Sensitivity of model-based signal processing to parameter uncertainties in normal modes estimation

The sensitivity of model-based signal processing to various parameter uncertainties in normal modes estimation problem was examined. A low-frequency source was considered whose field was generated by a normal mode model. Based on a state-space representation of the normal mode propagation model and a vertical array measurement system, the extended Kalman filter (EKF) was used to estimate parameters of the normal modes for the purpose of shallow ocean environment identification. The EKF was sensitive to the initial values of the state vector and easy to diverge when the modeling of the ocean environment was not so accurate. The effects on the processor performance of different factors, such as, initial values of the state vector, sound speed profile (SSP) uncertainty and array configuration, were studied in detail. Simulations under a typical shallow water environment were performed, presenting some intuitive results and conclusions.