Premium
Estimating Soil Water Characteristics from Simpler Properties or Limited Data
Author(s) -
Ahuja L. R.,
Naney J. W.,
Williams R. D.
Publication year - 1985
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj1985.03615995004900050005x
Subject(s) - standard deviation , soil texture , water content , scaling , soil science , soil water , mathematics , approximation error , variables , standard error , statistics , environmental science , geology , geometry , geotechnical engineering
Broad‐based regression equations of Rawls and associates were investigated for estimating the spatially variable soil water content‐matric potential relationships in a 1.6‐ha watershed (Udertic Paleustolls) from soil textural and structural properties, with and without one or two known values of the relationships. Also examined were a simple log‐log line based on two known values and estimates obtained from one known value for each relationship and a complete relationship for one case using the similar‐media scaling concept. The results were compared with measurements on 189 soil cores representing different sites and horizons. With the equations based on soil texture, bulk density and organic matter content, the soil water contents calculated at different matric potentials were generally larger than the measured values. The mean relative error ranged from 8 to 29%, with the standard deviation of errors ranging from 17 to 36%. The model which incorporated one measured soil water content (at −1500 kPa potential) as an additional independent variable, did not improve the results much. The model which incorporated two measured soil water contents (at −33 and −1500 kPa potentials) as additional variables reduced the errors in calculated values considerably. A simple log‐log line drawn through the two known points gave nearly the same accuracy. The estimates from the method of scaling were better than those from the model based on textural and structural variables alone; the mean relative error in the calculated water contents ranged from −0.96 to 9.10%, and the standard deviation of errors was also reduced.