[Objective] In recent years, with frequent extreme climate events, the northwestern region of China has experienced large diurnal temperature difference in summer, with daytime high temperatures continuously rising and persisting for extended periods. The resulting water vapor migration significantly affects the subgrade moisture content, thereby influencing the engineering performance of the subgrade. Investigating the water vapor migration patterns in loess subgrades under sealed pavement structures subjected to high temperatures holds substantial practical significance and application value for scientifically predicting the engineering performance of loess subgrades and ensuring their long-term stability. [Methods] A self-developed one-dimensional soil column test apparatus was employed, with compacted loess columns from the suburbs of Xi’an as the study subjects. The MTD15 temperature and moisture sensors were utilized to monitor the temperature and moisture variations within the loess columns under a boundary heating condition of 55 ℃. The water vapor migration patterns in one-dimensional sealed soil columns were analyzed under two heating modes: diurnal temperature cycling and continuous heating. Furthermore, the effects of different heating modes on the temperature and moisture distribution characteristics of the soil columns were explored. [Results] Boundary heating caused the temperature of the soil columns to rise, with a faster increase in the upper part and a slower increase at the bottom. During the heating phase, the temperature distribution along the height exhibited an approximately linear pattern. Under the cyclic heating of diurnal temperature difference, the internal temperature of the soil columns dropped rapidly after the heating was stopped. The cooling rate in the upper part was significantly higher than that in the middle and lower parts. By 08:00 the next day, the soil column temperature ranged between 22-25 ℃, with the middle part slightly warmer than the two ends. At 20:00 each day and 8:00 the following day, the temperature distribution along the depth of the soil columns remained basically the same. Under continuous heating, the soil column temperature reached dynamic equilibrium after 10 days of heating, exhibiting a two-segment, piecewise linear distribution. The variation trends in the moisture content distribution curves of the soil columns under the two heating methods were basically the same—namely, a decrease in moisture content in the upper part, an increase in the middle, and a decrease in the lower part, with these trends becoming more pronounced as the experiment progressed. However, within the upper 22.5 cm of the soil columns, the moisture content under the cyclic heating of diurnal temperature difference was higher than that under continuous high-temperature heating. [Conclusion] Under boundary heating conditions, the moisture of sealed soil columns is primarily governed by the combined effects of liquid water vaporization and moisture migration within the pores, water vapor migration driven by vapor concentration gradients, and condensation of vapor in the pores when the vapor pressure exceeds the saturated vapor pressure. These mechanisms collectively result in an inverse “S”-shaped moisture distribution. For subgrades with cover layers, 30 days of cyclic heating of diurnal temperature difference during summer induces moisture at 5 cm depth to fluctuate between -2% and +2% of the initial moisture content. This cyclic fluctuation of the upper subgrade moisture induced by diurnal temperature difference will significantly affect the mechanical properties of the subgrade. Under extreme climate conditions characterized by continuously rising heating temperatures and prolonged duration, the moisture content variations in the subgrade become more pronounced. These significant variations in moisture will lead to dehydration, shrinkage, and cracking in the upper subgrade layers, thereby compromising the service performance and lifespan of the subgrade and pavement structures.