We use the movement velocities of GPS stations in western Sichuan province, China, to determine the distribution of horizontal strain accumulation before the 2008 Wenchuan Ms8.0 earthquake with a piecewise approximation approach and the coseismic displacements to determine the fault slips of the earthquake with an inversion analysis method. The results show that the distribution of the principal strain rates is strongly related to the active faults in the region, but along Longmenshan fault where the earthquake occurred, the strain rates are much lower than the others. The fault slip distribution shows mainly a thrusting with dextral striking, and the fault slips in the upper parts of the fault plane are in general bigger. Using the current strain accumulation rate and the released energy by the earthquake, we predict such a big earthquake in Longmenshan fault zone will happen in 460 to 1380 years.
On May 12, 2008, a disastrous earthquake of magnitude Ms8.0 occurred in Wenchuan, Sichuan province, China, killing more than 69,000 people and injuring over 370,000 people [
Geological setting. WMF is Wencuan-Maoxian fault, YBQF is Yingxiu-Baichuan-Qingchuan fault, GAF is Guanxian- Anxian fault, XSHF is Xianshuihe fault, ANHF is Anninghe fault, and LQSF is Longmenshan fault. The red circles represent the main- and aftershocks with magnitude greater than 6.0. The red lines show the active faults based on [
After the earthquake, the area attracted a lot of attentions in scientific community. Zhang et al. [
However, the derived strain rates are contaminated by the errors in GPS measurements. Their significance must be carefully assessed to distinguish between those computed strain rates generated from GPS observation errors and those of crust strain accumulation in the region. Moreover, the earthquake fault slips can also be estimated by the inversion analysis of coseismic displacements with a dislocation model in addition to the analysis of seismic waves as studied in [
Under a research project “Dynamic Process and Strong Earthquake Prediction of Block Boundaries,” 61 campaign-based and 4 continuously operating GPS stations in western Sichuan region were established and surveyed in 2005 by China Earthquake Administration (CEA). In addition, 22 GPS stations of the Crustal Movement Observation Network of China (CMONOC) were surveyed twice before 2005. All of these 83 GPS stations were resurveyed in 2005, 2006, and 2007. The surveys were carried out using Trimble 5700 GPS receivers with choke ring antennas and lasted for 22–24 hours each day for 3 consecutive days in each survey campaign. GAMIT and GLOBK software were used for the determination of the velocities of GPS stations in ITRF2005 frame based on the campaign mode and continuous GPS measurements. The rotation of the Eurasian plate was eliminated by carefully choosing Eurasia-fixed frame [
Movement observations of GPS stations. (a) The movement velocities of GPS stations before Wenchuan earthquake (In ITRF2005 with Eurasian plate fixed). (b) Horizontal coseismic displacements at GPS stations; red star is the locations of Wenchuan Ms8.0 earthquake.
Upon the earthquake, the State Bureau of Surveying and Mapping (SBSM) and CEA conducted GPS surveys in the area. The coseismic displacements at the GPS stations are obtained from the results of the continuous and campaign GPS surveys before and after the earthquake. Dual-frequency GPS receivers with choke ring antenna were used, and each station was occupied for two observation sessions of 23 hours each. The GPS surveys were completed on June 2, 2008. Also CEA resurveyed at their GPS stations in the area. These GPS observations were processed with GAMIT and GLOBK software to estimate the coseismic displacements. Figure
The approach of piece-wise approximation divides the area of interest into a number of triangles with GPS stations as their vertexes. The Delaunay method is applied to form triangle network by using the GMT software [
Let the velocities and their standard deviations of each station of a triangle be
We then test the significance of the estimated strain tensor. The null hypothesis
Okada [
The methodology discussed in Section
Principal strain rates distribution before Wenchuan earthquake. Double arrows show the principal strain rate labeled at the center of individual triangles. Red circles represent the main- and aftershocks with magnitude greater than 6.0. The Delaunay triangle network is formulated with GPS points as their vertexes. The red lines show the active faults based on [
We can see from Figure
The coseismic displacements in Figure
Fault slip distribution by inversion of horizontal coseismic displacements. Top left side is fault slip distribution in horizontal geodetic coordinate system, and bottom is fault slip distribution in the fault plane, and the circles at the heads of arrows show the 95% uncertainty error ellipses. Top right side is the histogram of data residuals.
Figure
Using the obtained fault slip distribution, the seismic moment
The proposed approach can provide the distribution of significant principal strain accumulation using the velocities of GPS stations, which is important information for studying crustal deformations and can help identify the areas of geological hazards. There is no significant strain accumulation in most parts of LMSFZ before Wenchuan Ms8.0 earthquake comparing to nearby active faults, such as XSHF and AZHF; this could be an indicator of earthquake potential. The fault slip distribution estimated from the coseismic displacements shows that Wenchuan Ms8.0 earthquake is mainly a thrusting in southwest part and with a significant dextral striking in the northeast part, and the fault rupture is mainly distributed at the depth less than 20 km. Using the estimated fault slip distribution, the energy released is approximately estimated, and magnitude of the earthquake is calculated, which agree well with the results derived from seismic waves. Under our assumption of uniform fault deficit accumulation rate of 1 to 3 mm/a, a large earthquake will take place in 460–1380 years.
This study was supported by the Hong Kong Research Grant Council (PolyU A/C: B-Q02B) and by China National Science Fund (no. 40674004). The figures are prepared using the GMT software.