Predictive models for the residual saturation zone of the soil–water characteristic curve

Abstract

Purpose

Soil–water characteristic curve (SWCC) models have generally been developed based on measurement results in the low suction range. However, different mechanisms govern the SWCC in the low suction range (desaturation zone) and high suction range (residual zone). Accurate estimation of the residual zone is critical for analysing the behaviour of soils in arid regions. This study aimed to develop prediction models for the residual zone based on selected properties of soils.

Materials and methods

To achieve the abovementioned purpose, 162 total suction (in pF) and gravimetric water content (%) measurement pairs in the high suction range were made on 40 cohesive soil samples with known physical, chemical, and spectral properties. A semi-logarithmic linear model was used to define the residual saturation zone of the SWCC. The model parameters were the slope of the SWCC (${s}_{r}$) and total suction at zero water content (${\psi }_{dry}$). Correlation and stepwise regression analyses were carried out between the model parameters and selected soil properties. The regression equations were validated using the four-fold cross-validation procedure. Water content (%) estimation models were developed using combinations of different regression equations for ${s}_{r}$ and ${\psi }_{dry}$, and their estimates were evaluated using performance metrics.

Results and discussion

The ${s}_{r}$ values for the soils studied ranged from 1.589 to 13.035, with an average of 6.007. Although some studies have shown strong correlations between clay content and ${s}_{r}$, no significant relationship was found between clay content and ${s}_{r}$ for the soils in this study. However, significant correlations were found between the consistency limits, some spectral parameters, and ${s}_{r}$. The R² was 0.88 when the liquid limit (LL) and depth of the 1900 nm wavelength band (D₁₉₀₀) values were used as descriptive variables. The ${\psi }_{dry}$ values for the soils studied ranged from 6.483 to 7.370 pF, with an average of 6.855 pF. The relationships between the selected soil properties and ${\psi }_{dry}$ were weak, consistent with previous research.

Conclusion

The spectral absorption characteristics of the soils in this study had a high potential for estimating the SWCC. Prediction models based on various ${s}_{r}$ and ${\psi }_{dry}$ equation combinations could predict measured water content values with varying degrees of accuracy. The SWCC’s residual saturation zone was accurately estimated using soil properties such as liquid limit, electrical conductivity, and spectral characteristics that can be determined quickly and inexpensively.