Two assimilation systems developed using two nearly independent wave models are used to study the influences of various error sources including mode bias on wave data assimilation; a statistical method is explored to make full use of the merits of individual assimilation systems for bias correction, thus improving wave analysis greatly. This study opens a door to further our understanding of physical processes in waves and associated air–sea interactions for improving wave modeling.