In the stochastic method of estimation, the vertical structure of the ocean is modeled through the quasi-geostrophic theory while the horizontal structure is assumed to have a gaussian covariance with a length scale of 100 km. A good agreement has been observed between the assumed and the reconstructed sound speed perturbation field. Acoustic intensity losses due to spreading and absorption due to the presence of boric acid and magnesium sulfate etc. have been estimated. For a propagation range of 300 km, computed intensity loss due to spreading for different eigen rays varies between 92 and 116 dB while that due to chemical absorption is about 4.25 dB at 0.2 kHz and 15.25 dB at 0.4 kHz Source frequencies. As the Arabian Sea undergoes seasonal changes in accordance with the changes in the atmospheric forcing, the tomographic reconstruction is applied to model these changes through numerical experiments. Using the annual mean and winter mean sound speed profiles, eigen rays (17) have been identified for inversion analysis. For the reference state, the data kernel (consisting of ray path lengths in a pre-set number of layers) has been constructed. The acoustic rays scan the vertical water column between 180 m (upper limit) and 3300 m (lower limit). The data kernel was used to construct the generalized inverse operator to operate on travel time perturbations. The predicted travel times of eigen rays computed using the mean sound speed profile for the reference state and the assumed profile (winter season mean) have been used to generate possible perturbations in travel time. These travel time deviations ranged from 0.7 to 117 ms. These were operated by the generalized inverse operator to get the model parameter perturbations in different layers. Similarly, the sound speed profiles associated with a subsurface cold core eddy, observed during southwest monsoon period in the northwest Bay of Bengal were used in model simulations and profile reconstruction. The effect of the eddy has been found to be in lowering the ambient sound speed by about 10 m/s. Under its influence, the depth of the SOFAR channel axis remains constant at 〜1600 m, which otherwise should have shown a deepening in this region. Simulated ray arrival structure depicts the typical characteristics of a weak acoustic wave guide in the Bay of Bengal - the early arrival cluster of near axial flat angle rays, Iater arrivals of deep turning rays, followed by the latest arrivals of near-surface turning rays. The arrivals of the acoustic rays are delayed by about 100-200 ms under the influence of the eddy. The intensity computations show that when a ray passes through the eddy, it suffers an additional loss of 20-25 dB . From the simulated travel time delays, the eddy profile is reconstructed through a generalized inversion, based on the singular value decomposition technique, the numerical experiment shows that 18 eigen rays with 9 Iayers enable reconstruction of the eddy profile adequately using 9 eigen modes. Among the various methods available to describe propagation of acoustic energy in the ocean, parabolic equation method has been proven to be one powerful and effective means for estimation of sound field. This is so for (a) range dependent environment, (b) variable bottom topography, (c) specific input source field patterns and (d) source frequency spectra. Using PE method, two cases were analyzed assuming a source depth, frequency and propagation ranges (a) 150 m, 400 Hz and 650 km and (b) 1000 m, 70 Hz and 5000 km for simulation of propagation involving a medium range in the Bay of Bengal and (2) a long range along Perth (Australia) -Madras transect (case 2).
