Chaotic behaviour and non-linear prediction of airborne radar sea clutter data

The potential to model sea clutter radar returns using chaos theory is examined. Chaotic systems display qualitative similarities to sea clutter returns such as broad flat spectra, boundedness and irregular temporal behaviour. In this report several key parameters of chaotic systems, namely correlation dimension, Lyapunov spectrum and Lyapunov dimension are calculated from real sea clutter returns and found to be consistent with a chaotic interpretation. The airborne high resolution data (less than one metre) produces a correlation coefficient with an average value of 4.63 and an embedding dimension of 6-7. Lyapunov dimensions are consistent with correlation values. A local linear technique and a radial basis function (RBF) are used to construct a one step non-linear predictor. A mean square error (MSE) of approximately 0.0032 between the predicted and normalized (i.e. maximum +/-1 range) real time series is measured.