Gradients and Assumptions Affect Interpretation of Laboratory-Measured Gas-Phase Transport

A fundamental principle of measurement and analysis is that methods must be self-consistent. For example, if a soil's gas diffusion properties were precisely known, then measuring gas diffusion in that soil should yield the known result. We tested the self-consistency of a standard protocol for analyzing gas diffusion measurements and found that it has a tendency to introduce artifacts. We examined the impacts of four assumptions: the presence or absence of a gradient in air-filled porosity epsilon across the sample, the kind of average used to give the effective diffusivity D(epsilon)/D-0 across the sample, the physical presence of a gas fraction threshold epsilon(t) > 0 for transport, and the inclusion of such a threshold in the model fitted to the data. We examined the effects of correcting the gradient artifact and using different models in analyzing two published data sets. If the soil has a non-negligible epsilon(t) > 0, but the model fitted to the data is of the form D(epsilon) = A epsilon(N) (forcing epsilon(t) = 0), then the fitted exponent N will artifactually increase with increasingly narrow pore size distributions. This artifact of forcing epsilon(t) = 0 calls into question models that prescribe higher N in coarser soils.