Understanding the utility of aerial gamma radiometrics for mapping soil properties through proximal gamma surveys

Abstract

Historical aerial gamma radiometrics have been proposed to be an important covariate for characterizing soil properties because it provides information about soil parent material. Pre-existing aerial gamma radiometrics within the United States, however, can exhibit coarse spatial resolutions and therefore may be unhelpful for soil mapping studies. Therefore, the objective of this work was to test the hypothesis that aerial gamma radiometrics can reliably map soil properties. The hypothesis was tested using proximal radiometrics and soil sampling. Proximal or ground surveys were conducted within four different heterogeneous landscapes, and 112 soil samples were collected and characterized for texture (i.e. particle size fractions) and/or calcium carbonate equivalent. Proximal and soil relationships were assessed in terms of significance using Pearson correlation coefficient testing and stepwise backwards linear regression. Proximal data were then weighted and averaged within the aerial sensor field of view and subsequently tested for significance with historical aerial data using Pearson correlation tests. Proximal and aerial sensors were then compared in their ability to predict nearby measurements of clay and sand content. Relationships between texture and proximal gamma measurements were significant (p<0.01) when variability in clay content was present, while calcium carbonate equivalent and proximal gamma signatures were significantly correlated when clay content variability was low. Relationships between proximal and aerial measurements were only meaningful when the latter were adequately located in geographic space. When aerial spatial positions were corrected, they significantly explained about half as much of clay and sand content variability relative to proximal gamma measurements. Therefore, aerial gamma information has value in characterizing soil spatial variability, but attention should first be given to the data quality of these historical surveys.