Accounting for soil spatial variability in soil water capacitance probe calibration

Field calibration of a capacitance probe can become a challenge in long-term studies in which spatio-temporal behavior of soil water content is investigated for field-scale crop water uptake. The small physical sphere of influence can be complemented by the spatial range of representativity in presence of considerable soil spatial variability. For long-term studies on spatiotemporal processes of soil water focusing on spatial variability at the field scale, appropriate calibration techniques for capacitance sensors are lacking. In an ideal case, soil water content samples would need to be taken in the close vicinity of the access tube, that is, within the physical sphere of influence, for sensor calibration. This destructive sampling procedure is prohibitive because the measurement site of that particular access tube would be destroyed for any further capacitance probe measurements. On the other hand, despite the advantage of controlled boundary conditions, it is difficult or impossible to conduct a lab calibration of the sensor that would be valid for the site specific conditions in a heterogeneous field soil. The objective of this study was to propose a sampling and data processing procedure to compare six calibration scenarios with respect to the RMSE values of their associated calibration functions and to describe the spatial series of calibration parameters in an autoregressive state-space model. All scenarios were based on a data set obtained during the installation of access tubes and four subsequent soil sampling campaigns in the vicinity of 1-m radius around the tube. Site- and depth-specific calibration yielded the best calibration results, ranging among the best calibration results presented in the literature. Layer-specific stratification of measurement points according to the soil clay content yielded RMSE values approximately twice as high as the site- and layer-specific calibration and ranged among many other published studies. The presented sampling and data-processing procedure in combination with site- and layer-specific calibration are critically discussed and show a promising way for field-scale studies focused on spatiotemporal processes of soil water content and underlying calibration parameters. This study shows that the spatial range of representativity can substantially exceed the small physical sphere of influence of the capacitance sensor and supports its calibration.