Models for Estimating In‐Situ Potential Extractable Water using Soil Physical and Chemical Properties

Publisher Website
Google Scholar
Abstract
Accurate evaluation of the upper and lower limits of soil water availability is vital for accurate computation of the water balance in marginally dry regions. Laboratory determined water availability limits are often not available for all depths in a soil profile and may not be sufficiently accurate for some field applications. The objective of this work was to develop empirical equations for calculation of the limits of insitu potential extractable soil water using routinely measured soil properties. Field measurements of the upper and lower limits of extractable water were available for 61 soil profiles from 15 states. For each soil profile, physical and chemical properties were determined on samples collected from depth increments coinciding with measurements of extractable water limits. Physical properties included complete textural characterization, percent coarse fragments, bulk density, organic carbon content, and water content at −15 and −0.33 bar matric potential. Chemical properties included pH, cation exchange capacity (CEC), and percent calcium carbonate.Simple regression analyses were used to determine correlation of individual soil properties with each extractable soil water limit. Properties with the highest simple correlation with the extractable water limits were selected as possible variables to use in multiple regression equations for estimating the extractable water limits. There was a total of 401 observations for the analysis. In order to provide equations for estimating potential extractable water limits where data for only a limited number of soil properties are available, four levels of equations were derived from the data, each level corresponding to a different number of soil properties used as independent variables. Two, four, nine, and ten soil properties were used for the four respective levels. More accurate estimates of the in‐situ potential extractable water limits were obtained by developing separate regression equations for certain textural groupings. Two schemes were used. The first left the 401 observation data set intact. The second scheme grouped the soils data into three texturally based subsets (coarse; moderately coarse and medium; and fine textured) and then equations were developed for each subset. Independent variables in the equations include percentages of clay, silt, fine silt, sand, fine sand, medium sand, and very fine sand; water content by weight held at −15 bar matric potential, CEC, and an estimate of the percentage of particles passing through a no. 200 sieve. The potential extractable water calculated as the difference between the upper and lower limits based upon the equations developed was correlated with measured plant‐extractable water (differences between the field‐measured upper and lower limits) for the different levels of input information and textural separations. The coefficients of determination for potential extractable water ranged from 0.375 for the two‐variable input level with no textural separation to 0.758 for the 10‐input level with three textural separations. Standard deviation for the latter, best correlated system to estimate potential extractable water was about ± 2 percentage units when compared with field‐measured extractable water.