136 A High-SNR Doppler Beam Swing Method for VHF Atmospheric Radar

Tuesday, 29 August 2017
Zurich (Swissotel Chicago)
Koji Nishimura, National Institute of Polar Research, Tachikawa, Japan; and T. Hashimoto, M. Tsutsumi, K. Sato, and T. Sato

The Doppler beam swing technique, in which the antenna beam is altered typically towards 5 directions at every pulse transmission cycle of several 100 micro seconds, is commonly used for wind profiling with 50 MHz band (VHF) atmospheric radars. With this methd an observed signal in respect with each direction is decimated by 5 as a result of cyclic alternation of the beams. An echo from the troposphere and stratosphere typically has a correlation time of about 1 second for 50 MHz band radar and, in principle, data samples within a correlation time contributes for gaining the SNR. This means, we are able to obtain a signal that has the SNR of 5 times as much if we observe a single direction at a time for 1 second without decimation. For observing all directions, the antenna beam should be altered at every 1 second, on condition that the "simultaniousness" of the observation in respect with the 5 directions can be relaxed to the order of seconds. At the end, an observed signal with this method thus has 12 burst sampling windows with a length of 1 second each within a sequence of observation of 60 seconds, for example. We then need to take care of the order of the alternation of beams. In case that a regular scanning order from 1 through 5 is applied, the resulting spectra have insensible regions in the frequency domain. In order to avoid the insensible regions to appear in the frequency domain, a random-like scanning order should be applied to alter the beam directions. Thus obtained signal spectra are subjected to convolutional spectrum broadening due to the 1-second window. Considered the effect, there are some choices in spectral parameter estimation methods such as the weighted correlation method, parametric inversion, and sparse spectrum estimation. While our proposed technique has a positive impact on the SNR of the signal obtained, it has a slight negative impact on the accuracy of the parameter estimation due to the spectral broadening. Hence, the technique is supposed to be applied to the targets of very low SNR. In this talk, we present the concept of this whole scheme and quantitative evaluation on the SNR and accuracy of estimation, with some preliminary results from demonstrative observations.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner