Characterization of effective saturated hydraulic conductivity in an agricultural field using karhunen-loève expansion with the Markov chain monte carlo technique

Process-based soil hydrologic models require input of saturated hydraulic conductivity (K_sat). However, model users often have limited access to measured data and thus use published or estimated values for many site-specific hydrologic and environmental applications. We proposed an algorithm that uses the Karhunen-Loève expansion (KLE) in conjunction with the Markov chain Monte Carlo (MCMC) technique, which employs measured soil moisture values to characterize the saturated hydraulic conductivity of an agricultural field at a 30 m resolution. The study domain is situated in the Walnut Creek watershed, Iowa, with soybean crop (in 2005) and well-defined top (atmospheric) and bottom (groundwater) boundary conditions. The KLE algorithm parameterizes and generates Ksat fields with random correlation lengths that are used in the SWMS-3D model for predicting the soil moisture dynamics for two different scenarios: (1) the van Genuchten soil hydraulic parameters (except K_sat) are constant and are based on the soil type of the grid block within the domain, and (2) Ksat is correlated with the van Genuchten parameter a as K_sat α α2. The predicted soil moisture fields for both the scenarios are evaluated with the measured soil moisture in the MCMC algorithm for acceptance (or rejection) of the Ksat fields. The accepted Ksat fields are evaluated against the laboratory-measured K_sat at specific locations as well as with a large K_sat data set measured in situ in a nearby field with similar hydrologic conditions, and the comparisons show reasonably good agreement. The KLE-MCMC algorithm was further tested in the same study domain for another year (2002) having different vegetation (corn) and local forcings. The algorithm shows potential to characterize the effective saturated hydraulic conductivity fields at 30 m resolution using inexpensive and more regularly measured soil moisture measurements. Further studies are required to incorporate variability in different hydroclimatic regions and diverse topography to extend the application of this algorithm.