Results of the horizontal slice inversion at 150m and 200m depth are presented in Fig. 3, and the inversely estimated temperature shows similar patterns in terms of the location of the warm eddy observed by AXBT.
To obtain inverse solution of the vertical slice, the water column is divided into 8 layers: one layer from the surface to 150m depth, five 50m-thick layers between 150m and 400m depth, and two layers from 400m to the bottom. Results of the vertical slice inversion are presented in Fig. 4 where the symbols “+” represent the inverse est imation and the solid line represents the observed sound speed which has been range-averaged over sources and receiver positions. The sound speeds reconstructed by inversion are higher than the observed one in the upper layers above 250m depth. The maximum error of the inversion is about 10m/s (2.1 ﾟC in temperature) in the second layer where the ray traveled least and it created an underdetermined case of fewer equations than unknowns in the inverse problem. However, the vertical gradient of the so und velocity in the upper layers is in good agreement with the observed one.
In stead of measurements of the travel time, the sound pressure levels of different frequency components were monitored azimuthally for the circular geometry in '98-ex periment. Fig. 5 shows the azimuthal distributions of the pressure level received at the center of the two circles. It clearly shows that the pressure level amomaly exists along the direction where the warm eddy was appeared in the observation.
