For accurate measurements of soil water content with sensors that measure the apparent dielectric permittivity of the soil, we estimated the measurement errors using the manufacturer's default calibration function for soils with different properties. A high clay and organic matter content and a high soil bulk density had the greatest influence on the errors. Soil physical properties were evaluated at three depths on a plot with different tillage systems and soil water status was measured continuously during two growing seasons. Soybean (Glycine max (L.) Merr.) was grown in the first growing season with more frequent rainfall, and maize (Zea mays L.) in the second. Field soil water retention curves were determined and their spatial and temporal variability was evaluated. The shape of the curves was described using the parameters of the van Genuchten model. The field retention curves varied spatially, within and between growing seasons. Based on continuous measurements of soil water content in space, we evaluated the measurement errors with decreasing spatial and temporal resolution of the measurements. For each plot (tillage system and depth), an error in estimating the spatial mean of less than ± 0.02 m3 m⠒3 is achieved when using four measurement locations. When using only two spatial locations, the maximum error in the estimation of the spatial mean was up to ± 0.08 m3 m⠒3. The maximum differences in daily soil water content are adequately captured up to a temporal resolution of 4 hours. Based on a model of the spatio-temporal correlation structure of soil water content at a depth of 20 cm during the 2021 growing season, we used a spatio-temporal model to predict values at unmeasured loactions and to evaluate the uncertainty of the prediction.
|
