Assessing the Eeects of Data Selection with Dao's Physical-space Statistical Analysis System

A global, physical-space statistical analysis system is used to assess the eeects of data selection in optimal interpolation (OI). Both systems use identical error covariance statistics and rst guess, and are presented with the same observational data. Spectral decomposition of analysis increments in terms of spherical harmonics reveals that OI increments have an excessive amount of energy in higher wavenumbers and an unrealistically large ratio of divergence to vorticity. At least in part, dynamical imbalances in the OI analyzed state can be attributed to approximate method of solution of the analysis equation, and are not entirely due to the crude geostrophic constraint built into the covariance model.