Recovery And Refinement Of The Earth Temporal Gravity Fields From GRACE Observations

Since the launch of Gravity Recovery and Climate Experiment(GRACE)space gravity mission jointly by NASA and DLR in 2002 with a planned 5-year lifetime,GRACE data have been widely used in a number of disciplines to study geophysical processes including earthquake events,melting of icesheets,as well as oceanic and hydrologic processes.The majority of these studies relied on the monthly estimates of the Earth's gravity fields,which are publicly available as Level-2(L2)products officially released by Center for Space Research at the University of Texas(CSR),NASA's Jet Propulsion Laboratory(JPL),and the German Research Center for Geosciences Potsdam(GFZ).The popularity of GRACE in scientific community has fostered the proposal of Chinese GRACE mission,while the key technology of L1b data processing was once a great challenge for the last decade.Until recently,several Chinese institutes have got a breakthrough but it is still far from the level of the foreign institutes.In addition,the latest reports on GRACE indicate that GRACE L1b data has still potential to be improved.To develop a better algorithm towards a higher accuracy of gravity models from the current or next generation of GRACE data is one of the most attracting issues.As advertised by ICGEM(International Centre for Global Earth Models),over 150 static or temporal gravity fields have been established since 2000,while the majority of them are represented in terms of spherical harmonics(SH).This dissertation shall first outline the theory and methodology of establishing a temporal gravity field that is comparable to the one produced by official organizations.Subsequently,we will discuss several approaches of refining current GRACE gravity fields,including the use of kinematic orbits,the regional basis RBF as well as its higher version MBRF,and the effect of latest AOD products.The main contents and results are listed below:1.The nature of GRACE L1b raw data,the background models within orbit propogations,the orbit calibartion and gravity recovery through the variation-equation approach are firstly outlined.Based on the GRACE L1b raw data covering from 2005 to 2010,we have successfully produced an unconstrained monthly gravity field model(Hust-IGG01)up to d/o 60.Unlike the official data-processing centers,we employ the kinematic orbits instead of the GPS measurements as the pseudo observations.Meanwhile,an alternative model(Hust-IGG02)in use of the reduced-dynamic orbits as the pseudo observations is provided as well.We aim to understand the impacts that orbit pseudo observations cause on the accuracy of the ultimate gravity products.To this end,Hust-IGG01 and Hust-IGG02 are fully compared to each other,such that we are able to identify which type of orbit pseudo observations is more desired for the gravity inversion.Experiments demonstrate that Hust-IGG01 performs a better signal to noise level than Hust-IGG02.Over three typical regions of interest(Amazon,Greenland and Sahara),the mass variation derived from Hust-IGG02 has been under-estimated by about 5%-10%with respect to those from the official products,while Hust-IGG01 has achieved a fairly comparable accuracy.2.The regional geopotential representation RBF(radial basis function)is studied,which features to be highly localized and publicly known as a more appropriate base than spherical harmonic,as it is more prone to incorporate with regional geophysical a-priori information in regularization to model detailed gravity field accurately.The study assumes RBF scaling factors rather than SHC(spherical harmonic co-efficients)as the unknowns during the gravity inversion,and we successfully generated the RBF-based unconstrained model(namely,Hust-IGG03)as well as its constrained version(namely,Hust-IGG04).Comparisons among 2009-2010 time series of monthly gravity fields for GFZ RL05a,Hust-IGG03 and Hust-IGG04,demonstrate that RBF-based solution has several advantages over spherical harmonic solu-tion:(1)with an unit Tikhonov regularization matrix applied,Hust-IGG04 has evidently eliminated the striping error that severely bias the real gravity signal in unconstrained model Hust-IGG03,and performs a similar noise level as GFZ RL05a dealt by Gauss filtering with radius of 400km.Therefore users don't necessarily undergo the post-processing filtering on Hust-IGG04 to suppress noise any more;(2)Hust-IGG04 has revealed the gravity signal in a higher resolution than filtered GFZ RL05a product,on both of annual amplitude map and trend map during the period of 2009-2010,for instance Hust-IGG04 has shown an enhancement of ice-melting rate over southern Greenland amounting to 24%compared to the filtered GFZ RL05a product.Above results have demonstrated that RBF are not only capable of detecting gravity signal contents shown by classical spherical harmonic,but also achieving a higher signal resolution due to its feasibility of introducing a-priori information into regularization.3.The conventional spherical harmonic or regional geopotential representation like radial basis func-tion(RBF)doesn't provide an appropriate treatment of physical consistency between continental mass and passive ocean response.In this paper,we propose a MRBF approach by embedding the known coastal geometries in the RBF parameterization and imposing global mass conservation and equilibrium behavior of the oceans.Our hypothesis is that,with this physically justified constraint,the GRACE-derived gravity signals can be more realistically partitioned into the land and ocean contributions along the coastlines.The numerical results from GRACE L1b data inversion indicate that:(1)MRBF-based gravity modelling reduces the number of parameters by approximately 10%,and allows for more flexible regularization when compared to ordinary RBF solutions;and(2)the resulting MRBF mass flux is shown to better confine the coastal mass variability within the continents.The latter is particularly tested in the southern Greenland,and our results indicate that the trend of mass loss from the MRBF solution is approximately 2%?5%larger than that from the RBF solution.4.The official atmospheric de-aliasing(AD)products for GRACE gravity recovery are routely gen-erated by GFZ(called ATM),in order to minimize the potential temporal and spatial aliasing in Earth gravity field.However,the mis-modelling of the high frequent non-tidal effect still exists and has become one of the major error sources of GRACE mission.Moreover,developing the more precise AD prod-uct is especially critical for the next generation of GRACE-like mission.In addition to the ATM RL05 produced by GFZ,some alternative institutes make public their AD products as well,such as the ITG3D product by University of Bonn and the ESM product by ESA(European Space Agency).In this study,we carry out an evaluation of these three products(ATM,ESM and ITG3D)by analysis of K band range-rate(KBRR)residuals,and the numerical results show that:(i)compared to ATM,the KBRR-residuals over 2006 derived from ESM and ITG3D has been reduced by 2nm/s and 2.5nm/s respectively;and the spatial performance of ESM and ITG3D has been improved at majority of the world(78.4%and 78.9%);(ii)ESM has no evident differences with ITG3D,although ITG3D can reduce the KBRR-residuals by 1.8nm/s at regions of high latitude;(iii)ITG3D and ESM are consistent,while the jump in ATM product is identified and specified at Jan 29th,2006.We,therefore,are inclined that ITG3D is more preferred for removing atmosphere aliasing effect from current or even future GRACE mission.