Abstract: The problem of estimating adaptively and tracking two-dimensional direction of arrival(DOA) of coherent sources with a sparsely spaced electromagnetic vector-sensor array is studied.A new pre-processing method called polarization difference smoothing algorithm(PSDA) is firstly proposed to decorrelate sources'coherency,and is coupled with the propagator to estimate the coherent sources'DOA without eigenvalue decomposition into the signal and noise subspaces.The proposed PSDA-propagator can outperform its counterparts based on polarization smoothing algorithm(PSA) in a spatially nonuniform noise environment.In order to estimate adaptively DOA or track the time-varying DOA promptly,the leastmean -square(LMS) or normalized LMS(NLMS) algorithm is combined to estimate the instantaneous propagator,and the approximate Newton method is used to update the azimuths/elevations.Simulation results show that the proposed algorithm has good adaptive estimation and tracking abilities.