We investigate the robustness of the linear matched filter (MF) operating i
n a Gaussian environment in the presence of a mismatch between the design c
lutter-power spectral density (PSD) shape and the actual one. The Gaussian,
the power-law (PL), and the double-exponential spectral models have been c
onsidered with the goal of investigating which one fits best for windblown
foliage, We analyze the MF performance in terms of improvement factor, prob
ability of false alarm, and probability of detection by making use of the t
heoretical models and measured X-band ground clutter data. The numerical re
sults validate the double-exponential spectral model for windblown foliage
by showing that the differences in performance prediction between using mea
sured clutter data and modeled clutter data of various spectral shapes (viz
,, Gaussian, FL, and double-exponential) are minimized when the spectral mo
del employed is of double-exponential shape.