Bias correction of an ocean-atmosphere coupled model

A serious problem in the initialization of a climate forecast model is the model-data incompatibility caused by systematic model biases. Here we use the Lamont model to demonstrate that these biases can be effectively reduced with a simple statistical correction, and the bias-corrected model can have a more realistic internal variability as well as an improved forecast performance. The results reported here should be of practical use to other ocean-atmosphere coupled models for climate prediction.