Baseline and distribution estimates of complicated spectra

In processes with many lines, such as those encountered in climate, space physics, and communications systems, it is often found that the underlying base spectrum(1) also has a complicated shape. In this paper I discuss some methods to estimate the base, and detail spectra by using the probability distributions of such spectra. The methods include a robust estimate of the central part of a mixture of central and noncentral chi-square distributions and, second, a direct estimate of the mixing and noncentrality parameters made by minimizing the Kolmogorov-Smirnov D statistic between the empirical and theoretical cumulative distribution functions.