Tomato pot experiment was used to arrange straw capillary barrier at 17cm below the soil surface, which effectively cut the continuity of soil capillary. Combined with the leaching effect of irrigation water, the soil salt moved downward, and the potted soil tended to be in a non-uniform distribution state in the vertical direction, which was lower in the upper layer and higher in the deep layer, and the internal relationship between tomato yield and root system and salt distribution characteristics at different growth stages was explored. The results showed that in the treatment of vertical nonuniform distribution of soil salt, the excessive salt content in the soil below the straw layer inhibited the growth and distribution of tomato roots, while the roots in the soil above straw layer with low salinity showed compensatory growth. When the salt content of potted soil was the same, the root dry matter mass and fruit yield of salt vertical non-uniform distribution treatment were significantly higher than those of salt uniform distribution treatment (P<0.05). Differential distribution of salt content, which showed low salinity in the upper layer and higher in the deep layer, alleviated the salt stress of whole crops. Based on the traditional water production function, i.e., Stewart model and Jensen model, the crop salt production function was constructed by using surface salt factor, average salt factor and root weighted average salt factor respectively. After the function accuracy evaluation, it was found that the crop phased production function constructed by root weighted average salt factor combined with Jensen function had the highest prediction accuracy for tomato yield, while the soil salt content and root distribution in tomato fruit setting stage had the greatest impact on the final yield in different growth stages.