Reducing noise in radiometric multi-channel data using noise-adjusted singular value decomposition (NASVD) and maximum noise fraction (MNF)

Abstract

The airborne gamma-ray spectrometric method is primarily used for the detection of ore bodies, fallout from nuclear accidents and as a lithological mapping tool. Modern gamma-ray spectrometers can record at least 256 channels of data in the energy range 0-3 MeV.\rAfter applying concentional processing, there is still noise in the spectra. Several approaches are suggested to remove this remaining noise fraction (MNF). The methods are applied to raw spectra data prior to integrating counts in the K, U and Th windows and proceeding to conventional processing.\rThe three methods are applied to a survey data set for comparison. All methods are shown to reduce noise in the original data set.