The solution of the current wind retrieval algorithm for scatterometers has several wind vector ambi-guities, due to the bi-harmonic relationship between normalized backscattering cross section and the relative wind direction and the existence of the measurement error. In order to remove the ambiguities for a unique wind field, a circular median filter approach (CMF) is usually adopted. But under the con-dition for clustering distribution of the false ambiguities in some local areas, the CMF fails and thus engenders block ambiguities, which degrade the precision of the retrieved wind field. For such a situation, a technique of identification and removal of the block ambiguities is presented to further optimize the retrieved wind field after CMF. It is demonstrated in experiment that this technique can identify and remove most of the block ambiguities. The solution of the current wind retrieval algorithm for scatterometers has several wind vector ambi-guities, due to the bi-harmonic relationship between normalized backscattering cross section and the relative wind direction and the existence of the measurement error. In order to remove the ambiguities for a unique wind field, a circular median filter approach (CMF) is usually adopted. But under the con-dition for clustering distribution of the false ambiguities in some local areas, the CMF fails and thus engenders block ambiguities, which degrade the precision of the For such a situation, a technique of identification and removal of the block ambiguities is presented to further optimize the retrieved wind field after CMF. It is demonstrated in experiment that this technique can identify and remove most of the block ambiguities.