A reduced-dimension Propagation Method(PM) based algorithm is proposed for central Direction-Of-Arrival(DOA) estimation of coherent distributed sources. Based on the Generalized Array Manifold(GAM) model, the central DOA can be decoupled from the original array manifold, achieving separation from the angular spread. To avoid eigenvalue decomposition of the sample covariance matrix, a propagation operator matrix is computed in the orthogonal space of the generalized array steering vectors to construct the objective function. With the introduced dimensionality reduction technique, the proposed algorithm only requires a one-dimensional spectral peak search to determine the central DOA of the sources, which significantly reduces the computational complexity. Additionally, the Cramér-Rao lower Bound(CRB) for this scenario is derived in detail to provide a benchmark for the estimation performance of the algorithm. Performance analysis and numerical simulation results demonstrate that the proposed algorithm maintains excellent angle estimation performance while reducing complexity.