Abstract:In view of the phenomenon of time-varying signal weakening and leakage in the process of GRACE data preprocessing, the scaling factors method and the forward modeling method are used to recover the lost signals. CSR Mascon products are adopted to compare the effects of the two recovery methods from the perspective of time series and spatial distribution. Results show: 1) The scaling factors method is relatively simple to implement, but the result of the scaling factors is completely dependent on the prior model, and when the reliability of the selected model is poor, the reliability of the GRACE inversion results are affected. 2) The restoration process of the forward modeling method is constrained by the original GRACE observations, and can restore the long-term trend and periodic signals, but there may be signal over-recovery in some areas. In restoring GRACE time-varying signals, we suggest using the forward modeling method.