Spatio-temporal structures of sound fields in Lake Ladoga were measured using an extended vertical receiving array. Results of processing of these data are presented. The processing was aimed at extracting such field components which are stable with respect to unavoidable inaccuracy of the environmental model. In terms of a ray-based approach the stable compo-nent is formed by a narrow beam of rays. In terms of the normal mode approach it is formed by a small group of modes. In the experiment under consideration, wave fields excited by a point source emitting wideband sound pulses and fields of wave beams excited at fixed fre-quencies using a vertical emitting array were analyzed. Comparison of results of the data processing and computed wave fields shows that, as expected, the stable components can be predicted more accurately that the total sound field.
$^1$Federal research center Institute of Applied Physics