Modelling sound propagation under ice using the Ocean Acoustics Library's Acoustic Toolbox

Acoustic propagation in the Arctic and Antarctic is largely characterised by the presence of a highly variable ice canopy. To model sound in these environments requires both a way of effectively representing the ice layer and modelling its effect on signal transmission. The Ocean Acoustics Library has a powerful open source Acoustics Toolbox that contains Fortran code for running Ray, Normal Mode, and Wavenumber Integration models. There are two parts to modelling a sea ice environment: modelling the ice as an elastic acoustic medium, and modelling the roughness of the ridging characteristics of the ice. This work considers the scenario of an Autonomous Underwater Vehicle (AUV) producing a survey under ridged sea ice. This specifies a range of interest of 10km and a frequency band of interest of 3kHz-13kHz. An overview of methods for modelling ice as an acoustic medium and as a ridged surface is provided, and the applicability of different propagation and ice models for this scenario is discussed. The scenario is then implemented as a specific test case for two example ice canopy profiles. The ice canopy profiles used are sea ice draft measurements recorded in the Arctic using an upward looking SONAR on a nuclear submarine. Beam and ray methods are the only computationally fast propagation codes for this frequency range and are included in the BELLHOP module of the Acoustics Toolbox. With these methods the options for including the elastic properties of the ice are limited and only include reduction in the coherent field on reflection. Two methods for including the ridging of the ice canopy are implemented, one statistically based and one using direct input of measured ice canopy data. The statistically based method uses Twersky boss scattering, and the direct method inputs the draft data as an altimetry file. Gaussian beam tracing using BELLHOP is run to generate ray trace and coherent transmission loss estimates of this environment. The advantages and limitations of these implementations are discussed with suggestions for future improvements to the Acoustics Toolbox to better model the ice scenarios outlined. The improvements identified from this review and test case are: the capability to include specific ice condition data where available, better consideration of the elastic properties of the ice in BELLHOP; and new statistical methods for modelling unknown variable surface boundaries that provide statistical distribution information as well as mean field values.