Soil Water Retention Curve Development with Continues Time Series Soil Moisture Data

The monitoring and understanding of the movement of water through soil has broad applications not limited to irrigation and food security, flood risk, landslide rick assessment and wildfire risk management and many other applications relating to water resources.
The Richards Equation has become one of the most common ways to describe and quantify the transient movement of water in unsaturated porous media. The unsaturated hydraulic conductivity is important for both physical based models such as HYDRUS and for machine learning statistical based models to predicting soil moisture.  Hydraulic conductivity is determined from a water retention curve which is a plot of soil moisture verses unsaturated hydraulic head.  The numerical solution for the unsaturated hydraulic conductivity from the water retention curve is the van Genuchten equation. The water retention curve, however, is often difficult to obtain in the laboratory and difficult to apply to soil moisture data in the field. Development of the soil water retention curve while continuous logging soil moisture on the individual soil samples could reduce error, speed up the analyses time, and can be applied directly to soil moisture data collected with permanently installed soil sensors in the field   

 A fixture has been developed to hold a soil sensor in a soil sample during the laboratory development of a soil water retention curve. This fixture consists of a soil core sample in a metal ring with a volume of 400 cc, an SDI-12 HydraProbe soil sensor, and a mesh on the bottom holding the soil sample in place while allowing the exchange of water in and out of the sample.  The fixture was placed on ceramic plates and equilibrated to pressures from 0 to 1500 kPa while collecting data every minute. A special bulkhead was developed to get sensor communication out of the pressures plate extractor. Data on a smaller soil sample 21 cc in volume were collected alongside the lager soil sample for comparison. Gravimetric data was collected on all samples at the end of each cycle. Because it takes longer for a larger 400 cc soil sample to de-water, equilibrium was approximated using an exponent regression.   

Six soil samples 400 cc and twelve 21 cc soil sample were analyzed. The soil was an alpine soil   from the Siera Nevada Mountain range in Southern California representing three depths of 10, 50 and 100 cm. The 10 cm depth was highly organic while the 50 and 100 cm depths were mostly sand.   

The van Genuchten parameters, alpha, n, saturation and water residual were determined and comparisons were made between the larger 400 cc samples with the soil sensors and the smaller 21 cc soil samples. The site-specific soil moisture calibrations for the HydraProbe soil sensor had   R² values from 0.98 to 0.9985 with RMSE values from 0.002 to 0.015 wfv.