Evaluation of the Consistency of Three GRACE Gap-Filling Data

• Published in: Remote sensing (Basel, Switzerland) Switzerland), 2022-08, Vol.14 (16), p.3916
• Main Authors: Qian, An, Yi, Shuang, Li, Feng, Su, Boli, Sun, Guangtong, Liu, Xiaoyang
• Format: Article
• Language: English
• Subjects: Annual variations; Artificial intelligence; Collocation methods; Consistency; Data recovery; Datasets; Decomposition; Empirical analysis; gap-filling; Geodesy; GRACE (experiment); Gravitational fields; Gravity; Immunoglobulin G; Machine learning; Missing data; Noise levels; Orthogonal functions; Remote sensing; Robustness (mathematics); satellite gravity; Satellite laser ranging; Satellite observation; Satellite tracking; Satellite-to-satellite tracking; Satellites; sea level change; Seawater; spatial analysis; Spectrum analysis; Spherical harmonics; Time series; time variable gravity; time-series analysis; Water masses; Water storage
• ISSN: 2072-4292; 2072-4292
• DOI: 10.3390/rs14163916

The Gravity Recovery and Climate Experiment (GRACE) gravity mission has become a leading platform for monitoring temporal changes in the Earth’s global gravity field. However, the usability of GRACE data is severely limited by 11 months of missing data between the GRACE and GRACE Follow-on (GRACE-FO) missions. To date, several approaches have been proposed to fill this data gap in the form of spherical harmonic coefficients (an expression of the Earth’s gravity field, SHCs). However, systematic analysis to reveal the characteristics and consistency of the datasets produced by these latest gap-filling techniques is yet to be carried out. Here, three SHC gap-filling products are systematically analyzed and compared: (1) Combining high–low satellite-to-satellite tracking with satellite laser ranging (SLR) observations (QuantumFrontiers, QF), (2) SLR-based recovery incorporating the GRACE empirical orthogonal function decomposition model proposed by the Institute of Geodesy and Geoinformation at the University of Bonn (hereafter, denoted as IGG), and (3) applying the singular spectrum analysis approach (SSA). The results show that (1) the SHCs of the QF, IGG, and SSA data are consistent up to degree 12; (2) the IGG and SSA data give similar results over the 11 gap months, but the IGG shows a faster increase in the mean ocean water mass and the SSA appears to better capture the interannual variation in the terrestrial water storage; and (3) the noise level increases significantly in the high-degree terms (l > 16) of the QF data, so these data are only applicable for large-scale mass migration research. These results provide a reference for users to select a gap-filling product. Finally, we propose a new scheme based on the triple collocation method to derive a weight matrix to fuse these three datasets into a more robust solution.