Abstract: In deep-sea areas at mid-latitudes with depths exceeding 4000 m, a significant acoustic shadow zone exists when both the sound source and receiver are located within 200 m of the surface. Within a range of several tens of kilometers, this region exhibits substantial propagation loss, covering 80%–90% of the horizontal spatial extent. Detecting surface vessels is critical for underwater platforms. Conventional sonar systems have limited processing gain and struggle to distinguish between surface and underwater targets. Therefore, achieving detection and three-dimensional passive localization of ship targets within the acoustic shadow zone is of paramount importance. This study employs a signal processing method analogous to passive synthetic aperture sonar to enhance the output signal-to-noise ratio and azimuth resolution. Additionally, based on the time-frequency interference structure, field coherent matching localization method is proposed and deep learning techniques are utilized for search to estimate target range and depth. Finally, the bearing error in the deep ocean shadow zone has been corrected based on target range and depth. The effectiveness of the proposed method was validated through an experiment conducted in the deep sea.