Investigating the Potential of Cosmic-Ray Neutron Sensing for Estimating Soil Water Content in Farmland and Mountainous Areas

Abstract

The conventional methods of estimating soil water content (SWC) are mainly based on in situ measurements at sampling points and remote sensing measurements over an entire region. In view of these methods, cosmic-ray neutron sensing (CRNS) has received increasing attention in recent years as a mesoscale, noncontact SWC estimation technology that can provide more accurate and timely estimates of SWC over a larger area. In this study, we estimated SWC using both CRNS and soil-mounted detectors in farmland and mountainous areas, and evaluated the accuracy of the estimations at two experimental sites. Ultra-rapid adaptable neutron-only simulation (URANOS) was used to simulate the detection radius and depth of the two experimental sites and to obtain the spatial weights of the CRNS footprint. The results show that the theoretical range of detection was reduced in farmland compared to mountainous areas during the experimental period, suggesting that farmland retained more SWC even with less precipitation. Spatial weights were simulated to calculate the SWC of sampling points, and the weighted and averaged SWC were then correlated with CRNS. The weighting calculation improves the accuracy of CRNS estimations, with a determination coefficient (R2) of 0.645 and a root mean square error (RMSE) of 0.046 cm3·cm−3 for farmland, and reproduces the daily dynamics of SWC. The R2 and RMSE in mountainous areas are 0.773 and 0.049 cm3·cm−3, respectively, and the estimation accuracy of CRNS cannot be improved by the weighting calculation. The estimation accuracy of CRNS is acceptable in both regions, but the mountainous terrain obstructs neutron transmission, causing a deviation between the actual and theoretical neutron footprints in mountainous areas. Thus, the accuracy of SWC estimation is limited in mountainous terrain. In conclusion, this study demonstrates that CRNS is suitable for use in farmland and mountainous areas and that further attention should be given to the effects of topography and vegetation when it is applied in mountainous environments.