In this paper, the bearing, elevation, and range estimation of near‐field wide‐band sources from data observed across an array of sensors is considered. Coherent signal‐subspace processing for bearing estimation is tailored for localization of wide‐band sources located in the near field. For sources located in multiple groups (clusters), a matrix beamformer for dimensionality reduction is utilized to simplify wideband processing. The matrix beamformer is used to maximize the average signal‐to‐noise ratio over cells where sources are likely located. A version of coherent signal subspace processing in cell space, which combines beamforming and focusing, are developed and comparative performances are carried out by means of Monte Carlo simulations.