Measuring and modelling the soil shrinkage characteristic curve

Abstract The macroporosity, and to a lesser extent the microporosity, of swelling and shrinking soils is affected by their shrinkage behaviour. The magnitude of the changes in bulk volume in response to changes in water content is usually described by the soil shrinkage characteristic curve (SSCC), i.e. the relation between the void ratio and the moisture ratio. At present, many techniques have been described for determination of the SSCC. We have applied the core method, the rubber-balloon method and the paraffin-coated method on respectively undisturbed soil samples, disturbed soil samples and soil clods collected from seven horizons of a Vertisol and a Lixisol under sugar cane in the Havana province, Cuba. We demonstrated that the balloon and paraffin-coated method showed similar results, whereas the core method produced less pronounced shrinkage. The latter was due to the anisotropic shrinkage as was confirmed by the change of the geometry factor with the moisture ratio, to a possible reorientation of particles when collecting undisturbed soil cores, and to the occurrence of small cracks upon drying. We have further shown that the core method produced much higher scatter, which was explained by higher measuring errors and crumbling of the samples as they dried out. Because of its superior behaviour, the balloon method was then selected to test nine different parametric models that describe the SSCC. A group of four models which performed best in terms of RMSE, coefficient of determination and Akaike Information Criterion could be distinguished. These models include the three linear equations model of McGarry and Malafant [McGarry, D., Malafant, K.W.J., 1987. The analysis of volume change in unconfined units of soil. Soil Sci. Soc. Am. J. 51, 290–297], the combined linear and exponential five equations model of Braudeau et al. [Braudeau, E., Costantini, J.M., Bellier, G., Colleuille, H., 1999. New device and method for soil shrinkage curve measurement and characterization. Soil Sci. Soc. Am. J. 63, 525–535], a modified version of the theoretical three equations model of Chertkov [Chertkov, V.Y., 2000. Modeling the pore structure and shrinkage curve of soil clay matrix. Geoderma 95, 215–246] and a simplified version of the logistic model of Groenevelt and Grant [Groenevelt, P.H., Grant, C.D., 2001. Re-evaluation of the structural properties of some British swelling soils. Eur. J. Soil Sci. 52, 469–477]. Though performing very well, the McGarry and Malafant model does not describe the complete SSCC, whereas the Braudeau et al. model contains a relatively large number of parameters. Overall highest performance was observed for the modified Chertkov model. The modified Groenevelt and Grant model, however, has the advantage of being the most elegant as it consists of only one single equation.