Towards improved TDR soil water sensing for optimizing irrigation water management

Authors: Schwartz, Robert; Klopp, Hans; DOMINGUEZ, ALFONSO - University Of Castilla-La Mancha(UCLM)

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 4/23/2023
Publication Date: 4/24/2023
Citation: Schwartz, R.C., Klopp, H.W., Dominguez, A. 2023. Towards improved TDR soil water sensing for optimizing irrigation water management [abstract]. EGU General Assembly 2023, April 24-28, 2023, Vienna, Austria. Paper No. EGU 23-3383.

Interpretive Summary:

Technical Abstract: Decreasing water resources available for irrigation will require a thorough reconsideration of how water is allocated and managed for crop production. Electromagnetic (EM) soil water sensing is an important tool that can facilitate spatial and temporal allocation decisions to increase crop water productivity. Accuracy of volumetric water content measurements in the field, however, is problematic with EM sensors, especially in soils with high clay contents and pronounced horizonation. Under many circumstances, measurement uncertainties are large compared with the range of managed allowed depletion. Soil specific calibrations can improve accuracy although the procedures required to achieve this are normally impractical for routine field deployment of sensors. Herein we present our current efforts in improving the accuracy of TDR soil water sensing and their utility in irrigation management, especially under conditions of limited water availability. Earlier work using a quasi-theoretical model to describe the complex permittivity of soil demonstrated that bound water near clay surfaces and high frequency filtering of the broadband signal were major sources of error for TDR water content estimation. The specific surface area of the soil is partly responsible for these effects, which can also vary in the field because of the dependency of volumetric bound water on bulk density. Although theory can describe how soil apparent permittivity changes with respect specific surface area and bulk electrical conductivity; this does not necessarily reflect how these properties influence the measured travel time. Bound water polarization and dc losses result in signal attenuation of the high frequency components thereby increasing travel time greater than that expected from changes in apparent permittivity. To circumvent these difficulties, we are currently using a supervised machine learning approach to develop an empirical soil water content calibration based on measured travel time, measured state properties (temperature and bulk electrical conductivity), and inferred properties based on TDR waveform features (specific surface area). For example, at a given water content, the shape of the waveform reflection for a soil dominated by kaolinite is distinct from the reflection of a soil dominated by 2:1 phyllosilicates. Essentially the bulk density x specific surface area modifies the waveform features which in turn can be used to develop in essence an in-situ soil specific calibration. Introduction of measured soil water contents into crop models provides a way to facilitate real-time yield predictions of alternative water allocation decisions. The Richards Equation will necessarily be incorporated into crop models to permit a mechanistic basis of redistributing soil water in the profile. Soil water sensing can permit the accurate determination of both irrigation application efficiency and infiltration. Incorporation of measured soil water into crop models allows for “course corrections” of simulated profile water and potentially improvements in the estimation of evapotranspiration and yields.