The ensemble Kalman filter (EnKF) is a relatively new technique proposed for the data assimilation schemes of atmospheric and oceanic numerical models. The advantage of this approach over conventional data assimilation systems is the fact that it produces a flow-dependent background error covariance matrix in addition to the optimal estimate of the initial state. The EnKF has also the additional advantage of being much less numerically costly than the traditional Kalman filter as it uses the members of an ensemble to estimate the covariance matrix.

This study focuses on two applications of the EnKF to regional and mesoscale meteorological phenomena. One application is to an East Coast snow storm event. The challenge for such a case is the complicated dynamics of the system where convective and baroclinic instabilities both play equally important roles in determining the predictability. For this case, discussion will generally focus on how realistically-initiated background error covariance is handled by the filter, the characteristics of the evolution of the error covariance matrix, and how moist processes determine the overall predictability of this system. Second application of the filter is to a sea breeze event in Houston/Galveston area where it is believed that the high-ozone events of recent years are meteorologically linked to the characteristics of the land/sea breeze circulation prevalent for that geographical location. From a data assimilation point of view, modeling of such thermally-driven circulations imposes challenges that have not been previously addressed in the literature. One such very important challenge is the estimation of the parameters relevant to the parameterization schemes for surface-air interactions. The potential advantage of using the EnKF in a land/sea breeze modeling environment will be demonstrated and preliminary results will be presented. In addition, an evaluation of the EnKF will be presented in the context of its performance under such structurally different mesoscale conditions as winter snow storms and land/sea breeze circulations. Key differences between these phenomena pertaining to mesoscale error growth dynamics will be outlined and consequences of such differences to the application of the EnKF will be discussed.