, phase SsPmp, relative to the direct S-wave arrival, phase Ss, to minimize the trade-off between bulk P-wave speed (VP) and thickness of the crust (H). It is then straightforward to use the timing of the phase Sp to constrain VS. As examples, we show that this method works for data from both temporary and permanent seismic deployments in contrasting tectonic settings. Specifically, VS under station FORT in western Australia and H1620 in central Tibet are 3.77±0.08 and 3.42±0.11 km/s, respectively. This development complements the undertaking of using information from only the S-wave train to extract all three seismic parameters of the bulk crust, VP, VS, and H. These parameters are important for constraining overall silica content of the crust.

We present 805 reliable estimates of crustal thickness in central Tibet, revealing an undulating Moho beneath a near-uniform surface of the highest and the largest plateau in the world. Our results are derived from wide-angle reflections of P-waves from virtual seismic sources near seismographs (Virtual Deep Seismic Sounding, VDSS). The virtual sources are produced by large earthquakes well-distributed in back-azimuth; and we also utilize obvious changes of measured arrival times as a function of rayparameter to constrain independently bulk thickness and P-wave speed of the crust (about 6.3 km/s), so there is little trade-off between the two parameters. Additionally, we consider effects of known heterogeneities in the upper mantle. In regions surrounding the Hi-CLIMB linear array, there is an eastward trend, in addition to the previously proposed northward trend, of crustal thinning. Respectively, these trends are approximately parallel and normal to the Indian mantle front (IMF), or the northern, leading edge of underthrust lithospheric mantle of India, deduced from a collection of additional geophysical evidence. The Moho near the IMF also shows large, 3-D variations in depths, prompting us to propose that the leading edge of intact Indian lower crust, or the Indian crustal front (ICF) that accompanied the underthrusting Indian lithospheric mantle beneath Tibet, lies along the zone of disturbed Moho. Based on results from about a dozen permanent broadband seismographs of the China National Seismic Network, more-or-less flat-lying, and thus presumably stable, cratonic Indian Moho south of the ICF apparently extends eastward for another 500 km to about 93 degrees E. (C) 2020 Elsevier B.V. All rights reserved.

We use virtual deep seismic sounding (VDSS) and data from ~1000 broadband seismic stations to provide high‐resolution estimates of crustal structure in the western Cordillera of the United States (U.S.). The most robust result is the geographic distribution of residual topography (that is, the difference between observed elevation and that expected from crustal buoyancy alone) and, by implication, thermal or petrologic anomalies in the mantle. Overall, residual topography of the western U.S. Cordillera varies considerably; with contrasts of up to about 3 km across distances of 200 km or less. High residual topography, indicating large mantle effects, is evident along the periphery of the Colorado Plateau and the surroundings of the Great Basin. In contrast, the central Colorado Plateau and the Wyoming Basin show low residual topography, close to what is expected of a geologically stable lithosphere. Overall, in regions to the east of the Wasatch hinge line (the eastern limit of significant extension in the North American cratonic basement) patterns of high residual topography and anomalies of low seismic wave speeds in the upper mantle are similar, suggestive of a common, thermal origin. In contrast, such a similarity is absent in regions to the west of the hinge line, suggesting substantial effects of petrological heterogeneities in the mantle. Finally, joint analyses of VDSS and conventional receiver functions reveal a wide range of crustal P wave speeds, locally as high as 6.7 km/s, perhaps indicating magmatic modification of the crust.

The method of virtual deep seismic sounding (VDSS) uses the large amplitude, postcritical reflection off the Moho, the SsPmp phase, to probe the Moho. In this study, we augment VDSS using the change in differential travel times between phases SsPmp and Ss ( T ) as a function of distance (moveout) to determine simultaneously both the thickness ( H ) and overall P wavespeed ( V P ) of the crust. Tests using synthetic data show that for typical uncertainties in measuring T (1 standard deviation of ±0.2  s), a minimum uncertainty of ∼±1.1  km and ±0.06  km/s can be reached for nominal values of H and V P , respectively. We then demonstrate its utility with field data recorded by two permanent stations in Australia.

In the past 500 years, 14 historical earthquakes, including one that caused a maximum intensity of IX, occurred over a distance of more than 300 km along the Longmen Shan thrust belt, including portions that did not rupture during the devastating Wenchuan earthquake sequence of May 12, 2008. Estimated locations of epicenters and trends of faulting during historical events complement information gathered after the 2008 sequence. In particular, in addition to the two fault splays that ruptured in 2008, there are additional seismogenic faults across the Longmen Shan belt, including those in the hinterland and in the foreland. In the latter case, the occurrence of moderate-sized historical earthquakes and the disproportionally wide region of reported damage in the Sichuan basin from large events within the Longmen Shan belt, probably an effect of low attenuation in the stable basin, calls for attention to the potential of seismic hazard in the heavily populated Chengdu basin. Moreover, long recurrence-intervals of great events, estimated from trenching of colluvium along the ruptures of the 2008 sequence, should not be taken as reliable estimates for somewhat smaller, but nonetheless highly destructive events along the Longmen Shan thrust belt.

The seismic data incorporated in global Moho models are sparse and therefore the interpolation of global Moho depths on a local area may give unrealistic results, especially in regions without adequate seismic information. Gravity inversion is a useful tool that can be used to determine Moho depths in the mentioned regions. This paper describes an interactive way of local Moho depth determination using the gravity inversion method constrained with available seismic data. Before applying inversion algorithms, the Bouguer gravity data is filtered in various stages that reduce the potential bias usually expected in Moho depth determination using gravity methods with constant density contrast assumption. A test area with reliable seismic data is used to validate the results of Moho computation, and subsequently the same computation procedure is applied to the Sri Lankan region. The results of the test area are in better agreement with seismically determined Moho depths than those obtained by global Moho models. In the Sri Lankan region, Moho determination reveals a fairly uniform thin crust of average thickness around 20 km. The overall result suggests that our gravity inversion method is robust and may be suitable for local Moho determination in virgin regions, especially those without sufficient seismic data. (C) 2013 Elsevier Ltd. All rights reserved.

Temperature beneath Tibet is poorly understood, constituting a critical gap in understanding the dynamics of the most prominent case of active continental collision. Here we present results of numerical simulations to provide new insight into the thermal evolution of Tibet. We utilize collective constraints provided by several large-scale field experiments in the past 20 yr, and include recent rock physics data to fully account for important feedback processes among temperature, shear stress, shear heating, thermal conductivity and specific heat. We show that while the collision system as a whole is cooled by the northward-advancing Indian lithosphere beneath Tibet, the upper and middle crust is warmed by shear heating between the overlapping lithospheres. Such a thermal structure readily explains the longstanding enigma of a very warm upper crust over a cold upper mantle. Gradual northward warming of the system is also consistent with a bimodal distribution of seismicity in the upper crust and the upper mantle beneath southern Tibet and the absence of deep seismicity further north. We emphasize the localized nature of shear heating, which is self-sustaining yet self-limiting, therefore does not depend on precise values of various input parameters, such as the rate of convergence and the amount of radiogenic heating. This heat source may have further implications for late-stage magmatic activities and variations of crustal rheology under Tibet.

This chapter contains sections titled: Introduction Background and Data Analysis Multiple Ruptures Implications on Seismogenesis Conclusions

SUMMARY We present a method that extends both the applicability and the quality of virtual deep seismic sounding (VDSS)—a technique for estimating crustal thickness that is robust even if the crust– mantle transition is complex or the crustal thickness is large. The results are important for studies of crustal contributions to isostasy and for understanding dynamic topography due to mantle convection. VDSS uses S -to- P conversions beneath seismic stations as virtual sources for large, post-critical reflections off the Moho, that is, the seismic phase SsPmp . Original applications of VDSS rely on deep earthquakes as sources of illumination to circumvent strong, near-source scattering (e.g. depth phases) and are, therefore, limited by the uneven distribution of deep seismicity. The method presented here effectively removes effects of the earthquake source wavelet (SW, including complexities arising from long, complicated source time functions and near-source scattering) and can be applied to signal from shallow and deep earthquakes. It involves two steps. First, based on analyses of particle motion, we separate ‘pseudo- P ’ and ‘pseudo- S ’ wave trains from the vertical and the radial component of ground motion. The latter is then used as the appropriate reference time-series for the deconvolution of the vertical and the radial component of ground motion. Since the reference time-series contains both the SW and S -type signals due to scattering near the receiver, the deconvolution also effectively removes S -type multiples, such as the phase SsPms and related reverberations. Applying this method to synthetic seismograms verifies that it is robust in removing complex SWs, even in the presence of random or signal-generated noise. The method is further validated using data recorded by the Hi-CLIMB array from both deep and shallow earthquakes. Impulsive signals are now routinely achieved, significantly improving both the quality and quantity of results from VDSS.

The Moho is not always a sharp interface; but seismic phase SsPmp yields robust, physically averaged estimates of crustal thickness (virtual deep seismic sounding, VDSS). In S. Tibet where the Moho is as deep as 75km, bimodal distribution of earthquake depths, with one peak in the upper crust and the other below the Moho, generated much interest in how lithological contrast affects seismicity and rheology. Generally seismicity is limited by distinct temperatures (Tc): 350±50°C in the crust and 700±100°C in the mantle (Earthquake Thermometry). Laboratory experiments show that distinct Tc reflect the onset of substantial crystal plasticity in major crustal and mantle minerals, respectively. Above these Tc, frictional instability ends due to velocity weakening of slip. So the seismic to aseismic transition is closely linked with brittle-ductile transitions in the crust and in the uppermost mantle, where the strength of the continental lithosphere is expected to peak (“Jelly Sandwich”). Plasticity depends exponentially on temperature (which evolves over time), so interplay between the geotherm and crustal thickness could result in concentrated seismicity in the upper crust — the only portion of a very warm lithosphere where temperature is below ~350°C (“Crème Brûlée”). Conversely, where the entire crust is below ~350°C (and the uppermost mantle is also below ~700°C), then earthquakes could occur over a wide range of depths, including the entire crust and the uppermost mantle (“Caramel Slab”).

Surrounded by seismicity and other manifestations of active deformation, the Ordos plateau, or the western portion of the North China craton (NCC), is a uniquely stable terrane in Asia. Results from virtual deep-seismic sounding and crustal receiver functions suggests that the crust under the eastern Ordos is thicker (at least 60km) than expected from previous studies and from its modest elevation (∼1500m above sea-level). Receiver functions also reveal a pronounced elastic impedance contrast within the crust (at ∼40km depth), which we interpret as the Conrad discontinuity. The presence of a 20km thick layer of mafic lower crust between the Conrad and Moho discontinuities would maintain crustal isostasy. The ∼1000km long seismic profile from the Ordos plateau in the west to the North China basin in the east reveals that crustal thickness changes by almost a factor of two across the active Shanxi rift in central NCC (over a distance of only about 100km). Insofar the current configuration of the lithosphere under the Ordos plateau might serve as a proxy for the initial condition prior to reactivation of the eastern part of NCC—where a cratonic keel no longer seems to exist—our results support the hypothesis that lower crust foundering was due to transformation of a thick mafic lower-crust to a garnet-rich assemblage (possibly caused by hydration associated with subduction during and/or before mid-Mesozoic times).

While the surface of Tibet is undergoing pervasive pure shear, stable terranes, straddling subsurface sutures, remain in the sub-continental lithospheric mantle (SCLM), attesting to its strength. Furthermore, sub-horizontal, cohesive remnant of Indian SCLM is traced northward from the Himalayan deformation front for about 600km, exemplifying the longevity of buoyant, strong SCLM of Archean shields. Bimodal distribution of earthquake depths, with peaks concentrating in the upper/middle crust and near the Moho, has been a longstanding evidence for strong SCLM. Recent results from the Himalayas—Tibet and along the East African rift system not only corroborate the bimodal distribution but also firmly established that large earthquakes occur below the Moho. Intriguingly, non-volcanic tremors—newly discovered mode of elastic strain release—also occur near the Moho but well below the seismogenic zone in the upper/middle crust. Considering recent field observations and laboratory experiments of viscosity contrast across the Moho, the SCLM must be strong enough to accumulate elastic strain, a prerequisite for earthquakes, over geological time. Moreover, under laboratory conditions, recent advances that link the termination of frictional instability, an analogue for earthquakes, and the onset of crystal plasticity, provided a physical basis for limiting temperatures of crustal (~300–400°C) and mantle (~600–700°C) earthquakes. While any single rheological model cannot possibly account for all tectonic settings (which also evolve with time), lithological contrast across the Moho is important in shaping the bimodal distribution of strength in the continental lithosphere.

Using data from regional earthquakes recorded by the Hi-CLIMB array in Tibet, we model P-wave arrival times to constrain the velocity structure in the crust and the upper mantle in central and western Tibet. Of more than 30 high-quality, regional seismic profiles that have been assembled, we have selected 10 that show excellent crustal and Pn arrivals for further analysis. Travel times from four events along the Hi-CLIMB array provide details on crustal velocities, and six events at regional distances to the array provide further constraints on Moho structure and upper-mantle-lid velocities. We use three-dimensional ray tracing to model the travel times, and the results indicate that the Moho beneath the Lhasa terrane of southern Tibet is over 73 km deep, with a Pn speed of about 8: 2 km/s. The Qiangtang terrane north of the Bangong-Nujiang suture (BNS) shows a thinner crust, by up to 10 km, and a lower Pn speed of 7.8-7.9 km/s. Travel times from events to the west and east of the array indicate that both Moho structure and mantle-lid velocities in the region are three-dimensional in nature but approximately follow the trend of the BNS. Although only a limited number of events were used for the travel-time modeling, the results are consistent with earlier results from teleseismic imaging using the Hi-CLIMB array.

We discuss an innovation in traveltime tomography that combines wavelet-based, multiscale parameterization and finite-frequency theory to solve two outstanding issues that inevitably arise from uneven source station distributions and from the three-dimensional (3-D) nature of wavefront healing: how to objectively address the intrinsically multiscale nature of data coverage while simultaneously maintain model resolution at each scale level. We apply the new, integrated methodology to investigate 3-D variations of P and S wave speeds (delta InVP and delta InVS) beneath the Himalayan-Tibetan orogen. In particular, we are able to constrain variations in the Poisson's ratio via delta In(V-P/V-S). The formulation is naturally data adaptive, resolving features at each scale only if the required data converge is available. The very first, long-wavelength feature that emerges is a clear anomaly of high delta InV that extends over more than 500 km beyond the northern edge of the Lhasa terrane at places. Farther northward, a strong negative anomaly underlies the region where recent volcanism occurs in northern Tibet. Regions of negative delta In(V-P/V-S) delineate a slab-like, subhorizontal feature concentrated between depths of similar to 100-250 km. Such characteristics are consistent with the notion that chemically refractory, and therefore buoyant, mantle lithosphere of the Indian shield ("Greater India") has advanced subhorizontally northward far beyond the surficial Bangong-Nujiang suture. In the crust, two isolated regions of low delta InV, each extending to depths near 100 km, occur along the Lunggar and the Yadong-Gulu active rifts in southern Tibet. Deep penetrating rifts imply that only a limited amount of horizontal displacement is being accommodated on subvertical structures.