Elsevier

Geoderma

Volume 137, Issues 1–2, 31 December 2006, Pages 179-191
Geoderma

Measuring and modelling the soil shrinkage characteristic curve

https://doi.org/10.1016/j.geoderma.2006.08.022Get rights and content

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.

Introduction

Swelling and shrinking clay soils change in volume with water content changes. These volume changes depend on the amount and type of clay minerals and are characterized by their magnitude and geometry. They result in the occurrence of shrinkage cracks and surface subsidence. Assessing the magnitude and geometry of cracks is important for modelling infiltration at the soil surface and subsequent redistribution of water within the soil. Both processes are different compared to non-shrinking soil due to surface runoff and preferential flow in the cracks. The latter process is known as bypass flow or short circuiting (Hoogmoed and Bouma, 1980). Modelling of water transport in the vadose zone of the soil hence not only requires the soil–water characteristic curve and hydraulic conductivity, but also a third relationship, being the soil shrinkage characteristic curve (SSCC), which relates a water content related variable to a pore volume related variable.

Generally, when a swelling and shrinking clay soil dries out, four shrinkage stages can be distinguished (Haines, 1923, Stirk, 1954, Bronswijk, 1991): (1) structural shrinkage, (2) normal shrinkage, (3) residual shrinkage and (4) zero shrinkage (see Fig. 1). In the first stage, the large inter-aggregate pores and the biological tubular pores – worm and root channels – are emptied without considerable change in bulk volume, and air enters these relatively large pores. This stage only occurs in structured well-aggregated soils or soils with considerable biological activity. In the second stage, the decrease in water volume results in an equal decrease in bulk soil volume, with the intra-aggregate pores still being fully saturated. In the case of a structureless clay paste, the slope is equal to 1 (Sposito and Giráldez, 1976, Chertkov, 2000, Chertkov, 2003), whereas in structured soils, the slope can be much smaller than 1 and as low as 0.1 (Braudeau et al., 1999). Mitchell (1992) therefore suggested calling this stage basic shrinkage rather than normal shrinkage, whereas Groenevelt and Grant (2001) used the term proportional. In the third stage, air enters the intra-aggregate pores and a further decrease in water upon drying hence exceeds the volume change of the aggregates (and the bulk soil). Finally, in the fourth stage, the soil particles have reached their densest configuration and the volume of the aggregates remains unaltered as the water volume further decreases (Bronswijk, 1991). Bruand and Prost (1987) demonstrated, however, that during this fourth phase, reorganisation of clay particles does occur, leading to the formation of microscopic cracks. Since both phenomena have a compensating effect, the aggregate volume remains unaltered (Bruand and Prost, 1987). They further observed during stages 2 and 3, shrinkage of the millimetric domains which build up the soil aggregates. This shrinkage results in millimetric cracks which enclose those millimetric domains. As the number of millimetric domains increases with aggregate size, larger aggregates show a higher volume of voids relative to the volume of solids (Bruand and Prost, 1987) and the shrinkage behaviour of the sample becomes less pronounced.

Determination of the SSCC requires simultaneous measurement of the pore volume and the volume of water in a known volume of soil over the whole range of water contents, from saturation till oven dryness. Numerous techniques for determining the SSCC have been proposed. Many authors suggested determining the bulk volume of soil by measuring the weight or volume of a fluid displaced by variably saturated soil samples using Archimedes' principle. Soil samples were submerged in fluids such as kerosene or petroleum (McIntyre and Stirk, 1954, Monnier et al., 1973), toluene (Sibley and Williams, 1989) and mercury (ASTM, 2005), or were first coated with paraffin (Lauritzen and Stewart, 1941, Johnston and Hill, 1944, Lauritzen, 1948) or saran resin dissolved in methyl ethyl ketone (Brasher et al., 1966). Pellissier (1991) used a combination method in which toluene was used as the submergible fluid for the wet part of the SSCC, and water for the dry end when air enters into the soil. Before submerging the clods in water, they were dipped in molten wax, and at further drying, the wax was removed and the clod was sprayed with a freezing liquid.

Another approach for measuring the bulk soil volume is by measuring the dimensions of the soil sample directly. Berndt and Coughlan (1976) recorded the height and diameter of undisturbed soil cores as they dried out. A similar procedure was followed by Yule and Ritchie, 1980a, Yule and Ritchie, 1980b. Schafer and Singer (1976) filled columns with disturbed soil and followed the decrease in length of the drying rod-shaped soil column. Braudeau (1987) measured the reduction in sample dimensions upon drying by using a retractometer, which was later modified by Braudeau and Boivin (1995) and Braudeau et al. (1999). It allows continuous monitoring of the diameter of a cylindrical soil sample in the vertical direction using several laser sensors.

Michel et al. (2000) adapted a triaxial apparatus to allow the tracing of the isotropic character of shrinking soil. The limitations associated with many of these methods lead Tariq and Durnford (1993a) to design an alternative and very simple method to determine the SSCC. Disturbed or undisturbed soil samples were saturated and surrounded by an ordinary rubber balloon. The sample was dried by air flowing at low pressure over the sample. At given times, the soil sample and the balloon were submerged in water and the bulk soil volume was then determined by the volume of water which it displaced. All the above methods allow soil water to be related to pore volume.

To model water and solute transport in the soil, a continuous SSCC is required, rather than a set of discrete experimental data pairs that can be obtained experimentally using the approaches listed previously. Several models that can be fitted to a set of discrete data pairs are reported in literature and include polynomial models (Giráldez et al., 1983, Giráldez and Sposito, 1983), linear models consisting of different straight lines for the different shrinkage phases (McGarry and Malafant, 1987), logistic models (McGarry and Malafant, 1987), and sigmoid models (Groenevelt and Grant, 2001, Groenevelt and Grant, 2002, Cornelis et al., 2006). Kim et al. (1992), Tariq and Durnford (1993b) and Braudeau et al. (1999) suggested combining exponential or polynomial function with linear ones. These models all require greater verification using data sets independent from those applied in previous studies. Since those models have been previously tested using different methods, their evaluation using one single method is highly recommended.

Besides the magnitude of volume changes upon wetting and drying, which is described by the SSCC, the geometry of swelling and shrinking is of equally importance for modelling water transport in a swelling and shrinking soil. When representing a given soil volume by an isolated cube (Bronswijk, 1990, Chertkov et al., 2004, Chertkov, 2005), changes in volume can be described as:1ΔVV=(1Δzz)rswhere V is the original soil bulk volume, ΔV is the volume change upon shrinkage, z is the original height, Δz is the surface subsidence, and rs is the geometry factor first defined by Rijniersce (1983) in a study on pedogenitically unripe soils. For three dimensional isotropic shrinkage, rs is equal to 3. When cracking dominates subsidence, rs > 3, and in the case of subsidence only, rs is 1. Modelling of changes in soil bulk volume, crack area and surface subsidence as a function of the void ratio was demonstrated in Cornelis et al. (2006).

The objectives of this paper were

  • (1)

    to compare three methods to determine the volume change of the soil matrix, i.e. a soil volume without cracks, upon shrinkage. These methods were the core method of Berndt and Coughlan (1976) on undisturbed soil cores, the balloon method of Tariq and Durnford (1993a) in which disturbed soil samples were applied, and the paraffin-coated method of Lauritzen and Stewart (1941) for individual soil clods;

  • (2)

    to test the performance of the models of Giráldez et al. (1983), McGarry and Malafant (1987), Kim et al. (1992), Tariq and Durnford (1993b), Olsen and Haugen (1998), Braudeau et al. (1999), a modified version of the Chertkov, 2000, Chertkov, 2003, and the simplified notation of the Groenevelt and Grant, 2001, Groenevelt and Grant, 2002 model as suggested by Cornelis et al. (2006).

The above methods and models were tested on soil samples collected on a Vertisol and a Lixisol which were under sugar cane in the Havana province, Cuba. Before describing these soils and the followed methodology, the selected models will be first described in more detail.

Section snippets

Some considerations about the selected SSCC methods

In comparing the methods to determine the SSCC, the paraffin-coated method of Lauritzen and Stewart (1941) is considered in our study as a reference. This method was selected because clods can be obtained by breaking soil after it has been subjected to shrinkage. As such, the method represents the shrinkage behaviour of the soil matrix, rather than the bulk soil with cracks, relatively well. This allows modelling changes in crack area and surface subsidence (outside the soil matrix) by

Description of the selected SSCC models

It should be noted first that all models described here are expressed in terms of the void ratio as a function of the moisture ratio, which are respectively the volume of voids and the volume of water over the volume of solids. We therefore had to rewrite some of the models compared to their original notation reported in literature.

Soil sampling and soil properties

The study was based on soil samples taken from seven horizons of a Eutric Vertisol and a Ferri-Gleyic Lixisol (WRB, 1998) under sugar cane in the Havana province, Cuba. In the Vertisol, A and B horizons were distinguished within the upper 1.5 m of the soil profile. The B horizon was further divided into three subhorizons (B1, B2 and B3). The A horizon had a blocky structure and showed a strong vertical shrinkage behaviour, though there was also evidence of less pronounced horizontal shrinkage.

Assessment of the SSCC

In Fig. 3, the SSCCs determined using the core method of Berndt and Coughlan (1976), the balloon method of Tariq and Durnford (1993a) and the paraffin-coated method of Lauritzen and Stewart (1941) are compared for the seven horizons considered in this study. Larger void ratios indicating less shrinkage can be observed when comparing the core method with the balloon and paraffin-coated method, except for water contents near saturation. This was also observed by Crescimanno and Provenzano (1999)

Summary and conclusions

This study showed significant differences in the SSCC measured on undisturbed samples using the core method (Berndt and Coughlan, 1976) on one hand and on disturbed samples using the balloon method (Tariq and Durnford, 1993a) and soil clods using the paraffin-coated method (Lauritzen and Stewart, 1941) on the other hand. All samples and clods were taken from seven horizons of a Vertisol and a Lixisol which were under sugar cane in the Havana province, Cuba. The paraffin-coated method was

Acknowledgements

The research work was conducted in the framework of the project “Improving soil salinity management under sugar cane using Geoinformatica” which was funded by the Flemish Interuniversity Council, Belgium, to which we are greatly indebted. The helpful comments and suggestions of Ary Bruand and an anonymous reviewer are also greatly acknowledged.

References (57)

  • E. Braudeau et al.

    New device and method for soil shrinkage curve measurement and characterization

    Soil Sci. Soc. Am. J.

    (1999)
  • J.J.B. Bronswijk

    Shrinkage geometry of a heavy clay soil at various stresses

    Soil Sci. Soc. Am. J.

    (1990)
  • J.J.B. Bronswijk

    Relation between vertical soil movements and water-content changes in cracking clays

    Soil Sci. Soc. Am. J.

    (1991)
  • A. Bruand et al.

    Effect of water content on the fabric of a soil material: an experimental approach

    J. Soil Sci.

    (1987)
  • Y.-M. Cabidoche et al.

    Non-uniform volume and water content changes in swelling clay soil: II. A field study on a Vertisol

    Eur. J. Soil Sci.

    (1995)
  • V.Y. Chertkov

    The shrinkage geometry factor of a soil layer

    Soil Sci. Soc. Am. J.

    (2005)
  • V.Y. Chertkov et al.

    Modeling the crack network of swelling clay soils

    Soil Sci. Soc. Am. J.

    (1998)
  • V.Y. Chertkov et al.

    An approach for estimating the shrinkage geometry factor at a moisture content

    Soil Sci. Soc. Am. J.

    (2004)
  • Corluy, J., 2001. Studie van de waterbalans in verzilte bodems van een suikerrietplantage in Cuba. M.Sc. thesis, Ghent...
  • W.M. Cornelis et al.

    A simplified parametric model to describe the magnitude and geometry of soil shrinkage

    Eur. J. Soil Sci.

    (2006)
  • G. Crescimanno et al.

    Soil shrinkage characteristic curve in clay soils: measurement and prediction

    Soil Sci. Soc. Am. J.

    (1999)
  • J.V. Giráldez et al.

    A general soil volume change equation: II. Effect of load pressure

    Soil Sci. Soc. Am. J.

    (1983)
  • J.V. Giráldez et al.

    A general soil volume change equation: I. The two-parameter model

    Soil Sci. Soc. Am. J.

    (1983)
  • P.H. Groenevelt et al.

    Water retention in soil

    Soil Sci.

    (1972)
  • P.H. Groenevelt et al.

    Re-evaluation of the structural properties of some British swelling soils

    Eur. J. Soil Sci.

    (2001)
  • R.B. Grossman et al.

    Linear extensibility as calculated from natural-clod bulk density measurements

    Soil Sci. Soc. Am. Proc.

    (1968)
  • W.B. Haines

    The volume changes associated with variations of water content in soil

    J. Agric. Sci. Cambridge

    (1923)
  • W.B. Hoogmoed et al.

    A simulation-model for predicting infiltration into cracked clay soil

    Soil Sci. Soc. Am. J.

    (1980)
  • Cited by (128)

    • Earthen-based building: In-situ drying kinetics and shrinkage

      2023, Construction and Building Materials
    View all citing articles on Scopus
    View full text