An integrated framework for interpolating airborne geophysical data with special reference to radiometrics

Abstract

New processing techniques for airborne radiometric data make use of the information contained in all 256 channels of a radiometric spectrum, improving the final quality obtained. However, visualisation and interpretation of the processed data require interpolation to a regular grid and current methods for doing this are generally unsatisfactory. We highlight alternative interpolation techniques (kriging, radial basis functions, tension splines, smoothing splines) that overcome some of the disadvantages of existing methods. These techniques are formulated in a common mathematical framework and can be used for exact or smooth interpolation of the processed data. The resulting grids can be made to inherit certain desirable characteristics, such as smoothness or minimum variance. Further, the framework generates a continuous model of the data that can be updated rapidly when the image is visualised at different scales. Until recently, the main impediments to the application of the technique to large geophysical surveys have been the computer memory and effort needed to solve the resulting matrix equations. We describe some recent advances that reduce the computational requirements to acceptable levels. We describe an extension to the usual multi-channel technique that, during processing, preserves the original signal (as embodied in the 256-channel spectra) for as long as possible. We show that better images can result if the spectral components are gridded and the spectra reconstructed at the grid points. The reduction to standard 4-channel count-rates and conversion to ground concentration occur in the final processing step.