Abstract:Abstract:Drying-wetting cycle caused by rainfall and evaporation has an obvious influence on soil pore structure, particle surface, and mineral composition. It seriously affects the soil permeability and threatens the ability of soil to resist nutrient loss. This study was carried out to investigate pore-water form of soil during the wetting process and after experiencing different drying-wetting cycles. Eight soil specimens were prepared. Two of them were used to determine the critical value of the water form, three specimens were used for nuclear magnetic resonance (NMR) tests under treatment of different water content, and the other three specimens were used for NMR tests after 0-4 drying-wetting cycles. A saturation-suction method was used to obtain the critical value of relaxation time. The threshold value between adsorbed and movable water was obtained by comparing the relaxation time curves at the saturated state and the critical suction state. The results showed that the critical suction value of adsorbed and movable water was 71.12 MPa and the critical relaxation time was 1.96 ms. The pore water was adsorbed water when the relaxation time was less than 1.96 ms and it was movable water when the relaxation time was not less than 1.96 ms. When soil water content was 10%- 28%, the relaxation time were mostly 0.04-10.72 ms. The relaxation time peaks moved to the right of the curves, which indicated that both forms of water (adsorbed water and the movable water) were increased with increasing water content. When the water content was lower than 19.0%, the proportions of the two forms of water were similar. When the water content was higher than 19.0%, a larger proportion of movable water was observed while the proportion of adsorbed water increased slightly. The relaxation time was mainly 0.03-38.72 ms after 0-4 drying-wetting cycles and the relaxation time curve exceeding 1.96 ms moved to the right as the cycle number increased. After four cycles, the integral area of adsorbed water had a slight fluctuation (about 8%), however, the integral area of movable water increased by about 150%. Drying-wetting cycles enlarged the space among soil particles, and the cycle number had a linear relationship with the movable water and the total water. The movable water, both in the wetting process and after multiple drying-wetting cycles, was mainly in charge of soil permeability because it could lose at a lower hydraulic gradient. The dissolution of solvable cements, the loss of micro-particles, the change of pore wall and the fissures were the factors causing the change of water forms. Based on the Coope model, a simplified model containing only the initial porosity and cycle number was established to estimate the permeability of clay soil that experienced different drying-wetting cycles. Under the condition that the dryingwetting cycles was 0-4 in clay soil, the soil permeability was proportional to the sixth power of the cycle number. The study would provide valuable information for formulating optimal water and fertility schemes in farmland.