Estimation of average under‐ice reflection amplitudes and phases using matched‐field processing

Average low‐frequency under‐ice reflection amplitudes and phases in the central Arctic were estimated from long‐range (260‐km) signals from fixed continuous wave (cw) sources detected on a long (1‐km) vertical array (the FRAM IV experiment), using conventional and minimum variance estimators in conjunction with matched‐field processing. Theoretical computation of the trial source fields for all assumed ranges and depths was based on a normal‐mode code into which were incorporated the under‐ice reflection amplitudes and phases by functional dependence on grazing angle. These were varied iteratively to achieve maximum matched‐field processor gain with minimum range and depth errors. The resultant best data‐fitting amplitudes and phases, which differ significantly from expectations based on existing under‐ice scattering models, provide a new benchmark for testing and, it is hoped, a stimulus for the development of refined theories.