Use of neutron and gamma radiation meters to estimate bulk density and correct for bias of sampling for water content in a swelling clay soil

Two radiation methods were used to estimate the bulk density of a three-dimensionally swelling grey clay soil used for furrow irrigation at Narrabri, N.S.W. Firstly, gamma ray scattering was calibrated with measurements of wet bulk density derived from undisturbed soil cores. Secondly, a high correlation between neutron counts and gravimetric water content in this soil provided a basis for predicting bulk density corrected for bias in sampling of shrinkage cracks by using a published theoretical model of three-dimensional soil shrinkage. Gamma ray backscattering was poorly correlated with wet bulk density (�w), possibly because dry bulk density and water content are negatively correlated in swelling soil, which restricted �w to a relatively narrow range of values. This technique is therefore not recommended for use in this soil. High correlation (0 82 < r < 0.98, all P < 0.001) between neutron counts and bulk density corrected for three-dimensional shrinkage was found at all soil depths between 0.1 and 1.5 m. A precision of k0.01 Mg m-3 required from three to six samples per mean, depending on soil depth. The mean relative difference between predictions of bulk density from neutron counts compared with independent estimates by the core method was <4.1% at depths below 0.3 m. The recommended procedure is therefore to predict bulk density from neutron counts in order to correct for sampling bias and bulk density effects associated with the neutron attenuation method. The method eliminates the need for additional sampling for bulk density in conjunction with the neutron moisture meter in soils that shrink and swell three-dimensionally. However, the method is not appropriate for detecting differences in bulk density between soils with different structure unless the constants used in the model and the shrinkage behaviour are known for each soil. The latter requirements would usually preclude the technique for this purpose. At low water contents near the permanent wilting point for cotton, neutron escape through shrinkage cracks did not cause problems at depths below 0.3 m. The neutron method should therefore be appropriate for use at depths below 0.3 m in dryland hydrological studies in this soil.