The potential of the ensemble Kalman filter for NWP - a comparison with 4D-Var

The ensemble Kalman filter (EnKF) is reviewed for its expected assimilation characteristics and ease of implementation, and compared to the currently more popular four-dimensional variational assimilation (4D-Var). The EnKF is attractive when building a new medium-range ensemble numerical weather prediction (NWP) system. However it is less suitable for NWP systems with uncertainty in a wide range of scales; it may not use hieh-resolution satellite data as effectively as 4D-Var. For limited-area mesoscale NWP systems a hybrid method is attractive.