What constrains the growth of spread in ensemble Kalman filters?

The spread of an ensemble of weather predictions initialized from ensemble Kalman filters often grow slowly relative to other methods for initializing ensemble predictions. Several possible causes of the slow spread growth are evaluated in perfect- and imperfect-model experiments with a 2-layer primitive equation spectral model of the atmosphere. The causes examined are covariance localization, the additive noise used to stabilize the model, and model error itself. In this model, it is found that the additive noise is the biggest factor in constraining spread growth. A method for making the additive noise more flow dependent through by pre-evolving additive noise perturbations is demonstrated as a possible remedy.