POWERS generates the periodogram power spectrum estimate from the 1D or 2D fourier transform of a real image. The periodogram estimate is just a way to normalize the power spectrum such that Parseval's theorem holds (i.e. the sum of all the power terms equals the mean squared amplitude of the transform).
POWERS only operates on fourier transforms generated with VISTA command FFT. If you want to generate the power spectrum from a more general complex image (p.e. an image synthesized with command CMPLX) which is supposed to be the fourier transform of a real image, use command CABS to get the modulus of the transform and multiply it by itself (with MULT not CMULT since the modulus of a complex image is real). The periodogram estimate and the modulus-squared are basically identical (they may only differ in the 0th and Nyquist frequencies).