In order to equip unmanned underwater vehicles with the capacity to passively localize long-range sources in shallow water, a localization approach which combines the flank array of unmanned underwater vehicle and matched-field processing was proposed. The proposed approach can be used as an unmanned underwater vehicle cruising at a low speed along a straight line at a certain depth and gathering data at sampling positions. Matched-field processing data received by the flank array at each sampling position provide the primary localization capacity. Furthermore, combining and analyzing already generated matched-field localization outputs can improve the localization performance against the lack of spatial gain due to the limited physical size of the flank array. It enhances the robustness to ocean environmental mismatch and position errors of the flank array, the stability in the estimation process, and the output signal-to-noise ratio. The three-dimensional coordinates of the source of interest can be gradually estimated over time. The ability of the proposed approach is examined through the synthetic test cases at different signal-to-noise ratios, and satisfactory results can be obtained in all cases. Additionally, the localization approach is proper for practical situations because the system requirement is not high.