For a given precision requirement, a wavelet-based data compression technique was applied to the temperature and relative humidity fields of the Eta-12 model. With the maximum allowable error less than 2^-3 degrees Kelvin, the wavelet compression procedure achieved an average compression factor of 45:1 for the quasi-stationary temperature field. The same procedure was applied to the relative humidity field, which is non-stationary compared to the temperature field. For this field the maximum error was set at less than 1%. With this pre-defined precision, the data compression procedure achieved an average compression factor of 30:1.

In this paper, details of the compression procedure and some preliminary test results are presented.