Abstract Estimation of soil water content (SWC) in deep soil profiles is of crucial importance for strategic management of water resource for sustainable land use in arid and semi-arid zones, as well as for soil and water conservation. Soil properties have a very important effect on SWC. This study aimed to analyze the influence of soil particle size on SWC, for the first time using soil particle size to estimate SWC in deep soil profiles. SWC was measured mainly in farmland, natural grasslands and plantations of Caragana from the surface to more than 20?m depth. The same soil samples were also tested for particle size. The results show that the soil desiccation is formed in the caragana forest in 3–18?m soil layers, but almost no formation in 18–24?m layers; water content of farmland and grassland is different in all soil profiles although they are both shallow rooted plants. Correlation analysis indicated that SWC could be well predicted by clay content and the close correlation between SWC and clay content yielded a coefficient of determination ( R 2) of 0.82 and 0.72, respectively, for farmland and grassland. After multiple regression analysis, a regression model was built using SWC, clay content and sand content data, giving R 2=0.66. The model provided reliable estimates of SWC profile based on textural class. This can assist in estimating water depletion by vegetation, by comparing moisture of farmland and grassland soils with that of plantation forests, and in selecting sustainable land use of arid land. Keywords Clay content ; Field capacity ; Sand content ; Soil water content (SWC) ; Soil particle size prs.rt("abs_end"); 1. Introduction Soil water content (SWC) is used to calculate “the available water storage capacity, which is defined as the moisture held between field capacity (FC) and permanent wilting point (PWP)” and is critical for practical application which is related to agricultural, water and soil resources management ( Rao (1998) and Starks, Heathman, Ahujab, & Ma (2003) ). It is also a critical factor in evaluating the suitability of the given vegetation in that region. The transmissivity parameters (e.g., soil hydraulic conductivity vs. SWC relationships) used in physically-based models that make basic assumptions of soil uniformity and homogeneity, are also highly sensitive to SWC ( Givia, Prasher, & Patel, 2004 ). However, direct measurement of SWC is usually hard, expensive and time-consuming for most researches and management applications, especially on a relatively large scale. When the researched area is relatively homogeneous in its physical soil makeup and topography, SWC is related to other physical characteristics such as particle size distribution, structure, bulk density and organic matter content ( Rao, 1998 ). It is possible to develop empirical relationships that provide adequate estimates of SWC through numbers of sampling sites which are inexpensive and easy to access. Most methods are named pedotransfer functions (PTFs) ( Bouma & Van Lanen, 1987 ). Correlations between soil properties (SWC, organic carbon content, and percentage of sand, silt, clay etc.) have been studied since early in the twentieth century ( Briggs & Shantz (1912) , Salter & Williams (1965) and Doorenbos and Pruitt, 1977 ). With the development of computer modeling and databases, more PTFs have been developed ( Rawls, Gish, & Brakensiek, 1991 ; W?sten, Pachepsky, & Rawls, 2001 ). Rawls, Brakensiek, and Saxton (1982) developed PTFs using 5350 sets of soil data. Baumer (1992) developed PTFs using 18 000 soil horizon measurements from the US National Soil Pedon Characterization database to predict SWCs at FC and wilting point (WP). W?sten, Lilly, Nemes, and Le Bas (1999) proposed PTFs based on the HYPRES database that contains 5521 sets of soil data. Bruand, Perez Fernandez, and Duval (2003) formulated PTFs use particle size and bulk density to calculate gravimetric water content at 7 water potentials. Furthermore, the accuracy of PTFs in predicting the SWC has been evaluated. Givia et al. (2004) showed that the PTFs developed for soils having similar characteristics to those being studied generally perform better than others. Cornelis, Ronsyn, Van Meirvenne, and Hartmann (2001) presented that a PTF performs much better if it is used to the developed region. However, concrete measurement of the required soil characteristics is not practicable and the present PTFs are most be developed for estimating water retention and available water content in surface soil ( Schaap, Nemes, & Van Genuchten, 2004 ). Currently, we can only get a crude spatial distribution of soil textural composition by field survey ( Starks et al., 2003 ). Particle-size composition could be related to FC, WP, and available water content via regression equation ( Pidgeon, 2006 ). The relationships predicting SWC have been developed from those found in different countries, including England ( Pidgeon, 2006 ), USA
展开▼