Target detection in the presence of sea clutter and ionospheric interference using joint spectral/time-series analysis and autocovariance procedures

Joint spectral-temporal methods for target detection are not uncommon, but generally involve application of spectral processing on short sub-sections of the original time series. Usually a sliding window is applied, so that the region of truncation slides through the superset. Final analysis and interpretation are usually performed in the spectral domain. Here a new method of temporal-spectral analysis is presented. Truncation is used first in the spectral domain, and then multiple sets of resulting time series are employed for identification purposes. The method is called the ETISTI method (-enhanced truncated interleaved spectral-temporal interferometery-), and is adapted and extended from earlier meteor studies. It uses a combination of band-pass spectral filtering, auto-correlative algorithms and time-series analysis for target detection. It especially utilises long data sets of the order of hundreds to thousands of seconds, in order to improve spectral resolution, but at the same time achieves temporal resolution of the order of seconds. Signal-to-noise levels are determined locally rather than globally, using dynamic auotocovariance methods, thereby allowing adaptive time- and range-dependent noise-level determination, and hence better target discrimination. The method works especially well for accelerating targets, and for targets obscured by ionospheric interference, lightning and intermittent RF noise.