摘要：Core Ideas Bulk density changes caused large uncertainties in neutron moisture meter calibrations. Air‐ or water‐filled annuli around access tubes simulated voids to measure error. Error magnitude depended on void size and contrast between soil and void water content. Experimental results showed much smaller errors than previous numerical simulations. Larger voids reduced sensitivity of neutron moisture meter to soil water content. Air‐ and water‐filled voids around neutron moisture meter (NMM) access tubes have been cited as sources of volumetric water content (θ v ) measurement error in cracking clay soils. The objectives of this study were to experimentally quantify this potential error stemming from (i) uncertainty in bulk density (ρ b ) sampling and (ii) the impact of air‐ and water‐filled voids. Air‐ and water‐filled voids were simulated using ∼0.6‐cm (small) and ∼1.9‐cm (large) annuli around access tubes. After NMM measurements were taken in a tightly installed access tube, either a small or large annulus was installed in the same borehole. Additional NMM measurements were taken with the annulus filled with air, and then water and ρ b and θ v were measured. The RMSE of the NMM calibration using all 11 installations was 0.02 m 3 m −3 . However, if two cores were used for calibration, the ratio of NMM‐measured θ v to in situ θ v was significantly different ( p < 0.05) from measured θ v half the time (RMSE, 0.012–0.05 m 3 m −3 ). Small air‐filled voids created drier estimates of θ v (bias, −0.039 m 3 m −3 ; p < 0.001), wherease small water‐filled voids were not significantly different from the calibration. Air‐ and water‐filled voids from larger annuli were significantly lower and higher ( p < 0.001) than core‐measured θ v , with biases of −0.068 and 0.080 m 3 m −3 , respectively. Although this work does not correct NMM‐predicted θ v to matrix θ v , it does bound NMM error under field conditions in a cracking clay soil.
