The main idea of this manuscript consists on applying signal processing methods along with statistical approaches to achieve the characterization of a complex and hostile environment such as the oceanic underwater environment. In order to study, characterize (i.e. the measurement of the pressure, the salinity, the temperature, etc.) and extract the topographical features (depth, etc), Ocean acoustic tomography methods are widely used.For many reasons, Passive ocean Acoustic Tomography (PAT) configuration becomes more attractive than the classic active one. However the new passive configuration is more challenging. In our project, some of the facing problems have been addressed. In this manuscript, major problems are described and a new separation scheme based on modified ICA algorithms is proposed. Finally, our experiment results corroborate the effectiveness of our proposed approach.KEY WORD: Underwater acoustic applications, passive acoustic tomography, second order statistics in frequency domain, multipath channel, sparseness or non-stationary signals.