Submitted to: World Congress of Soil Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: August 16, 2002
Publication Date: August 16, 2002
Citation: Wosten, H., Nemes, A., Pachepsky, Y.A., Rawls, W.J., Zeiliguer, A. 2002. Translating available basic soil data into missing soil hydraulic characteristics. World Congress of Soil Science. Vol.4:135-143
Interpretive Summary: Any forecast of water use efficiency or water budget requires information about ability of soils to retain and to transmit water. This information is usually compressed in soil water retention and hydraulic conductivity functions. Those functions are costly to measure, and it is impractical to determine them in large-area projects or in pilot studies. Therefore, soil hydraulic pedotransfer functions, that transfer simple-to-measure soil survey information into soil hydraulic characteristics, offer an attractive alternative to dense sampling of the project area. Because pedotransfer functions are empirical equations, caution has to be exercised when those functions developed, evaluated, and used. This paper discusses and illustrated pitfalls that may be encountered. There is a room for improvement of pedotransfer functions accuracy by using novel database analysis tools, such as artificial neural networks, group method of data handling, and regression trees. A preliminary grouping of soil can help to decrease the accuracy. An increase in accuracy with data from which pedotransfer function is developed is not necessarily associated with the increase in reliability, i.e. in accuracy with data other than the development data set. Both accuracy and reliability of a pedotransfer function have to be evaluated with respect to a particular application, that is, to the sensitivity of a predictive model to the errors given by the pedotransfer function in use. The most reliable pedotransfer functions to-date are built from large regional databases, it is recommended that large and reliable global databases of soil hydraulic data have to be created.
Soil hydraulic pedotransfer functions transfer simple-to-measure soil survey information into soil hydraulic characteristics, that are otherwise costly to measure. Examples are presented of different equations describing hydraulic characteristics and of pedotransfer functions used to predict parameters in these equations. Grouping of data prior to pedotransfer function development is discussed as well as the use of different soil properties as predictors. In addition to regression analysis, new techniques such as artificial neural networks, group method of data handling, and classification and regression trees are increasingly being used for pedotransfer function development. Actual development of pedotransfer functions is demonstrated with a case study. Accuracy and reliability of pedotransfer functions are demonstrated and discussed. In this respect, functional evaluation of pedotransfer functions proves to be a good tool to assess the desired accuracy of a pedotransfer function for a specific application. Functional testing of uncertainty in the soil hydraulic input data reveals to what extent variability in modeling results is explained by uncertainty in pedotransfer functions. Linking pedotransfer functions to soil survey information results in much needed soil hydraulic characterization of large areas. Since pedotransfer functions offer sufficiently accurate predictions for many applications, it is recommended that large and reliable global databases of soil hydraulic data have to be created.