Global Mass Variations Inversed By Variance Constraint Point-mass Method

The Gravity Recovery and Climate Experiment(GRACE)mission was launched in March 2002,provides a direct approach to continuously monitor the variation of the timevarying earth's gravity field.However,affected by observation instruments error loaded in GRACE satellite,the high-frequency background force model and the near-polar orbit of GRACE satellite,the high-order terms of spherical harmonic potential coefficients of timevarying earth gravity field model of GRACE satellite are seriously attenuated by noises.The spatial distribution of surface quality variations inversed by the traditional spherical harmonic potential coefficient method shows serious north-south strip errors,for which smooth filtering approaches needed to be used.But it will inevitably filter out true signals and even change its spatial distribution while filters noises.Many scholars found compared to apply smooth filters,solutions with higher accuracy and spatial resolution can be inversed by applying time-space constraints to geophysical signals.Another mascon method used to calculate the variation of earth's surface quality was proposed,and has extracted the high precision and high resolution surface quality variation signal by applying time and space constraints in the solution process.The point-mass method is a novel method to calculate the variation of earth's surface quality in recent years.It establishes the functional relation between the quality variations and disturbation imposed in satellite through Newton's law of universal gravitation.Many scholars have calculated the quality variations of Chinese mainland,Amazon,Greenland and other regions by radial point-mass method and three-dimensional acceleration point-mass method,and consider the solutions better than the traditional spherical harmonic potential coefficients method.But current point-mass method also faces the problem that the north-south strips error and leakage error can not be completely removed.This problem probably induced by most studies only using the unit matrix or the constraint matrix based on spatial smoothing as the regularization matrix.In order to solve this problem,the regularization matrix is designed by using prior variance information of mass block's quality variations,and variance constraint radial point-mass method and variance constraint three-dimensional acceleration point-mass method are developed respectively.Differences of the two methods in inversion of surface quality variations are compared in detail.In order to solve the problem that a single regularization parameter can't adapt to the great difference of quality variation signals in different regions of the world,in this paper,three different iterative regularization matrices are designed to extract the quality variation signals contained in the observations step by step.In order to validate the effectiveness of the method developed in this paper,the global equivalent water height variation results from January 2003 to November 2014 are inversed,and solutions from GRACE spherical harmonic potential coefficient method,GLDAS2.1 hydrological model,CSR and JPL RL06 Mascon models are used to compare and analyze in cryosphere,land basin,global root mean square error,correlation coefficients,long-term trend,annual amplitude.The results show that variance constraint radial point-mass method and variance constraint three-dimensional acceleration point-mass method can effectively remove northsouth strips error,restrain leakage of land-sea signals,and greatly improve accuracy and spatial resolution of solutions,thus provides a novel method for calculating the quality variations of earth's surface with high precision and resolution.