Adaptive Multi-Scale Frequency-Domain Reconstruction for Multi-Source Gravity Data Fusion

The gravity navigation reference map is a foundational element of gravity matching-aided navigation systems, as it directly determines positioning accuracy and overall navigation performance. To improve fused gravity anomalies and reduce reliance on manually selected parameters, this study proposes an adaptive multi-scale frequency-domain reconstruction (AMFR) algorithm. Specifically, the target ocean covered by multi-source satellite gravity data is divided into multiple scales, and distinct fusion parameters are assigned according to the characteristics of multi-source gravity data in different sea areas to enhance the accuracy of the fused gravity navigation reference map. Seven fusion schemes with different filtering diameters and weighting strategies are designed for comparative analysis. We evaluate the algorithm’s performance using measured gravity data with varying spatial distributions from the South China Sea and the Western Pacific Ocean, employing both RMS comparison and matching positioning experiments. Experimental results demonstrate that the RMS of the gravity navigation reference map constructed by the proposed AMFR algorithm is reduced by 30.4%, and the matching positioning accuracy is improved by 23.4%. The AMFR algorithm effectively enhances the precision of the gravity navigation reference map and improves the navigation performance of the gravity matching-aided navigation system.