Abstract: Based on the traditional Green-Ampt infiltration model, the diffusion rate D(θ) was introduced to improve the model according to the saline soil infiltration characteristics. Considering the presence of air in the soil, the method of linearizing D(θ) expression was adopted to improve the accuracy of the model parameters. Five types of saline-alkali soils were used to conduct five one-dimensional water infiltration experiments. The data of infiltration rate and wetting front were used to verify the applicability of the improved model, and the result showed that the simulated values of the improved model were in good agreement with the measured values. The average value of the R2 for the five saline-alkali soil infiltration rate was 0.983, and all the average absolute errors for the five saline-alkali soil infiltration rate were less than 0.05 cm/min. The average value of the R2 for the five saline-alkali soil wetting front was 0.868, and all the average absolute errors for the five saline-alkali soil wetting front were less than 3.50 cm. The simulation data of the improved Green-Ampt model were not completely consistent with the measured value, which was due to the combined influence of various factors, the main reason was that the test data was easy to be changed by environmental disturbance. Notably, none of the model data is completely consistent with the experimental data. The experimental data was affected by the disturbance of environmental factors, and were often different from the theoretical results. The experimental data was different from the theoretical results due to the interference of environmental factors. Soil characteristics was the main factors affecting model parameters, and the model parameters of the improved model was affected by K++Na+ content of saline-alkali soil. The correlation between salt content and model parameters was also explored, the result showed that the suction head value decreased with the decrease of the content of K++Na+, and the saturated hydraulic conductivity increased with the decrease of the content of K++Na+. However, the S1 in the five saline-alkali soils did not conform to the above law, which was due to the influence of the texture of its heavy clay. The increase of clay particles in the soil reduces the infiltration performance, the water conductivity of the soil decreased and the pore structure of the soil became more complicated. Guo Green-Ampt model was used to simulate the infiltration process of saline-alkali soil. The results showed that Guo Green-Ampt model had a certain limitation in the simulation of saline-alkali soil infiltration, and the improved model had a better simulation effect than Guo Green-Ampt model, but the two models had different applicable conditions, so there were certain limitations in the comparison. The improved Green-Ampt model can simulate the infiltration process of saline-alkali soil with different levels of salinization, and it can be used in the study of the infiltration model of saline-alkali soil with relatively heavy clay texture. The improved model was still applicable when the ionic composition of saline-alkali soil was different from that of test soil. However, for the saline-alkali soil with different soil texture, especially the soil with very high sand content, the improved model was limited to some extent. This study can provide theoretical support for further study on the mechanism of water infiltration in saline-alkali soil and promotion of local agricultural production.