Relationship between specific surface area and the dry end of the water retention curve for soils with varying clay and organic carbon contents

Accurate description of the soil water retention curve (SWRC) at low water contents is important for simulating water dynamics and biochemical vadose zone processes in arid environments. Soil water retention data corresponding to matric potentials of less than -10 MPa, where adsorptive forces dominate over capillary forces, have also been used to estimate soil specific surface area (SA). In the present study, the dry end of the SWRC was measured with a chilled-mirror dew point psychrometer for 41 Danish soils covering a wide range of clay (CL) and organic carbon (OC) contents. The 41 soils were classified into four groups on the basis of the Dexter number (n = CL/OC), and the Tuller-Or (TO) general scaling model describing water film thickness at a given matric potential (<-10 MPa) was evaluated. The SA estimated from the dry end of the SWRC (SA-SWRC) was in good agreement with the SA measured with ethylene glycol monoethyl ether (SA-EGME) only for organic soils with n > 10. A strong correlation between the ratio of the two surface area estimates and the Dexter number was observed and applied as an additional scaling function in the TO model to rescale the soil water retention curve at low water contents. However, the TO model still overestimated water film thickness at potentials approaching ovendry condition (about -800 MPa). The semi-log linear Campbell-Shiozawa-Rossi- Nimmo (CSRN) model showed better fits for all investigated soils from -10 to -800 MPa and yielded high correlations with CL and SA. It is therefore recommended to apply the empirical CSRN model for predicting the dry part of the water retention curve (-10 to -800 MPa) from measured soil texture or surface area. Further research should aim to modify the more physically based TO model to obtain better descriptions of the SWRC in the very dry range (-300 to -800 MPa).