In recent years, dynamic gravity measurement technology utilizing optical-gyro inertial navigation systems has advanced rapidly, owing to its high efficiency capabilities in acquiring Earth's gravity field data. However, conventional filtering methods exhibit limitations in addressing the frequency-domain aliasing of gravity signals and noise, which constrains the accuracy and resolution of measurements. To address these challenges, a novel joint noise reduction method integrating empirical mode decomposition (EMD) and wavenumber correlation filtering (WCF) was developed. In this method, the raw gravity data were low-pass filtered initially. The filtered results then underwent EMD, and a threshold was applied to retain low-order intrinsic mode functions. Subsequently, correlation filtering was implemented on repeated line retention results to reconstruct signals and suppress noise. Simulation experiments demonstrated that the proposed method achieved a 44% improvement in signal-to-noise ratio and a 48% reduction in mean square error compared to traditional EMD processing. These results confirm the effectiveness of the EMD-WCF method in balancing measurement accuracy and resolution, offering a new strategy for high-frequency gravity signal denoising.