HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Figures Only Full Text Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lamorski, K. Articles by Walczak, R. T. Search for Related Content PubMed Articles by Lamorski, K. Articles by Walczak, R. T. GeoRef GeoRef Citation Agricola Articles by Lamorski, K. Articles by Walczak, R. T. Related Collections Soil Physics

SOIL PHYSICS NOTE

Using Support Vector Machines to Develop Pedotransfer Functions for Water Retention of Soils in Poland

K. Lamorski^a, Y. Pachepsky^b,*, C. Sawiski^a and R. T. Walczak^a

^a Institute of Agrophysics, Polish Academy of Sciences, Lublin, Poland
^b USDA-ARS Environmental Microbial Safety Lab., 173 Powder Mill Rd., Beltsville, MD

^* Corresponding author (Yakov.Pachepsky{at}ars.usda.gov).

Pedotransfer functions (PTFs), which estimate soil hydraulic parameters from easy to measure soil properties, are an important data source for hydrologic modeling. Recently artificial neural networks (ANNs) have become the tool of choice in PTF development. Recent developments in machine learning methods include the growing research and application of the alternative data-driven method called Support Vector Machines (SVMs). Support Vector Machines have gained popularity in many traditionally ANN dominated fields. Using the SVM eliminates the local minimum issue—the minimum found is always the global one. The objective of this work was to see whethers using the SVM to develop PTFs may have some advantages compared with the ANN. We have used the Soil Profiles Bank of Polish Mineral Soils that includes hydraulic properties for 806 soil samples taken from 290 soil profiles. This database was repeatedly randomly split into training and testing data sets, and both SVMs and ANNs were trained and tested for each split with bulk density, sand and clay as input variables, and water contents at 11 soil water potentials as the output variables. The PTF performance was evaluated by using the test datasets to compute the coefficient of determination, the root-mean-squared error, and the slope and the intercept of the linear regression "predicted vs. measured water contents." The three-parameter SVMs performed mostly better than or with the same accuracy as the eleven-parameter ANNs. The advantage of SVM was more pronounced at soil matric potentials where larger relative errors have been encountered and the correlation between predicted and measured soil water contents was lower. It is worthwhile to consider SVM as a tool to develop PTF.

This article has been cited by other articles:

				A. Nemes, D. J. Timlin, Ya. A. Pachepsky, and W. J. Rawls Evaluation of the Rawls et al. (1982) Pedotransfer Functions for their Applicability at the U.S. National Scale Soil Sci. Soc. Am. J., August 19, 2009; 73(5): 1638 - 1645. [Abstract] [Full Text] [PDF]