Study On The Recovery Of The Earth's Gravity Field From The Combination Of Multi-Type Satellite Gravimetry Data

The successful implementation of CHAMP, GRACE and GOCE satellite has opened up a new era for satellite gravimetry. It also provides an important technical support to explore the meticulous structure of the earth's gravity field and its temporal variation. With the deep understanding for the error characteristics of the satellite gravimetry observations and the continuous updating of the prior background geophysical models, how to further explore the effective information of multi-type satellite gravimetry data, and to improve the accuracy of the earth's static and temporal gravity field model, is one of the hot spots currently. In this context, the spectral characteristics of the multi-type satellite gravimetry data are analyzed in this paper. Based on the dynamic integral approach and the space-wise least square method, the GRACE static and temporal gravity field model and GOCE static gravity field model are inversed respectively. With the least-squares combined adjustment method and least-squares spectral combination method, the combined inversion of high accuracy and high resolution gravity field model based on multi-type satellite gravimetry data is realized. In addition, the inherent defects of current satellite gravity missions are analyzed, which can accumulate the useful experience for the next generation missions in the near future.The main work and contributions in this dissertation are as following: Based on the principle of dynamic integral approach, the spectral relationship between the multi direction observations and the earth gravity field is anlyszed in detail, and a new dynamic integral approach which takes the weight ratios of multi direction observations is built. The GOCE HL-SST observations are also introduced to verify this new approach. The results show that the spectral response between the multi direction observations in different frames and the earth gravity field informations are different, but the equal weighted solutions in different frames have a good consistency. In contrast, the weighted solutions which adequately take the relationships between each component into consideration have separate accuracy curves, and the solution in the LNOF is the best. Aiming at solving the large scale linear equations during the earth's gravity field model determination efficiently, the intensive task in the least squares solution process are analyzed at first. The parallel algorithm based on the combination of OpenMP, MPI and Intel MKL are created to build the design marix and normal equation, which provides an effective tool for high order gravity field model inversion with multi-type satellite gravimetry data.The principle of dynamic integral approach for combining the LL-SST and HL-SST data is analyzed, and the reliability of this approach and its relative software is verified by the simulation data. Based on this software, a new GRACE model WHU-GRACE2015s truncated to 150 degrees is inversed by 5 years of GRACE observations. The calculation results show that WHU-GRACE2015s model has a good consistency with GGM05s, Tongji-GRACE01S and AIUB-GRACE03S.The principle of space-wise least square method based on gravity gradient data is studied. After introducing the cascade filter, which is the combination of high pass filter and AR filter, the color noises in GOCE SGG data are whited efficiently. In order to solve the ill-posed problem in normal equation, the Kaula regularization is introduced. Eventually, a new GOCE model WHU-GOCE2015s, which is truncated to 210 degree, is invsered by the combination of GOCE HL-SST and SGG data. The final results demonstrate that our model is better than the first generation of space-wise solution at low degrees, and better than the first generation of space-wise solution and direct solution at high degrees. In addition, our model is better than first generation of time-wise solution at the whole frequency band.After introducing the principle of least-squares combined adjustment method and least-squares spectral combination method, the comparsion between this two mothods are implemented with both simulated and raw observations. Eventually the least-squares combined adjustment method is selected to inverse the combined model WHU-GOGRA2015S, which contains both GRACE and GOCE observation information. WHU-GOGRA2015S model is truncated to 210 degree, and its accuracy is close to GOCO01S model.Different strategise for processing low frequency noise in GRACE KBRR data is introduced. The practical calculation formulas for PPS and PSS strategies are demonstrated. Based on these two strategies, the filter strategies named FPS and FSS are also introduced. The simulated and raw observations are used to assess these four different low frequency noise processing strategies. The final results show that PPS strategy can lead to the increase of condition number of normal equation, the decrease of accuracy of inversed low degree spherical harmonic coefficients, and the temporal signal loss. Comprehensive considering the calculation accuracy and efficiency, the FPS is proposaled for low frequency processing.After analyzising the rank defect formative factor for nomal equation during the determination of spherical harmonic coefficiens with LL-SST data alone, the dynamic integral which only using the GRACE KBRR data is built, and a new temporal gravity field WHU-Grace01s truncated to 60 degrees is inversed. Since only the GRACE KBRR data in introduced, WHU-Grace04s is not sensitive to the orbit resonances, and its accuracy is better than other models at high degrees and orders.The influence of the important parameters, such as the truncated order, the arc length, the calibration parameters, and the number of iterations, on the gravity field model determination with dynamic integral approach is analyzed, and the temporal gravity field model WHU-Grace2015m is inversed with the combination of GRACE orbit and KBRR data. The truncated degree and order of WHU-Grace2015m is 60. The results from the time domain and frequency domain show that our model is quite accurate with the CSR RL05 model, JPL RL05 model and GFZ RL05 model.A new formula for orbit repeat period calculation is formed, and the fast algorithm for optimal orbit selection is introduced. At the same time, the relationship between orbit repeat period, GRACE, Pendulum, Bender configurations and inversion accuracy of earth's gravity field model is discussed. The results demonstrates that the rational altitude and inclinaton of orbit can avoid the sparse groundtrack orbits, while Pendulum and Bender configuration can obtain multi direction observations. Comprehensively taking the orbit repeat period and orbit configuration into consideration, the accuracy of inversed temporal gravity field models can improve efficiently.Based on the control variable method, the formative factors of stripe error for GRACE-type mission are analyzed. And the impact of the geophysical models error for temporal gravity field model inversion is estimated. The results demonstrate that, the special orbit characteristics of GRACE and the errors in observations and prior background geophysical models, including ocean tide model and non-tidal model, affect the inversed solutions accuracy. At the current range rate measurement of accuracy, the observation errors should be taken into account primarily. In the near future, we need to pay more attention to the accuracy and the way to use background geophysical models, which will affect the accuracy of inversed gravity field models seriously.With the simulated observations of e2.motion mission, a new temporal gravity field model truncated to 120 degree is determined. Due to the high accuracy of each payload and rational orbit configuration, e2.motion solution can reflect the area and location of temporal signals more precisely than GRACE solution, and its accuracy can increase about 60% to 90%.