The first maritime surveillance radars in World War II quickly discovered that returns from the sea, soon to be known as sea clutter, were often the limiting factor when attempting to detect small targets while controlling false alarms. This remains true for modern radars, where the detection of small, slow moving targets on a rough sea surface remains one of the main drivers for maritime radar design, particularly in the development of detection processing. The design, development and testing of radar signal processing for maritime surveillance requires a very detailed understanding of the characteristics of radar sea clutter and of the combined target and clutter returns. This book provides an updated and comprehensive review of the latest research into radar sea clutter and detection methods for targets in sea clutter. The emphasis is on understanding the characteristics of radar sea clutter as observed with different radars, viewing geometries and environmental conditions. This understanding is assisted by the development of mathematical models that are used in the radar design process. In recent years there has been an increased interest in operating at higher altitudes, resulting in the sea surface being illuminated with larger grazing angles than used in traditional airborne surveillance platforms or ground-based systems. There has also been significant research into bistatic operation, including passive radars using illuminators of opportunity. The use of coherent and multi-aperture systems in maritime radar are also of increasing interest and these new application areas are also covered in this book.
Inspec keywords: statistical analysis; Doppler radar; radar clutter; marine radar; radar detection; radar signal processing
Other keywords: Doppler radar; statistical analysis; backscatter; airborne radar; radar detection; radar signal processing; mathematical analysis; object detection; target detection; radar imaging; radar sea clutter
Subjects: General electrical engineering topics; Radar and radiowave systems (military and defence); Other topics in statistics; Signal detection; Radar equipment, systems and applications; Mathematical analysis
In this paper, we reported the requirements of airborne maritime reconnaissance, although many of the ideas and models can be equally applied to ground-based or ship-borne radars.
This chapter reviews the main characteristics of monostatic sea clutter and their representation in mathematical models. The main characteristics of the backscattered returns which need to be considered to assess radar performance include the normalised mean backscatter; the amplitude distribution; the Doppler spectrum; the spatial and temporal variation and clutter spikes.
This chapter has reviewed the available measurements of bistatic sea clutter and shown how their statistical characteristics can be represented in models for performance prediction and radar design purposes. This knowledge can now contribute to the wider task of designing bistatic radar systems.
This chapter presents a number of parametric models that relate to the statistical models presented in Chapters 2 and 3. The practical use of these statistical models requires a way to relate the model parameters to the sea conditions, collection geometry, polarisation and radar parameters, including spatial resolution and frequency. This can be done either empirically, based on measured data, or with electromagnetic modelling using physical models of the sea surface. However, electromagnetic modelling approaches are usually too computationally intensive to be used for most applications.
In this chapter, the simulation of sea clutter is restricted to representations using compound Gaussian models. As discussed in Chapter 2, compound models have a speckle component with an exponential distribution for intensity, and a texture component that is approximately constant over the tens of milliseconds typically associated with a radar dwell. When using pulse-to-pulse frequency agility with sufficiently large frequency steps, the speckle fluctuation will be approximately random and for a fixed frequency radar, will fluctuate according to a short-term temporal correlation, usually modelled by a Doppler spectrum. The texture will also have correlations that influence both the spatial and longer term temporal fluctuations of the sea clutter. Section 5.2 describes techniques for simulating coherent clutter samples with different statistical distributions, correlated texture and a uniform Doppler spectrum (i.e. constant centre frequency and width).
This chapter describes how sea clutter models are used in the prediction of radar detection performance. For these results to have value, the radar analyst must have confidence in their accuracy. If an analytic modelling approach is taken, considerable attention to detail is required to understand and model the radar operation over a wide range of environmental conditions and viewing geometries. These include the grazing angle, the wind speed and direction and the sea swell and direction. The models must also be able to accurately represent the effects of the radar waveforms and processing on the received signals.
This chapter describes some of the standard techniques that may be used to detect returns from targets against a background of sea clutter and added thermal noise. The goal of the radar signal processing is to maximise the chance of detecting a target, while controlling false alarms from clutter and noise at an acceptable level. The ideal outcome is a constant false alarm rate (CFAR) and the challenge is to achieve CFAR performance when the clutter characteristics are unknown 'a priori' and varying considerably over the region of sea being searched by the radar.
The work presented here has been extended in a number of further publications. The first looks at the case of multiple targets with the multiple target Bernoulli TBD algorithm developed for both closely and well separated targets. In both scenarios, the extended TBD filter shows good performance, characterised by a small number of false tracks and successful detection and tracking of targets with SIRs of 5 dB or more.