Soil desiccation poses a critical bottleneck constraining the sustainable development of the ecological environment in semi-arid loess hilly regions, where natural precipitation serves as the sole replenishment source for soil moisture. Desiccation of deep soil layers induced by artificial forest and grassland vegetation on the Loess Plateau has shown a persistent intensification trend. Particularly since the large-scale implementation of the Grain for Green Project in 1999, the severity of regional soil water deficits has been significantly exacerbated. Compounded by the region's thick loess deposits ranging from tens to over a hundred meters, the deep groundwater cannot replenish shallow soil through capillary rise, ultimately leading to the formation of permanent dry soil layers. Current research on soil desiccation remediation remains inadequate, predominantly focusing on surface-level interventions such as various mulching measures, bare land treatments, and cultivation of annual shallow-rooted crops. However, there are limited reports addressing the cultivation of perennial deep-rooted vegetation systems under such deep soil desiccation conditions. To explore the sensitivity of soil moisture to rainfall after replanting deep-rooted and perennial plants under the background of deep soil desiccation in loess hills, in this study, the CS650 soil moisture probewas used to continuously monitor the soil moisture from 0 to 600 cm soil in 2022?—2024, and the analysis results showed that: 1) The ratio of water infiltration into the soil to the soil in the loess hilly area increased with rainfall, light rain was basically ineffective for soil moisture storage below 10 cm, and rainfall above moderate rain had a significant effect on the water storage below 10 cm. The ratio of water infiltrated into the soil by moderate heavy, and torrential rain can be increased from 36% to 61%. 2) Under the background of deep soil drying, the soil moisture in the first year of replanting alfalfa was mainly deposited, with savings mainly concentrated in the 0—340 cm soil layer. In the second and third years of alfalfa planting, soil moisture consumption was the main one, accounting for 13.1% and 50.5% of the rainfall, respectively, and the water storage consumption of the 0—340 cm soil layer was 45.4 mm and 185.3 mm, respectively. 3) The water consumption in the first year of alfalfa planting was about 31.1% of the rainfall of the current year, the water consumption in the second and third years of planting exceeded the rainfall of the current year, and there was no significant difference in the yield and water use efficiency of alfalfa in the first and second years of alfalfa planting compared with the control (P>0.05), and the yield in the third year was significantly lower than that of the control (P<0.05), but the difference in water use efficiency was not significant (P>0.05). This study can provide a scientific basis for the sustainable replacement of artificial vegetation and the rational utilization of water resources in the loess hilly area.