A Signal-to-Noise Ratio Filter by Incorporating Spectral Characteristics of Temporal Gravity Field Signals and Varying GRACE Observation Conditions

Unconstrained monthly gravity field solutions of the Gravity Recovery and Climate Experiment (GRACE) and GRACE Follow-On (GRACE-FO) are predominantly influenced by correlated and high-frequency noise. To mitigate the effect of this noise, this paper proposes a signal-to-noise ratio (SNR) filter (SF) that incorporates the spectral characteristics of a priori monthly gravity field signals and varying observation conditions throughout the entire GRACE period. The performance of the SF filter was evaluated through a comparative analysis with the DDK filter and a combination filter of Gaussian and P4M6 (Gauss+P4M6). Compared to DDK3 and Gauss+P4M6 filters, the SF filter exhibits an improved SNR in mass change estimation under observation conditions characterized by poor data quality, repeat ground track, and normal observation periods. In global scale analysis, SF filtering exhibits a noise reduction of 51% and 81%, while retaining stronger amplitude and trend signals than DDK3 and Gauss+P4M6 filtering. Especially in surrounding regions such as the Greenland, Central Africa, and Amazon River Basin, higher SNR is achieved by the proposed SF filtering method. In small-scale regions like Greenland Sub Basins and other river basins worldwide, mass changes estimated using SF filtering demonstrate a better agreement with those from CSR Mascon solutions or GLDAS models. For extended analysis, the SF filter was further applied to GRACE-FO monthly solutions including CSR RL06.2, ITSG-Grace_op, and COST-G Grace-FO RL02, consistently achieving improved SNR in mass change estimation with respect to the other two filters.