Cluster of sound speed fields by an integral measure

Abstract

A technique to cluster large area oceanic predictions of sound speed into quasi-range independent areas is presented. Oceanic models produce high fidelity predictions of the oceanic sound speed fields that enable large-scale simulation of acoustic propagation with reasonable accuracy. Unfortunately, the oceanic models can produce sound speed fields quicker than can be digested by current technologies in underwater acoustic performance prediction systems. The speed bottleneck can be broken in two ways, a long term improvement in prediction technologies, and a interim process that allows similar acoustic areas to be aggregated into range dependent regions while maintaining a high degree of fidelity with the performance prediction resulting from using the complete oceanic model output. The interim process created must be capable of reflecting changes in sound speed field that control water born energy, and the changes in the field that effect the interaction with the oceanic bottom. This study uses vertically integrated gradient of the sound speed field as a basis for creating quasi-range dependent areas. The integrated gradient, when applied over a restricted latitudinal extent, gives an estimate of the mean ray curvature in the wave-guide. Since the field is integrated over depth, the effects of water depth are included in the calculation. The nature of the interaction with the oceanic bottom is not included in this calculation. The variations in bottom loss over the region of interest will be integrated into the analysis at a latter stage. The quantity cannot be used to predict transmission loss, but indicates where similar propagation conditions occur. The method is shown to be sensitive to the characteristics of the deep sound channel, and changes in the near surface structure of the sound speed field. The method is sensitive to the vertical integration weighting, which needs to be investigated further. The method is applied to an oceanic region where variations in bottom effects are simple and relatively well understood. Incoherent acoustic transmission loss predictions are made from a single receiver depth to a single target depth for each of the longitude-latitude mesh points in the computational domain of the oceanic prediction model where the initial water depth is greater than both the receiver depth and the target depth to a maximum range of 100km. The transmission loss predictions are fitted to a two-parameter family of curves. The two-parameter family of curves accomplish two closely related goals: (1) the number of parameters that must be compared is minimized, and (2) the slight differences in range of maximum and minimum that so often cloud prediction comparisons are eliminated. The relationships between transmission loss parameters, and integrated gradient of the sound speed field are relatively well constrained. The relationships cluster into groups that are characterized by bottom type. The relationships between the two-parameter family of curves and the vertically integrated index are thus far diagnostic, the variations in parameter coefficient are large enough such that a prognostic relationship does not appear to be supported.