and the dynamics of the real atmosphere contributes to the error in the

forecasts. When the model is employed to provide the background for an analysis scheme, forecast errors often lead to a slowly evolving systematic error component in the background. This type of error, which is called model bias, violates the common assumption of analysis schemes that the mean of the probability distribution of the background error is zero. Likewise, systematic errors in the observations and the observation operator violate the assumption that the observation operator maps the hypothetical true state of the model to a random variable that differs from the observations by a zero-mean random noise term. This paper evaluates a strategy to account for the background and the observation bias in the Local Ensemble Transform Kalman Filter (LETKF). This strategy is evaluated in both idealized and realistic settings. Numerical experiments are carried out with the simplified primitive equation model called SPEEDY and a reduced resolution version of the NCEP GFS.