Spatial variability of some soil shrinkage indices in hilly calcareous region of western Iran

• High variability of soil shrinkage indices mainly ascribed to high variability of soil properties.
• %ess had negative spatial correlations with those in %eps and %ers.
• Spatial variability of shrinkage indices was affected by particle sizes and carbonate.
• RK was more successful than OK, for spatial prediction of some shrinkage indices.

This study was conducted as a first one to explore the spatial variability of some soil shrinkage indices and their spatial correlations with soil chemical and physical properties in calcareous soils in western Iran. Ninety nine undisturbed clods were collected from 0 to 10 cm soil depth in hilly calcareous regions of Cherlgerd watershed in western Iran. Soil shrinkage curve was measured, according to Archimedes’ principle, by covering the undisturbed clods with an acrylic resin which was slowly permeable to water vapor and impermeable to liquid water. Then, void ratio (e) vs. moisture ratio (ϑ) pairs were plotted and the shrinkage curve data were modeled using sigmoidal Peng and Horn (2005) model. Derivatives of Peng and Horn (2005) model were solved to discriminate different shrinkage zones and to derive shrinkage indices related to the different zones. Geostatistical technique was employed to evaluate spatial variability of soil shrinkage indices including relative void ratio changes, mean slopes at various shrinkage zones, COLE index, and total and relative shrinkage capacities. Spherical and exponential models were successfully fitted to experimental semi-variograms. The studied shrinkage indices showed moderate to strong spatial dependency in the studied area. Results of co-regionalization analysis showed that relative void ratio change in the proportional zone had inverse spatial pattern with relative void ratio change in the residual and zero shrinkage zones. Also, spatial pattern of shrinkage indices had superior agreement with spatial variability of some soil properties (i.e., particle size distribution and calcium carbonate equivalent) which predominantly affected soil shrinkage behavior in the study area. Comparing regression kriging (RK) using environmental variables with ordinary kriging (OK) showed that RK could improve the spatial prediction of relative shrinkages in the structural, proportional and zero shrinkage zones, and total and relative shrinkage capacities.