Soil Hydraulic Modeling Outcomes with Four Parameterization Methods: Comparing Soil Description and Inverse Estimation Approaches. Issue 1 (29th March 2018)

• Record Type: Journal Article
• Authors: Graham, Scott L.; Srinivasan, M.S.; Faulkner, Nathalie; Carrick, Sam
• Is Part Of: Vadose zone journal. Volume 17:Issue 1(2018)
• Page Start: 1
• Page End: 10
• Publication Date: 2018-03-29
• Subjects: Soil science -- Periodicals; Zone of aeration -- Periodicals; Groundwater flow -- Periodicals; Groundwater flow; Zone of aeration; Periodicals; Electronic journals; 631.4
• DOI: 10.2136/vzj2017.01.0002
• Languages: English
• ISSNs: 1539-1663

Abstract

Abstract : Core Ideas: Soil hydraulic model outcomes are sensitive to the data source used for parameterization. Inverse parameter estimation from field data improved drainage estimates. All parameterization methods lead to underestimation of field‐measured evaporation. Different methods for parameterizing soil hydraulic models can lead to substantially varied predictions of soil–plant–atmosphere water fluxes. This study investigated, for a heterogeneous stony soil, four methods of soil hydraulic parameterization: (i) use of a pedotransfer function with a four‐layer soil profile based on detailed soil physical and textural description; (ii) use of a pedotransfer function with a single‐layer soil description; (iii) inverse estimation from soil moisture data; and (iv) inverse estimation from lysimeter drainage. Soil drainage, volumetric water content, and evapotranspiration were each modeled using HYDRUS‐1D for an irrigated pasture in New Zealand during the time period 1 July 2011 to 15 Mar. 2014. The first 15 mo were used for model spin‐up and inverse parameter estimation, while the remainder of the study period was used as a validation period, during which model results were compared against field data. Predictions from each model parameterization were compared with field‐measured fluxes from lysimeter, soil moisture sensors, and eddy covariance to determine the approach most appropriate for our site and application. The parameters estimated inversely from field data improved the modeling of soil drainage, leading to total drainage within 5 to 7% of the measured volume and prediction of 35 to 80% more drainage peaks than parameterizations based on detailed soil physical description. While all methods underpredicted evapotranspiration by 18 to 30% compared with eddy covariance, improvement in drainage estimates with inverse estimation from field data led to decreased capability for modeling evapotranspiration. We suggest this approach for application in other settings to select the most appropriate parameterization approach for a given soil hydraulic model application.