A combined EMD and MSSA model for the extraction of gravity tide correction from relative gravimetric data

Abstract

A combined empirical mode decomposition (EMD) and multichannel singular spectrum analysis (MSSA) model (EMD–MSSA model) was used for extraction of the gravity tide correction without a priori information (e.g., station coordinates) from static relative gravimetric data. Static observational data acquired using a CG-5 relative gravimeter over 16 days were used to investigate the feasibility and reliability of the proposed method. The singular spectrum analysis (SSA) method and empirical mode decomposition (EMD)–independent component analysis (ICA) method were also adopted for comparison. Experimental results show that the time series of the gravity tide correction estimated using EMD–MSSA, SSA and EMD–ICA methods are consistent with a theoretical reference (the Longman formula). The gravity tide correction estimated using the EMD–MSSA method is closer to the theoretical model than other methods, the root-mean-square difference of the residuals between estimated values and theoretical values are smallest, and the accuracy of the gravity tide correction time series derived using the EMD–MSSA method is thus highest. The correlation coefficient of extraction results and GT is highest for the results extracted using the EMD–MSSA method. The experimental results show that using the EMD–MSSA model, which combines the advantages of the MSSA and EMD signal processing methods, improves the extraction estimation accuracy and reliability of the gravity tide correction from relative gravimetric data.