Skip to main content
ARS Home » Research » Publications at this Location » Publication #93738
Title: PRECISION OF REGRESSION BASED PEDO-TRANSFER FUNCTIONS EVALUATED BY REPEATEDSAMPLING

Author
item Williams, Robert
item Timlin, Dennis
item PACHEPSHY, YA - 1270-15-00
item Starks, Patrick

Submitted to: Agronomy Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: 10/18/1998
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: Pedo-transfer functions are used to estimate unknown from known soil properties. Here psi(theta) at specific psi are estimated using regression equations derived from texture, bulk density and organic matter. Pedo-transfer functions derived in this manner are developed from large databases to make the regression models universal and robust. However, how do variations in techniques used to determine the soil properties, or the inherit errors in large databases, affect the estimates? Here we use two databases: a larger unmodified database, and a smaller subset of selected data. Part of each database is used to develop the regression model. The remaining data are used to test the regression model and determine psi(theta) estimation errors. The databases are then randomized and the procedure repeated. Using this approach the robustness and precision of the regression models, as reflected in the quality of the data, can be determined. Results will be presented.