A number of issues have yet to be addressed before the ensemble Kalman Filter (EnKF) can be successfully applied to operational atmospheric data assimilation. These relate to the required ensemble size and concern model-error representation and balance in a primitive-equation context.

To examine these issues, a sequential ensemble Kalman Filter (EnKF) has been used to assimilate simulated radiosonde, SATEM, and aircraft reports into a dry global primitive-equation model. The model uses the simple forcing and dissipation proposed by Held and Suarez. It has 21 levels in the vertical and a 144 x 72 horizontal grid. In total, about 80,000 observations are assimilated per day.

A method of generating (approximately) balanced model perturbations is used to generate the initial ensemble and to simulate model error. In this study, the model-error statistics (like the observation-error statistics) are assumed to be known.

A perfect-model experiment and experiments with simulated model error are performed in this environment to examine the issues mentioned above. These experiments include a series of data assimilation cycles with different configurations of the EnKF. The results indicate that the EnKF, with fewer than 100 ensemble members, performs very well in this experimental context.