| The classical theory of detection using the
Neyman-Pearson principle is applied to stratosphere-troposphere (ST) radar
signals. It is extended to provide information regarding the detection of weak
signals which complements the detectability method usually employed in ST radar
studies. It is shown that for ST radar signals of low amplitude and a
detectability around 3 (a value commonly invoked in literature), the probability
of detection is about equal to the probability of false alarm. The question of
threshold detectability is also discussed.
Spectral moments errors are evaluated by a method which is an
extension of the analytical method of estimation developed by Miller and
Rochwarger and the results compared to other statistical and analytical models.
As already known, three factors can affect the error on the estimated
parameters: the signal-to-noise ratio, the spectral width and the incoherent
integration number. For high signal-to-noise ratios, analytical results are in
good agreement with Barrick's and Denenberg's theoretical models and with
Yamamoto's statistical one. For low signal-to-noise ratios, the spectral
parameters are more sensitive to the selected model but overall variability is
similar. |