The analysis of multichannel airborne gamma-ray spectra

Abstract

The conventional processing of airborne gamma-ray spectrometric data uses 3 broad energy windows to estimate the ground concentrations of K U and Th. This thesis investigates the potential for using the full gamma-ray spectrum in an attempt to increase the amount of information currently extracted from airborne gamma-ray data. The observed spectrum is considered as the sum of 3 terrestrial and 3 background components. Given the shapes of the component spectra, the airborne gamma-ray spectrometric inverse problem is to determine the relative contributions of the components to the observed spectrum. The component spectra are determined through suitable airborne and ground calibrations. The limitations of the component spectra have necessitated a model-based approach to multichannel fitting. The components are fit to real data, and only those energies over which a good fit is achieved are used for multichannel processing. A parametric model based on a principal component analysis of the terrestrial component spectra as functions of simulated detector height is used to find the K, U and Th terrestrial component spectra that best fit the background-corrected airborne data. The simulated heights are mapped onto actual heights using airborne calibrations over a calibration range. This enables the terrestrial component spectra to be used for the calibration of multichannel background estimation methods. The component spectra are then fit to the background-corrected observed spectra to obtain elemental count rates. This strategy ensures the best possible fit between model and data, and minimizes the propagation of statistical errors in the observations into the estimates of the elemental count rates. The analysis of multichannel spectra using this model produces 3 new parameters - the effective height of the detector above K, U and Th sources. These effective heights may be useful for regolith mapping and for refining the data processing procedures. The multichannel processing results in significant reductions in the fractional errors associated with the estimated elemental count rates. For 3 surveys processed using the new methodology, the average deviations of the K, U and Th elemental count rates from the estimated mean elemental count rates at each observation point are reduced by 12.4%, 26.5% and 20.3%, respectively, when compared to the conventional 3-channel method. This results in a better structural resolution of small anomalies in enhanced images of the processed data. Show more