|GHANBARAN, BEHZAD - University Of Austin|
|TASLIMITEHRANI, VAHID - Nutrition Physiology Company, Llc (NPC)|
Submitted to: Catena
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/21/2016
Publication Date: 1/25/2017
Citation: Ghanbaran, B., Taslimitehrani, V., Pachepsky, Y.A. 2017. Accuracy of sample dimension-dependent pedotransfer functions in estimation of soil saturated hydraulic conductivity. Catena. 149:374-380.
Interpretive Summary: The majority of models concerned with soil water contents are using the parameter of soil saturated hydraulic conductivity, or Ksat, that characterizes the ability of soil to conduct water. Measurement of Ksat is time- and labor-consuming and is impractical in large-scale projects. For this reason, values of Ksat are usually estimated from more readily available soil properties. Recently we analyzed a mid-size international soil database and showed that values of Ksat depend on the size of the soil sample used in measurements. The objective of this work was to evaluate the importance of accounting for the sample size in Ksat predictions with the data from a recently developed ARS-USDA database covering the continental United States. We found that the accuracy of predictions that take into account sample size is substantially better than accuracy of six widely used estimation methods that do not include the size dependence. Results of this work are important for the modeling research and applications in meteorology, agronomy, hydrology, and related fields, in that they show the direction in which substantial improvement in soil Ksat prediction can be made.
Technical Abstract: Saturated hydraulic conductivity Ksat is a fundamental characteristic in modeling flow and contaminant transport in soils and sediments. Therefore, many models have been developed to estimate Ksat from easily measureable parameters, such as textural properties, bulk density, etc. However, Ksat is not only affected by textural and structural characteristics, but also by sample dimensions e.g., internal diameter and height. Using the UNSODA database and the contrast pattern aided regression (CPXR) method, we recently developed sample dimension-dependent pedotransfer functions to estimate Ksat from textural data, bulk density, and sample dimensions. The main objectives of this study were evaluating the proposed pedotransfer functions using a larger database, and comparing them with seven other models. For this purpose, we selected more than nineteen thousand soil samples from all around the United States. Results showed that the sample dimension-dependent pedotransfer functions estimated Ksat more accurately than seven other models frequently used in the literature.