Merging GRACE Follow-On gravity fields with various approaches

For the 2002–2017 period, an essential information about global monthly gravity variations was provided by the Gravity Recovery and Climate Experiment (GRACE) mission. Gravimetric mission have given a new opportunities to analyze fluctuations in the Earth's system, so GRACE observations is widely used in various fields of science. After a gap of almost a year, in May 2018, the GRACE mission successor, i.e., the GRACE Follow-On (GRACE-FO) mission was launched. Both mission observations are processed and supplied by various centers. Therefore, in the following research, we determine a monthly homogenous gravity fields for GRACE-FO mission based on recent Science Data System (SDS) solutions. We use all available 44 months (June 2018 to March 2022) of GRACE-FO monthly data in spherical harmonics form up to degree and order 96. Data is provided by SDS centers, i.e., the Center for Space Research (CSR; the United States), the German Research Center for Geosciences (GFZ) and the NASA’s Jet Propulsion Laboratory (JPL; the United States). To obtain merged gravity fields, we test algorithms based on non-iterative: (1) coefficient-wise and (2) field-wise weighting methods, and iterative: (3) variance component estimation (VCE) method, which help to eliminate signal noise left from each data sets after Gaussian spatial smoothing. In a case of weights, we obtained similar magnitude of them for three selected datasets for non-iterative methods, but CSR weights dominate for the VCE method. In the study, our solution was analyzed in spectral and spatial domain of gravity signal (spherical harmonics case) and land hydrology (total water storage case; TWS). The analysis of signal information contained in each degree and order of spherical harmonic coefficients notice that largest amount of information is contained up to 60 degree and order for SDS and weighted solutions. Above degree and order 60 is more than two times less information. The greatest signal differences between our and SDS solutions occur for the sectorial coefficients up to 40. We also show that the applied field-wise weights much more effectively remove remaining noise after spatial averaging than per-order/degree weighting. Moreover, the obtained values of signals variance for our solutions concur with the geophysical models. In case of land hydrology, the used weighting approaches reduce root mean square scatter of TWS by 5-15% for continental areas. The largest differences occur mainly in high latitudes for both hemispheres. Regional analyses (e.g. Amazon, Zambezi, Murray Darling river basins) show good agreement of monthly TWS series between original GRACE-FO (from SDS centers) and our solutions. We received the extreme TWS differences up to 15% depending on the data center (CSR, GFZ or JPL) and weighting method.