Spectral Component Analysis for Airborne Gamma-Ray Spectrometry

Introduction

Measurement of the ground concentrations of the naturally occurring radioelements potassium, uranium (series) and thorium (series) by airborne gamma-ray spectrometry (AGS or radiometrics) is a well established technique.

During the past several years there has been considerable interest in the development of techniques which employ the complete 0 to 3,000 kev spectrum for analysis of airborne gamma-ray spectrometer data. This work was aimed at improving the accuracy of radiometrics results over the traditional method, which uses narrow windows over 3 selected photo peaks to extract ground concentrations of potassium uranium and thorium. Until recently, these developments (mostly spectral fitting procedures) proved to be somewhat impractical because they required extensive analysis, complicated calibrations or didn't work well in all situations.

Fugro Airborne Surveys has tested and now employs a new multi-channel technique developed recently by Hovgaard and Grasty (see reference 1 below) to reduce statistical noise in radiometrics data. This method (described as noise adjusted singular valve decomposition or NASVD), analyses multi-channel survey data to identify all statistically significant spectral shapes. These "spectral components" are used to reconstruct new potassium, uranium and thorium window values, which then have significantly less noise than the original raw windows. For the uranium window, in particular, it is possible to achieve a reduction in statistical noise equivalent to an increase in detector volume by a factor of three or more. The spectral component method results in a more accurate measure of the ground concentration, which improves considerably the discrimination between different geologic units with similar radioelement concentrations.

Comparison with Spectral Fitting

Since this method uses all of the counts in the spectrum as well as the correlation between the radioelements, the reduction in statistical noise is greater than that achieved with multi-channel spectral fitting. This new method also has a distinct advantage over multi-channel spectral fitting procedures, in that no model experiments are required. Once the new window totals are determined, the traditional method of data reduction employing the traditional calibration factors is applied to achieve ground concentrations with lower errors due to statistical noise.

Benefits

The spectral component technique results in a reduction of the statistical error associated with each measurement point (usually one second of flying time) which results in an improved precision (and accuracy) of radioelemnt ground concentrations. No significant improvement occurs for the total count window since it already incorporates a major part of the gamma spectrum. The most significant benefit occurs for the lowest count-rate measurement, which is generally uranium and to a lessor extent thorium. The potassium result usually improves the least since it tends to have the highest count rate. The improved maps or images often reveal patterns and shapes previously hidden or barely discernable in the noise - particularly for uranium.

We apply spectral component analysis to improve the precision of the upward-looking-sensor technique for radon removal. Current developments indicate that it may also be applied to the low-energy peak method as well.

Another notable advantage of spectral component analysis is its effectiveness for quality control of radiometrics data. It is extremely sensitive to changes in the spectrometer performance and will detect such things as spectral drift, changes in sensitivity and noise in the spectrum. In many cases, post-flight detection of such problems allow them to be corrected. It should be noted, however, that spectral component analysis is not a cure-all for poorly acquired or improperly calibrated AGS data. Fugro Airborne Surveys continues to emphasise AGS acquisition and calibration procedures of the highest available quality.

Nuclear Environmental Surveys

The spectral component technique is well suited to application to environmental surveying for man-made radioactivity due to atmospheric weapons testing or other sources. In particular, low-levels of 137cesium are amenable to this technique (see reference 2, below). A significant advantage of this approach is that the identity of such gamma-emitting nuclides need not be known in advance of surveying.

Existing Survey Data

Fugro Airborne Surveys can apply NASVD spectral component analysis to any survey for which spectral data is available, enabling a significant improvement over traditional results.

Reference 1: Hovgaard, J. and Grasty, R. L.; "Reducing statistical noise in airborne gamma-ray data through spectral component analysis"; in Proceeding of Exploration 97, 1997; p.753-764

Reference 2: Hovgaard, J. ; "A new processing technique for airborne gamma-ray data (noise adjusted single value decomposition); Am. Nucl. Soc. Symp. on Emergency Preparedness and Response; San Francisco 1997.