Crustal structure of the northeastern Tibetan plateau,the Ordos block and the Sichuan basin from ambient noise tomography

We apply ambient noise tomography to significant seismic data resources in a region including the northeastern Tibetan plateau, the Ordos block and the Sichuan basin. The seismic data come from about 160 stations of the provincial broadband digital seismograph networks of China. Ambient noise cross-correlations are performed on the data recorded between 2007 and 2009 and high quality inter-station Rayleigh phase velocity dispersion curves are obtained between periods of 6 s to 35 s. Resulting Rayleigh wave phase velocity maps possess a lateral resolution between 100 km and 200 kin. The phase velocities at short periods （〈20 s） are lower in the Sichuan basin, the northwest segment of the Ordos block and the Weihe graben, and outline sedimentary deposits. At intermediate and long periods （〉25 s）, strong high velocity anomalies are observed within the Ordos block and the Sichuan basin and low phase velocities are imaged in the northeastern Tibetan plateau, reflecting the variation of crustal thickness from the Tibetan plateau to the neighboring regions in the east. Crustal and uppermost mantle shear wave velocities vary strongly between the Tibetan plateau, the Sichuan basin and the Ordos block. The Ordos block and the Sichuan basin are dominated by high shear wave velocities in the crust and uppermost mantle. There is a triangle-shaped low velocity zone located in the northeastern Tibetan plateau, whose width narrows towards the eastern margin of the plateau. No low velocity zone is apparent beneath the Qinling orogen, suggesting that mass may not be able to flow eastward through the boundary between the Ordos block and the Sichuan basin in the crust and uppermost mantle.

Public cloud computing for seismological research:Calculating large-scale noise cross-correlations using ALIYUN

The amount of seismological data is rapidly increasing with accumulating observational time and increasing number of stations, requiring modern technique to provide adequate computing power. In present study, we proposed a framework to calculate large-scale noise crosscorrelation functions(NCFs) using public cloud service from ALIYUN. The entire computation is factorized into small pieces which are performed parallelly on specified number of virtual servers provided by the cloud. Using data from most seismic stations in China, five NCF databases are built. The results show that, comparing to the time cost using a single server, the entire time can be reduced over two orders of magnitude depending number of evoked virtual servers. This could reduce computation time from months to less than 12 hours. Based on obtained massive NCFs, the global body waves are retrieved through array interferometry and agree well with those from earthquakes. This leads to a solution to process massive seismic dataset within an affordable time and is applicable to other large-scale computing in seismological researches.