Skip Navigation LinksHome > May 2001 - Volume 166 - Issue 5 > ESTIMATING WATER RETENTION CHARACTERISTIC FROM SOIL PARTICLE...
Soil Science:
Technical Articles

ESTIMATING WATER RETENTION CHARACTERISTIC FROM SOIL PARTICLE-SIZE DISTRIBUTION USING A NON-SIMILAR MEDIA CONCEPT

Zhuang, Jie1; Jin, Yan1; Miyazaki, Tsuyoshi2

Collapse Box

Abstract

Soil water retention characteristic is an important property of soil. Indirect estimation of this property using easily measurable soil properties remains the research focus of many soil scientists. In this study, a general approach was proposed to derive soil water retention characteristic from soil particle-size distribution and bulk density using a non-similar media concept. Effectiveness of the developed model was tested against measured water retention data of 130 sieved soils with textures ranging from sand to heavy clay. The values of the root mean square error (RMSE) between estimated and experimental pressure heads ranged from 0.4340 to 0.7287 Log cm H2O for all 11 soil textual classes tested in this study. In addition, the new model was compared with the Arya-Paris model, and the results showed that the proposed approach could reduce the overall estimation deviation significantly. The comparison in terms of RMSE values indicated that for 61% of the soils, estimations by the new model were better than those by the best one of the three approaches of the Arya-Paris model, and for about 10% of the soils, the results using the two models were similar. Evidently, the proposed model employs a simple calculation procedure and applies well to soils of all textures except sandy soils.

Soil water retention is a basic soil property necessary for the study of plant-available water, infiltration, drainage, and solute movement. However, the high variability and the complexity of soil make direct determination of the soil water retention property costly, time-consuming, and subject to significant sources of error, especially in the hydrologic survey of a large area. Therefore, an alternative to measurement is to estimate this property indirectly using the more easily available information, such as particle-size distribution, particle density, pore-size distribution, bulk density, mineralogy, and soil morphology (Rawls et al., 1991; van Genuchten and Leij, 1992; Zeiliguer et al., 2000). Numerous attempts have been made to relate particle-size distribution and other properties to soil water retention data (Gupta and Larson, 1979; Rawls et al., 1982; De Jong et al., 1983; Cosby et al., 1984; Saxton et al., 1986; Vereecken et al., 1989; Williams et al., 1992; Schaap et al., 1998). However, most of the methods developed are based on statistical techniques. The applicability and accuracy of those models are limited for various reasons (Rawls et al., 1991; Williams et al., 1992; Tietji and Tapkenhinrichs, 1993; Kern, 1995).

Significant contributions were made by Arya and Paris (1981) to predict water retention curves using particle-size distribution data. They presented a "physico-empirical" approach combining physical hypotheses with empirical representations. This approach is based mainly on the similarity between shapes of particle-size distribution and water retention curves. The Arya-Paris model treats the water flow paths in a soil as a bundle of capillary tubes and assumes that sizes of soil particles are related to corresponding pore diameters of the capillary tubes. The capillary volume is taken to be a function of particle size, mass fraction of the particle size, and a scaling parameter, ∊ (this symbol is equivalent to α in the papers of Arya and Paris (1981) and Arya et al. (1999)). The Arya-Paris model (1981) assumed that the scaling factor ∊ is a constant for all soil textural classes. However, later investigations (Schuh et al., 1988; Mishra et al., 1989; Tyler and Wheatcraft, 1989; Basile and D'Urso, 1997) showed that the average ∊ values varied between fine and coarse textured soils. Recently, Arya et al. (1999) proposed three methods to estimate a variable scaling parameter, represented by ∊i for different particle sizes. Their recent work improved greatly the predictive ability of the original approach with constant ∊. In their study, using the experimental water retention characteristic data of three to six soil samples for each of the five representative soil textures (sand, sandy loam, loam, silt loam, and clay), they obtained statistically the parameters for calculating ∊i for each of the five soil textures. Although it is reasonable to make predictions by means of calibration, adequate measurements of water retention curves prior to model prediction are needed to estimate accurately the statistical parameters for the model.

Many soil physicists have been pursuing simple and efficient methods to describe hydraulic processes of soil in order to avoid laborious and costly measurements. Moreover, since soil heterogeneity is widespread, it is preferable to use soil properties that are measured easily and with a high degree of accuracy to estimate the properties that are difficult to measure and that are more prone to experimental errors. Therefore, it is important to develop a general and more direct approach for deriving water retention characteristic from particle-size distribution data while minimizing the use of empirical and statistical parameters so that prior information on water retention is not required.

In this study, based on a non-similar media concept (NSMC) presented by Miyazaki (1996), we propose a simple analytical model for estimating soil water retention characteristic directly from particle-size distribution, bulk density, particle density, and saturated water content. The NSMC model is then compared with the Arya-Paris model (Arya and Paris, 1981) with the variable scaling parameter, ∊i, calculated by the three methods presented by Arya et al. (1999).

© 2001 Lippincott Williams & Wilkins, Inc.

Login

Article Tools

Share

Search for Similar Articles
You may search for similar articles that contain these same keywords or you may modify the keyword list to augment your search.