| Autor: |
Hartstra, I.E., Salomons, E.M., Colin, M.E.G.D., Prior, M.K. |
| Jazyk: |
angličtina |
| Rok vydání: |
2019 |
| Předmět: |
|
| Zdroj: |
UACE 2019 in Hersonissos, Crete, 30 june-5 july, 2019 |
| Popis: |
Passive-sonar-performance modelling requires the accurate estimation of the acoustic field for a broad range of frequencies in various environments. In this paper, we assess the suitability of a Gaussian Beam (Bellhop) and Parabolic Equation model (RAM), whereby we focus primarily on computational efficiency. For Parabolic Equation (PE) models the computation time increases with the number of grid-points required to provide accurate results. Since the grid-size scales to the wavelength, the computation time generally increases with frequency. The computation time of Gaussian Beam models increases primarily with longer maximum range. Since lower frequencies result in relevant acoustic intensities at larger distances, this results in a computation time increase in case of lower frequencies. Our aim is to estimate a single-run ‘transition frequency’ that determines which model is optimal in terms of computational costs, while ensuring that each model is subjected to a convergence criterion we set at 1 dB. We only consider single-frequency runs in this paper. The transition frequencies are estimated for three different scenarios that vary in depth and sound speed profile (SSP). We find that RAM is the computationally efficient choice up to a transition frequency of about 11 kHz and 12 kHz in the case of the shallow water scenarios: Weston Case 1 and 4, respectively. For the deep water scenario, a Munk profile truncated at 1 km depth, RAM is found to be the most efficient choice up to about 4.5 kHz. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
