Shrink–swell behavior of soil across a Vertisol catena

SummaryAn accurate estimate of the degree of soil cracking is important in partitioning rainfall into soil infiltration and runoff in watersheds with clay soils that shrink and swell. Usually in the application of surface hydrology models, cracking is considered to be predicted by assuming equidimensional shrinkage and a 1:3 ratio of change in soil profile thickness to depth of water loss. Our objective was to use in situ measurements of soil profile subsidence with loss of water in an apparently uniform Vertisol to determine: (1) whether or not the soil profile subsidence followed the 1:3 ratio of change in thickness to depth of water loss and (2) how the ratio of subsidence to water loss varies with the Coefficient of Linear Extensibility (COLE). The research was conducted on a catena of Houston Black and Heiden clays. Vertical soil shrinkage and swelling along with soil water content were measured at the summit, shoulder, backslope, and footslope positions of the catena. Change in soil water content along with spatial variability in COLE were the primary drivers of temporal and spatial variability of shrinkage and swelling on the catena. The rate that soil shrank or swelled with change in the amount of water stored in the soil was related to COLE, which was negatively correlated with carbonate content of the soil. Shrinkage of the soil profile with water lost was less than the commonly used 1:3 ratio. Our data supported using COLE to predict the degree to which the ratio is less than 1:3. Incorporating COLE values into a hydrology model to simulate soil shrinkage with water loss is a useful approach because COLE values have been measured for soils with shrink–swell potential and that data is available in USDA NRCS Soil Survey databases.

► Soil subsidence across outwardly uniform soils varied spatially and temporally.
► The coefficient of linear extensibility estimated soil subsidence from water loss.
► Shrinkage potential of a Vertisol formed from marl is predicted by carbonate content.
► Hydrology models can include soil crack formation using published COLE data.