Robert S. Pritchard
IceCasting, Inc.
http://www.icecasting.com
Tel/fax: (415) 454-9899
Ambient noise models should accumulate sound received from different mechanisms and locations (both nearby and distant), and account for propagation loss [Pritchard, 1984, 1990, 1993]. The acoustic sources are estimated from velocity, deformation, stress, energy dissipation, and other fields that result from simulations and forecasts of sea ice dynamics.
The semi-empirical ambient noise model presented provides a framework. It allows for future improvements in propagation laws and source signatures. This model is based on the basic physical principle that sound intensities from multiple uncorrrelated sources accumulates linearly.
What is the algorithm?
Sound received at a hydrophone at location x and depth z is the sum of intensities
where the integral is performed over the horizontal domain of the ice,
Is is the intensity
of each source emanating from location y and depth 
, and T is the fraction of
source intensity that is received and therefore is the complement of the propagation loss,
Pritchard [1984, 1990, 1993] used energy dissipation measures as proxy variables to estimate noise signatures from pressure ridging, shear ridging, mixed layer shearing, and microcracking. An empirical propagation loss law developed by Buck and Wilson, [1984]. The semi-empirical approach means that the signature S(f) for each process must be estimated empirically or by another analytical means.
The pressure ridging source was proportional to power dissipated by stress during plastic stretching
and shear ridging was proportional to
The mixed layer shearing source was proportional to power dissipated by water stress
The microcracking source was proportional to strain energy release rate
New relationships can be developed for the
anisotropic plasticity model. It is essential however
that all key variables be available for the ambient noise model calculations.
The fields must be stored on tape for archiving and for separate processing. We need the
complete fields of velocity, stress, plastic stretching, and oriented thickness distribution.
Each of the above sources is likely to be replaced by improved relationships. additional
processes will be added. For example I am now hoping to develop a ridging model that
incorporates the processes of block sliding and falling.
What data are needed?
Substantially more research is needed to estimate source signatures from different
mechanisms. Furthermore, the semi-empirical model requires that unknown coefficients be
estimated by correlating measured acoustic signals while observing ice motion,
deformation, stress and ice condition. Ice condition data will be required for model
operations.
Is it practical?
Yes, the model is practical! It uses output form the ice dynamics model fields to force the
acoustic signature fields. It is a simple and inexpensive calculation that does not interfere
with the ice dynamics calculations.
Will it make a difference?
The primary argument for using this model in the new PIPS 3.0 is that it is based on
correct physical principles, and it will allow other researchers to add their algorithms. Even
though we have done little testing of the semi-empirical model, its physical correctness
argues strongly for its acceptance.
References
Buck, B. M., and Wilson, J. H. (1984) "Buck/Wilson Deep Water Arctic Transmission Loss Model," PRL TR-58, Polar Research Lab, Carpinteria, CA.
Pritchard, R. S. (1984) "Arctic Ocean Background Noise Caused by Ridging of Sea Ice," J. Acoust. Soc. Am., vol. 75, no. 2, pp. 419-427.
Pritchard, R. S. (1990) "Sea Ice Noise-Generating Processes," J. Acoust. Soc. Am., Vol. 88, No. 6, pp. 2830-2842.
Pritchard, R. S. (1993) "Sea Ice Constitutive Behavior and Under-Ice Noise," Natural Physical Sources of Underwater Sound, (ed.) B. Kerman, Kluwer Academic Press, The Netherlands, pp. 591-610.