High-sensitivity, low-power meteorological radar achieved through coherent signal processing

We describe a recently-developed S-band, scanning, polarimetric-Doppler weather radar which makes extensive use of coherent signal processing algorithms in order to achieve high sensitivity despite its relatively small antenna size (3 m diameter) and modest transmitter power (3 kW peak). The radar, which measures Z, LDR, mean-Doppler velocity and spectral-width, is designed for long-term, statistical observations of rainfall spatial structure and bright-band effects at St. Andrews in Scotland. These data will augment a study of precipitation-induced microwave and millimetre-wavelength attenuation on various paths within a dense network of communications links spanning the coverage area of the radar.

Because the radar will be remotely-operated, a highly reliable system is desired, so as to minimise the frequency of maintenance visits. Traditional, high-powered magnetron transmitters were considered insufficiently reliable for this application. Therefore, a high-reliability, low peak-power, travelling wave tube amplifier (TWTA)-based transmitter was selected instead. The fully-coherent, TWTA-based radar to be described compensates for its low transmitted power by employing FFT-based coherent integration techniques, so as to attain high sensitivity. We consider this approach to be preferable to the alternative technique of pulse-compression waveforms, as it avoids artifacts in the data due to range-sidelobes.

The relatively long de-correlation time of a typical rain target is exploited by operating the radar at the highest PRF consistent with the desired unambiguous range interval. This provides a large number, n, of correlated samples for coherent integration, resulting in an n-fold improvement in signal-to-noise ratio (SNR) at the output of the signal processor. Additional signal processing, comprising over-sampling in the range-domain and incoherent integration of the power in corresponding velocity bins of multiple spectral estimates, is employed to reach an overall improvement factor approaching 30 dB relative to the single-pulse SNR. Furthermore, the FFT-based processing scheme lends itself to the implementation of a simple, adaptive ground-clutter rejection algorithm.

In this paper, we describe in detail both the radar hardware and the signal processing algorithms. The extra sensitivity obtained by exploiting the signal phase information is clearly demonstrated by comparing the output of our signal processor with a simple, incoherently-integrated log. video channel. The new radar has been calibrated by performing simultaneous scans with the co-sited Chilbolton CAMRa radar. We also present the results of these calibration experiments, which demonstrate good agreement between estimates of Z, LDR and velocity made with these independent systems.