Potential of soil water content mapping using electromagnetic induction in a low conductive coniferous forest ecosystem catchment

Abstract:

Various studies have demonstrated the potential of electromagnetic induction (EMI) techniques to infer soil water content patterns at the field scale. In this study, we used EMI to characterize the spatio-temporal variability of soil water content (SWC) in a coniferous forest ecosystem catchment. To this end, four time-lapse EMI data sets were recorded within a period of one year using two integral investigation depths (80 cm and 160 cm) within the catchment with an area of ~ 38.5 ha. Independent information on SWC and porosity were provided by a wireless soil moisture sensor network with 110 measurement locations equipped with sensors at three depths (5, 20 and 50 cm). A linear regression between EMI data and SWC resulted in an R² of only ~ 0.4. Fitting of two more complex models (a quadratic function and the full Archie equation that including variable porosity) yielded progressively higher R² values (up to 0.66). The best model had a root mean square error between measured and predicted SWC of 4 vol. %. However, EMI measurements required individual calibration to soil water content measurements for each measurement day, which limits the applicability of EMI for mapping and monitoring SWC in this particular catchment. To explain the variability in calibration functions between different days and to understand the remaining uncertainty of the model predictions, we derived maps of the electric conductivity of the pore water (s_w) by assuming that the residuals between measured and predicted SWC in the Archie model are solely due to spatial variability of s_w. These maps were highly structured and showed similar repeating patterns in s_w for all survey days. The mean s_w varied considerably between survey days, indicating that the EMI-based SWC predictions are affected by spatio-temporal variations in s_w of the forest soils. This study indicates that the accuracy of EMI-based SWC predictions can be limited at certain test sites due to spatial and temporal variations of other soil properties, such as porosity and s_w.

Key words: soil water content, SWC, EMI, SWC prediction, porosity, soil water conductivity, coniferous forest ecosystem catchment, Archie law