A Lossy Floating-Point Compression Method for Meteorological Navigation Data Based on Mantissa Bit Optimization

Meteorological navigation data are critical to navigational safety and route planning, but their massive volume imposes a heavy burden on data storage and communication in intelligent shipping systems. Considering the spatiotemporal characteristics of such data and the fact that they allow lossy compression within an acceptable error range, this paper proposes an efficient error-guaranteed lossy compression method for floating-point data. The method modifies the binary representation of mantissa bits within a preset error bound to increase the number of trailing zeros in the XOR results between adjacent values, thereby reducing the meaningful bits that need to be encoded, and selects corresponding encoding strategies according to the resulting XOR patterns. Experimental results on public datasets and real meteorological navigation datasets show that, compared with existing mainstream compression algorithms, the proposed method incurs slightly higher compression time but achieves the highest compression efficiency under the preset error constraints, demonstrating its suitability for scenarios with strict error control requirements in large-scale meteorological navigation data storage and transmission.