Water Retention of Salt-Affected Soils: Quantitative Estimation Using Soil Survey Information

Abstract

Soil water retention (SWR) at −0.1, −33, −1500, and −150000 kPa matric potentials and available water content (AWC) were estimated from information available from 729 horizons of salt-affected soils in the Hungarian Detailed Soil Hydrophysical Database. Soil characteristics of the 1:10,000 scale Hungarian soil maps were used as input parameters. Ordinal and nominal (categorical) variables: texture, organic matter content, calcium carbonate content, soluble salt content, pH, and soil subtype classes of the soil map were used to develop a new prediction method based on the CHAID classification tree. Results of the model development were compared with results using conventional prediction methods (classification tree (CRT) and multiple linear regression (MLR)). Four types of pedotransfer rules were established by classification tree methods. The first rule uses continuous-type input parameters, the second uses soil taxonomical information in addition, the third and fourth one uses categorical-type input parameters. In addition, continuous pedotransfer functions (point estimations) were established as well. Results show that the root mean square error (RMSE) of the developed method is between 1.25 vol% (150000 kPa) and 6.40 vol% (−33 kPa). With the mentioned available input parameters, for salt-affected soils the prediction reliability is similar with categorical and continuous-type information. To predict SWR from categorical-type information the CHAID method is advisable. In the case of continuous-type input parameters MLR is suggested, based on this study. The established hydropedologic methods can be readily used to prepare available water content maps for the topsoil of salt affected soils based on solely soil survey information.

Supplemental materials are available for this article. Go to the publisher's online edition of Arid Land Research and Management to view the free supplemental file.

Keywords:

• CHAID
• hydropedology
• multiple linear regression
• pedotransfer function
• pedotransfer rule
• regression tree
• soil map
• Solonchak
• Solonetz
• water capacity

Acknowledgments

In the framework of Grant No. T048302 provided by the National Scientific Research Fund (OTKA) we had the opportunity to develop the Hungarian Detailed Soil Hydrophysical Database (Hungarian acronym: MARTHA) with the collaboration of the County Offices of the Hungarian Plant and Soil Protection Service. This work was supported by the MyWater project which is co-funded by the European Community's Seventh Framework Programme (FP7/2007- 2013). We would like to thank Dr. Csilla Farkas, Hilda Hernádi, and Péter Marth for their work during the database construction, Alessandro Rossi, for his support to the statistical analysis and Henk Wösten and Attilla Nemes for their valuables comments on the manuscript.

Notes

^a Standard deviation.

*Correlation is significant at 0.05 level.

**Correlation is significant at 0.01 level.

^a The plant available water content calculated from predicted SWR values.

^b Those samples had measured soluble salt content information which was necessary to predict SWR at −33 kPa and AWC by the developed MLR method.

CHAID, CRT_kat: classification tree with category type independent variables and Hungarian soil subtype. The intervals of the soil properties’ categories can be found in the appendixes (App. 1–5.). CRT1: regression tree based on soil properties similar to the input variables of the first prediction method (CHAID) but in continuous form and without Hungarian soil subtype: clay (<0.002 mm) (mass %), silt (0.002–0.05 mm) (mass %), sand (0.05–2 mm) (mass %), organic matter (mass %), calcium carbonate (mass %), pH_H2O, and soluble salt content (mass %). CRT2: regression tree with the input parameters of the CRT1 method and the Hungarian soil subtype.

Input variables correspond to soil properties available from soil maps. Description of the input variables: clay (<0.002 mm) (mass %), silt (0.002–0.05 mm) (mass %), sand (0.05–2 mm) (mass %), organic matter content (mass %), calcium carbonate content (mass %), and pH_H2O, soluble salt content (mass %).