Experimental design for geoacoustic parameter inversion

• Published in: The Journal of the Acoustical Society of America Vol. 109; no. 5_Supplement; p. 2423
• Main Authors: Tolsma, Mischa F. P., van den Bos, Adriaan, Blacquière, Gerrit
• Format: Journal Article
• Language: English
• Published: 01-05-2001
• DOI: 10.1121/1.4744579
• EISSN: 1520-8524
• ISSN: 0001-4966

Both model-based and nonparametric classification benefit from a theoretical investigation into the information contained in reflections from the subbottom. Specifically, it is difficult to find out if the properties of a certain bottom type are identifiable with a given acoustical signal due to the complexity of the reflection models used. This problem is approached with the Fisher information matrix of the stochastic model for the measured acoustic response. The problem is unidentifiable when this matrix becomes singular. Its inverse, the Cramér–Rao lower bound, gives the best possible precision of the parameter inversion. This lower bound can be used to design appropriate signals for precise measurement of the subbottom. This has been done for a model of a surface layer with varying density given by Lyons and Orsi [IEEE J. Oceanic Eng. 23, 411–422 (1998)]. It is found that precise measurement of the reported range of model parameters requires a signal with a frequency range of 1–50 kHz. This supports the finding that broadband or multifrequency techniques are needed for the classification of layers with continually varying properties. [Work supported by TNO TPD and Rijkswaterstaat, The Netherlands.]