Can soil water measurements at a certain depth be used to estimate mean soil water content of a soil profile at a point or at a hillslope scale?

- Vertical patterns of soil moisture (θ) for 0-3.8 m soil profile tended to be time stable.
- Prediction error for mean θ varied with sampling locations and soil depths.
- It was unreliable to use θ of surface layer to predict mean θ of a soil profile deeper than 0.2 m.
- Soil profile θ at point and hillslope scales can be predicted with measurements at the most time stable depth.
- The proposed approach was further verified in another area with a different climate.

SummaryProfile soil water contents (θ) are important for scheduling irrigation, determining root water uptake and energy partition between sensible and latent heat. Soil water content measurements for deep soil profiles are time consuming and costly. The objective of this study was to test whether θ at a certain depth at a point can be used to estimate profile θ at a point or at a hillslope scale by time stability analysis. A total of 37 datasets of θ from 0.1 m to 3.8 m depths were collected by a neutron probe at 28 locations over four years on a representative hillslope in the Chinese Loess Plateau. Time stability of vertical patterns of θ was assessed by Spearman's rank correlation analysis. Soil water contents of the first two years were used to identify the time stable depth using mean absolute bias error, and θ of the second two years were used to validate if the time stable depth identified and associated mean relative difference can be used to predict mean θ of a soil profile at a point or at a hillslope scale. Results showed that vertical patterns of θ were time stable. The prediction error for mean θ varied with sampling locations and soil profile depths. At the most time stable location (location 4) in terms of vertical patterns, mean θ of soil profiles (i.e., 0-1.0 m, 0-2.0 m, 0-3.0 m, and 0-3.8 m) was predicted well by the most time stable depth, with absolute bias relative to mean <0.05 at a point scale and <0.10 at a hillslope scale. Further application of this approach in the Canadian Prairies site indicated that mean θ of 0-1.0 m soil profile at a point or a transect scale was predicted well by the most time stable depth at almost all the locations. This study verified that soil profile θ at both point and hillslope (or watershed) scales can be predicted with the θ measurements at the most time stable depth.