Development Of Multi-surface Method And Its Application In Description Of The GPS Deformation Fields

The quick development of space earth observation technology represented by GPS provides rich and reliable crustal deformation constraints for seismic research. In GPS data products, strain rate field is widely used in earthquake prediction research because it can directly describe the crustal deformation rate and can be used to identify strain concentration region. On the other hand, because the actual GPS results include noise, and the distribution of the sites is inhomogeneous, so how to build a reasonable model to describe the GPS velocity and obtain reliable GPS strain rate field has been one of the hot issues in geoscience research.Aiming at the problem of reliable access to the GPS strain rate, this paper proposes and tests a variety of function models based on the classical multi-surface function formula, and the exponent function is selected to construct the kernel function, and then the model selection and programming are completed for the improved multi-surface function. On this basis, through a large number of experimental data, we obtained new kernel function parameter values range, and made preliminary application in strain rate field acquisition of the Chinese mainland, deformation identification before the Lushan earthquake and the Northridge earthquake, and the effect is good.1 Study on improvement and applicability of computing method of multi-surface function strain rateBy building the simulation data input(data distribution in 75°E-135°E,20°N-50°N,sampling density with 1°×1°) of known strain distribution, regular distribution, containing different amplitude error, and through a thousand data experiment, we obtained value range of the key parameters k and a of improved multi-surface function. In the cases of containing different error data input, we can obtain a better result when parameter a = 1.0, and this value is not affected by adjustment nodes and the additional error. Another key parameter k is affected by adjustment node density. When the adjustment nodes number is 53, the advantage value range of k is 0.0035-0.0080; when the adjustment nodes number is 100, the advantage value range of k is 0.008-0.02. Under the condition of the above parameters combination, based on the simulated data we obtained highly consistent results with analytical strain rate field distribution.Under the condition of a variety of data input, the improved method of multi-surface function showed stronger applicability compared with the classic methods, and the strain rate results are more stable. Compared with the least-squares collocation method, the improved method of multi-surface function has small deviation in the area of sparse data distribution and irregular geometric distribution, and there is a certain degree edge effect.In general, the kernel function of improved multi-surface function method is adopted index function, and it is a little more complicated than the classic methods, and the number of key parameters is 2 and is the same as the classic methods. According to our study, when a, one of the two key parameters of improved multi-surface function method, is 1.0, the effects is the best, and the value are fix in all examples of this article. Based on the current study, the improved multisurface function model can obtain better strain rate results when it gets only the value of parameter k, and this greatly enhances the operability of the improved method.2 The use of improved multi-surface function method for calculating strain rate in the actual deformation analysisUsing the improved multi-surface function method, we analyzed deformation characteristics before the Lushan earthquake and the Northridge earthquake. The 1999-2007 results show that compressive strain accumulation is relatively significant in the focal region of the Lushan earthquake, and shear deformation is not obvious along the rupture section plane. The surface deformation before the Northridge earthquake is dextral-compressive, and its strain accumulation rate is obviously bigger than that of the southern segment of Longmenshan fault where the Lushan earthquake happened.Coseismic displacement field of the Lushan earthquake is thrusting and has a small amount of sinistrogyration, which is incompletely the same as the strain accumulation of 1999-2007, and it reflects the strain accumulation characteristics of the whole seismogenic cycle. Maybe 10 years observation before earthquake cannot represent the average level of the whole seismogenic cycle. The co-seismic deformation of Northridge earthquake showed that thrust rupture was in dominant, and the horizontal movement in east and west directions and uplift movement were significant in near field which liked “dome” shape. And it is obviously that the earthquake rupture didn’t reach crustal surface.In general, the improved multi-surface function method has good effects in GPS strain rate calculation, and its effects are especially better than classical multi-surface function method in large spatial scale. In addition, the strain rate results have small deviation in the area of sparse data distribution and the edge of the study area, so we should pay attention to it in the practical application.