By considering the theoretical expression for plane wave reflection, the reflection loss, as a function of grazing angle and frequency, is strongly dependent on the ice thickness, roughness and the elastic properties. In order to obtain information about the internal properties and ice thickness, acoustic measurements have to be made at frequencies or in frequency bands which is sensible to the sea ice, not at low frequencies causing total specular reflection. According to simulations in Fig. 6, not including roughness, the optimum frequencies for retrieving ice information are between 100-3000Hz. This is well above the maximum lobe of ambient noise centered around 15-25Hz in the Arctic. In order to obtain optimal ice information two approaches has to be considered 1) intensity measurements in a broad frequency band to get information about the filtering processes 2) travel time/phase measurements using frequencies which are sensible to the actual range of ice thickness to be measured. The travel time approach has previously been found to be promising for sources using center frequencies at a couple of hundred Hertz (Jin et al., 1993, 1994).
From the above results it is obvious that there is a conflict between the optimum frequency for retrieving information about averaged ocean temperature and the optimum frequency for retrieving information about the ice cover, and also for the source location. Therefore two monitoring configurations have to be considered in the forthcoming work; one for averaged temperature and one for averaged ice conditions.
In the case of monitoring the effect of changing averaged ocean temperature narrow band sources centered at low frequency (well below 100Hz) will be used in order to avoid the strong reflection and scattering losses caused by the sea ice cover. While, in the case of monitoring changes in the sea ice cover we will consider broadband (100-3000Hz) sources positioned within the surface duct in order to trap as much as possible of the acoustic energy in the surface duct so that the signal/noise ratio is satisfactory to long enough ranges to give a reasonable estimate of averaged ice thickness.
Task 4. Acoustic modeling of the Fram Strait
A promising method of permanently monitoring temperature and current velocity of the Atlantic water penetrating into the Arctic Ocean through the Fram Strait is acoustic signal travel time measurement at the cross-section of the strait, which has a typical width of 300km and depth of 2700m. The method is based on the fact that the propagation time of acoustic pulses along the paths connecting a source and a receiver (eigen rays) is determined primarily by the distributions of temperature and longitudinal stream velocity components. This allows the acquisition of appropriate data from acoustic measurements in an acoustic tomography framework (Naugolnykh et al., 1998a).
In the sensitivity study of acoustic propagation to temperature changes, different methods of the inverse problem are used. One of them is based on selecting the stable rays and determining the travel time variation. The second one consists of defining the trend in the "signal-arrival" spectrum caused by the temperature changes, and computing average parameters of the arriving signal ensemble - the "cumulative sum". The third is based on the "collective arrival time" approximation.
The presence of the stable rays was established and the temperature influence on the signal propagation along the stable rays was evaluated as a gradient of 29 ms/℃ for Ray 4s and 37 ms/℃ for Ray 5s. The arrival time fluctuations were investigated for changing oceanographic data. Environmental data variation was provided by shifting the entire sound-speed pattern horizontally with respect to the transmitter-bottom-receiver configuration for a distance essentially more than the scale of ocean turbulence spatial correlation.
