Substrate/soil temperature (Ts) is a useful parameter in crop management. Unfortunately, it is also a key factor contributing to the inaccuracy of water content measurements by almost all commercial soil moisture sensors, because Ts changes soil moisture sensor readings (Wraith and Or, 1999; Seyfried and Grant, 2007). This can cause moisture readings to Go Up when the substrate/soil is actually drying out and vice versa.
Figure 1. Soil moisture measurements recorded by a commercial sensor from a major U.S. manufacturer. Soil moisture readings are clearly correlated with soil temperature. Soil moisture measurement errors are significant.
Temperature fluctuations can affect the final accuracy of soil moisture measurements by impacting three main processes:
Manufacturing (e.g. factory calibration)
Soil-specific calibration
In-field soil moisture measurements
If temperature fluctuations are significant in the lab in which the sensor is being assembled and calibrated, then the calibration coefficients are not reliable. This error can be minimized if a temperature chamber or a temperature-controlled room is used. Similarly, temperature fluctuations can undermine the process of soil-specific calibration (in the lab or in the field) as the change in water content values are driven by both temperature and water loss (evaporation and/or transpiration). This will result in in-field measurements showing a trend that is more correlated with Ts than it is with plant water use. This can potentially render any measurements at a resolution of less than 24 hour useless. Also, because the calibration process was affected by Ts, it may have not improved the accuracy of soil moisture measurements.
The temperature affects soil moisture measurements in two key ways:
by affecting the sensor electronic circuitry
by changing the electrical properties of the soil
Electronic components used in soil moisture sensors have properties that change as a function of temperature. Depending on the design, these changes can nullify each other or add up. In most cases the former is the case. If the temperature constant is known or determined, it can be applied to the output moisture readings. In digital sensors, it can be saved on the sensor memory and automatically applied to the readings.
Most commercial soil moisture sensors measure volumetric water content (VWC) by measuring the dielectric permittivity of the bulk soil. The dielectric is correlated with temperature, which means soil moisture measurements using any dielectric-based sensor are correlated with temperature. This problem is worse in sensors with prongs made of steel rods and those that measure both electrical conductivity and moisture by separating the real and imaginary components of the dielectric. This is because the two components have shown to have unpredictable behavior and often opposing correlation with temperature.
Figure 2. Soil moisture readings are automatically compensated in real-time for temperature effect and sent to the computer for further processing.
It is worth noting that the actual temperature sensitivity of commercially available soil moisture sensors of any kind are usually higher than indicated by the manufacturers. The answer most users of soil moisture sensors are used to hearing is that the problem of temperature dependency has a complex nature and it is impossible to resolve. Some solutions such as multiple regression analysis or averaging have been suggested by researchers (Saito et al., 2013; Kapilaratne and Lu, 2017) to remove the effect of soil temperature on probe outputs. However, no real-time automatic solution has been adopted or implemented by sensor manufacturers. These strategies are suggested and often required for sensors that are installed near the soil surface or are used to measure the moisture content of soilless media (e.g. rockwool, coco).
To deal with this issue, we added an accurate temperature sensor to the design of our soil moisture sensor, avoided using steel electrode arrays, and employed an adequately high measurement frequency. The effect of temperature on the electronics was minimized by working on the circuit design. In addition, we developed our own two-stage temperature-compensation algorithm based on multiple regression analysis that takes both temperature and moisture readings and compensates water content measurements for fluctuations that are caused by temperature difference from a baseline temperature (measurements are normalized to 25 °C). The algorithm resides on a dedicated reader or data logger (Fig. 2), and the whole process is worry-free, real-time and automatic. The algorithm is transparent, and the user has full control over it. The parameters can be adjusted, enabled/disabled to suit specific soil, substrate or condition.
References
Kapilaratne, R.G.C.J., Lu, M., 2017. Automated general temperature correction method for dielectric soil moisture sensors. Journal of Hydrology, 551:203-216. https://doi.org/10.1016/j.jhydrol.2017.05.050
Saito, T., Fujimaki, H., Yasuda, H., Inosako, K., Inoue, M., 2012. Calibration of temperature effect on dielectric probes using time series field data. Vadose Zone Journal, 12(8)http://dx.doi.org/10.2136/vzj2012.0184.
Seyfried, M.S., Grant, L.E., 2007. Temperature effects on soil dielectric properties measured at 50 MHz. Vadose Zone J., 6: 759–765. https://www.stevenswater.com/resources/documentation/hydraprobe/Seyfried3.pdf
Wraith, J.M., Or, D., 1999. Temperature effects on soil bulk dielectric permittivity measured by time domain reflectometry: Experimental evidence and hypothesis development. Water Resources Research, 35: 361–369. https://agupubs.onlinelibrary.wiley.com/doi/pdfdirect/10.1029/1998WR900006