Measuring and Estimating Soil Hydraulic Propertyin a Farmer’s Field, Western Kentucky

Xi Zhang Javier Reyes Ole Wendrothxi.zhang@uky.edu javier.reyes@uky.edu owendroth@uky.edu

Department of Plant and Soil Sciences, University of Kentucky, Lexington, KY 40546, USA

This research was conducted at the Hargis field,Hillview Farm located in Caldwell County,Princeton, KY. Wheat/ double-crop soybean/cornrotation is practiced in this farm. The field isclassified as silt loam and silty clay loam soils.Undisturbed soil cores were collected from 10soil profiles at five depths (topsoil: 7-13 cm and27-33 cm; subsoil: 47-53 cm, 67-73 cm, and 87-93 cm) in the field. Ks was determined with apermeameter based on Darcy’s law under con-stant and falling head conditions. Bulk densitywas measured with the core method. Bulk soil iscollected for texture (pipette method) and organicmatter (OM) (combustion method) analysis.Seven traditional regression models and onehierarchical artificial neural networks model wereused to calculate Ks. The Ks from soil survey wasalso considered in this research. The performanceof selected PTFs was evaluated based on rootmean squared error (RMSE).

Materials and Methods

Figure 1. Measured and estimated saturated hydraulic conductivity by pedo-transfer functions and soil survey.

IntroductionSaturated hydraulic conductivity (Ks) is anessential input parameter for hydrological modelsand often used as a matching point in thecalculation of the K(θ) curve. Ks is a stochasticparameter and exhibits high spatial variabilityand scale dependency. Therefore, directmeasurements are often arduous, time consumingand expensive. To avoid these problems, pedo-transfer functions (PTFs) have been developed toestimate Ks indirectly through more easilymeasurable soil properties that may already existfrom the soil survey. However, PTFs wereusually developed at large scales, includingnational, international and intercontinental scales.PTFs derived from large scales may not beapplicable for smaller scales (e.g., field scale).The application of PTFs for estimating Ks at thefield scale is still rarely reported, although mostagricultural management (e.g., irrigation) occursat the field scale.The objective of this research was to evaluatePTFs for estimating Ks at the field scale in a fieldin western Kentucky.

ReferencesArrington K., Venturaa S., Normana J. (2013) Predictingsaturated hydraulic conductivity for estimating maximumsoil infiltration rates. Soil Science Society of AmericaJournal 77:748-758.Parasuraman K., Elshorbagy A., Si B.C. (2006) Estimatingsaturated hydraulic conductivity in spatially variable fieldsusing neural network ensembles. Soil Science Society ofAmerica Journal 70:1851-1859.

ConclusionThe prediction of Ks at the field scale using a PTFdeveloped at large scale is inaccurate. The resultsuggests that existing PTFs need to be calibratedon local datasets or new PTFs should bedeveloped for local field application if necessarybefore relevant predictions can be made.

AcknowledgmentsWe appreciate Riley Walton for the technical support andTrevor Gilkey for allowing us to conduct this research onhis farm.

All of the selected PTFs and soil survey exhibitedlarge differences between the measured and thecalculated Ks, especially for topsoil. PTFs weresuperior to soil survey in general.The performance of PTFs in estimating Ks in thesubsoil was comparable with the evaluationsconducted by Tietje and Hennings (1996) andWagner et al. (2001), and was better than for thetopsoil.Ks in the subsoil was found to be significantly(0.05 level) related to silt (r=0.44), clay (r=-0.46),bulk density (r=-0.59) and OM (r=0.48).However, Ks in the topsoil was only weaklyrelated to texture, bulk density and OM.The Ks in the topsoil probably was stronglyinfluenced by soil structure, which developmentand stability are substantially affected by landmanagement, land use history, and biologicalactivity. Also, the presence of macropores andearthworm channels increased the uncertaintiesand further complicated the prediction of Ks inthe topsoil by PTFs.

Results and Discussion

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

(j) (k)

a Brakensiek et al. (1984)b Cosby et al. (1984)c Puckett et al. (1985)d Saxton et al. (1986)e Vereecken et al. (1990)f Wösten (1997)g Wösten et al. (1999)h Rosetta (textural class)i Rosetta (sand, silt and clay)j Rosetta (sand, silt, clay and bulk density)k Soil survey