Precision and Bias in Field Methods for Measuring Soil Saturated Hydraulic Conductivity
The Subsurface Flow Module within COMSOL Multiphysics® was utilized to simulate variable saturated subsurface flow for multiple field methods that estimate the saturated hydraulic conductivity (Ksat) of the soil. The subsurface flow module implemented the Richards equation with a user-defined soil-water retention function following Vogel et al., (2001) that improves the van Genuchten – Mualem model for fine textured soils while retaining the performance in coarse textured soils. Each simulated field method utilized a 2-D axisymmetric domain with a free draining lower boundary condition where aspects of infiltration within a ring type device are output. Field methods analyzed include methods that utilize a 1-dimensional flow assumption, and methods that use corrections to account for divergent 3-dimensional flow. Different field methods use either a constant head or falling head boundary condition within the ring or rings. Constant head conditions are simulated using a constant pressure head boundary condition and the volumetric flow rate into the soil domain within the innermost ring. Falling head conditions are implemented by converting the time variable boundary flux within the ring device to an equivalent depth and subtracting from the initial pressure head using the Global ODE interface and nojac operator; this change in pressure head is output from the simulation. Post-processing of the COMSOL® simulated output for each method was in accordance with the associated field method to identify a simulated Ksat value. A comparison of the simulated Ksat to the known input Ksat thus identifies the precision and potential systemic bias associated with each field method.
Natural heterogeneity in the field spanning orders of magnitude over small spatial scales and the lack of an independent reference standard for measuring Ksat prevent a field or laboratory analysis from identifying the accuracy of field methods. The American Society for Testing and Materials (ASTM) provides standard procedures for several industry standard field methods, however the ASTM asserts that no statement can be made regarding precision or bias and the measured infiltration rate should only be considered an index value. In practice, these field methods are used to make engineering decisions with substantial implications to local hydrology, water quality, and financial expense. As the precision and bias are not yet defined but the results are used for impactful decisions, these COMSOL® models were established to systematically model the field methods in an environment with known soil properties. Simulations included a range of realistic soil parameters to capture the range of conditions typically observed in the field. Results indicate the simulated Ksat may vary from the input Ksat by a factor ranging from 1.1 to greater than 3, with the magnitude of the bias factor depending on the method, soil type, and antecedent soil moisture.
