Abstract: Soil shrinkage and cracking are essential behavior during dehydration in the environmental geotechnical engineering of farmland. Their quantification can be necessary to determine the soil physical-hydraulic parameters and crack preferential flow in soils. In the present study, a new predicting model was proposed for the crack ratio with respect to soil water content using a VG-based soil shrinkage characteristic curve and soil shrinkage anisotropy factor. The migration and transition of pores were also considered in the solid-liquid-gas phase system in soil matrix-subsidence-crack domains. This model included a VG-based soil shrinkage characteristic model, a shrinkage anisotropy model using a Logistic curve, and the soil cracking ratio model. An indoor experiment was conducted to investigate the soil shrinkage characteristics and cracking behavior. The crack ratio was determined using image processing techniques and morphological features. The experimental data was used to evaluate the fitting of a model for cracking ratio evolution. The results showed that the VG-based shrinkage model (VG-PENG model), three fitting parameters, and two estimated parameters, well predicted the soil shrinkage characteristics in various types of soils (R2>0.97, RMSE<0.04). The Logistic model was first introduced into the expression of soil shrinkage geometry, which was previously used to describe the growth principles under limited resources. Quantification of soil shrinkage anisotropy showed that the soil shrinkage was highly anisotropic. The soil shrinkage exhibited only subsidence with shrinkage geometric factor approximately equivalent to 1 in the early phase of the soil dehydration. The shrinkage showed mainly vertical subsidence with a light horizontal shrinkage (or cracking) in the middle phase, with shrinkage geometric factor varying between 1 and 3. The shrinkage geometry factor tended to be stabilized in the late phase, indicating a residual state of shrinkage. The anisotropic shrinkage with rapid change occurred in the relative water content of 0.3-0.7. The logistic shrinkage anisotropy model well predicted the shrinkage geometric factor with respect to the water content. A new model was also proposed to predict the evolution of crack ratio with respect to water content. The curve of the model showed sigmoid characteristics, depending highly on shrinkage properties and anisotropy. The simulated data showed better agreement with the experimental one, indicating an extremely significant level (R2=0.974, P<0.001). Since the water content within the soil layer was assumed evenly distributed, this model was considered to be appropriate in a relatively limited height of the soil layer. The evolution of crack ratio was predicted from a perspective view of soil physics rather than a mechanical view. Consequently, the soil shrinkage anisotropy was fully integrated into the modelling of the cracking ratio. A significant innovation was also made to apply the VG-type shrinkage characteristic curve to crack ratio prediction. The crack porosity prediction belonged to the field of soil physics to describe the evolution of cracking ratio using the shrinkage curve and geometry factor. The proposed model well predicted the cracking ratio in surface soils with high accuracy and convenience. A further investigation was also needed to explore the efficacy of the crack ratio model on undisturbed soils, and the effects of substrate properties on cracking behaviors. The research can provide a promising theoretical basis and parameter prediction for soil water-solute movement in soils with variable-solid phase and preferential flow in soil physics and hydrology.