Predicting Neutron Attenuation from Bulk Density and Moisture for Soil Carbon Measurement

TL;DR

This work addresses the challenge of correcting neutron attenuation in INS-API soil measurements to enable accurate, non-destructive mapping of soil carbon. It develops an empirical attenuation model derived from Monte Carlo simulations that express neutron attenuation as a function of dry bulk density $\rho_{bd}$ and volumetric water content $\theta$, incorporating hydrogen sources from free water, lattice water, and organic matter. The model is calibrated and validated using 33 real soils and 4 synthetic soils, with attenuation predictions achieving typically $\lesssim 10\%$ error at 30 cm depth and experimental confirmation across moisture and depth variations. Practically, the approach enables field-ready attenuation corrections for INS-API data, paving the way for self-consistent, voxel-level soil carbon assessments in agricultural and rangeland contexts.

Abstract

Inelastic neutron scattering (INS) enables rapid, non-destructive in situ measurements of soil elemental composition over large soil volumes. Standard INS yields bulk elemental concentrations, but spatially resolved measurements require techniques such as Associated Particle Imaging (API), which pairs neutron detection with coincident alpha detection to reconstruct the location of the neutron interaction. One of the unique advantages of API is its capability to measure all major soil components simultaneously, allowing for the estimation of both bulk density and water content directly from the measured neutron-induced gamma-ray spectra. Accurate interpretation of bulk INS-API data depends on correcting for both gamma-ray and neutron attenuation in soil. Although gamma attenuation can be calculated from known mass attenuation coefficient data and density, neutron attenuation is more complex, depending on neutron energy, soil composition, bulk density, and hydrogen content from water and organic matter. We use Monte Carlo simulations of soils with varied compositions, bulk densities, and water contents to model neutron attenuation and develop a simple predictive model requiring only dry bulk density and volumetric water content. We validate this model experimentally using an INS-API system with controlled soil columns, finding agreement within 10 percent at 30 cm depth. This approach enables practical, field-ready correction of INS-API measurements for neutron attenuation, laying the groundwork for a self-consistent measurement framework that can address the elemental composition of soil carbon assessments.

Figures (9)

• Figure 1: Schematic of the INS–API system showing the emission of a 14.1 neutron and associated 3.5 alpha particle from the D–T reaction. The alpha particle is detected by a YAP scintillator, providing the neutron’s initial trajectory and time of emission, while gamma rays from neutron interactions in the soil are detected by LaBr$_3$ and NaI scintillators.
• Figure 2: Experimental setup for neutron attenuation measurements. A steel soil container is placed between the neutron source and a graphite block target. Soil depth and moisture were varied to assess their effect on neutron attenuation.
• Figure 3: Voxelized $XZ$ and $YZ$ cross-sections of the reconstructed carbon block signal. (a,b) Steel box only; (c,d) 10 cm soil at 40% VWC; (e,f) 18 dry soil. Voxel boundaries capture the carbon block signal without crossing the API cone edge.
• Figure 4: Mass attenuation coefficients for a 4.43 gamma ray plotted vs. total density of solids and liquids in soil (a). Multiplication by total density yields the linear attenuation coefficient (b).
• Figure 5: Normalized 4.43 gamma production vs. depth for four synthetic soils (a) and for uncollided 14.1 neutrons only (b).