Compressing atmospheric data into its real information content

Abstract Hundreds of petabytes are produced annually at weather and climate forecast centers worldwide. Compression is essential to reduce storage facilitate data sharing. Current techniques do not distinguish the real from false information in data, leaving level meaningful precision unassessed. Here we define bitwise content theory for Copernicus Atmospheric Monitoring Service (CAMS). Most variables contain fewer than 7 bits per value highly compressible due spatio-temporal correlation. Rounding without zero facilitates lossless compression algorithms encodes uncertainty within itself. All CAMS 17× compressed relative 64-bit floats, while preserving 99% information. Combined with four-dimensional compression, factors beyond 60× achieved. A Turing test proposed optimize compressibility minimizing loss end use data.