A preliminary greenhouse air-conditioning method demonstrates that there is phreatic water evaporation of deeply buried phreatic water in an extremely arid area. Soil-moisture monitoring shows that most daily fluctuation of moisture in the shallow soil layer of 0-60 cm enters deeper layers as water vapor; only 2‰ of this daily fluctuation water enters the atmosphere and evaporates. Monitoring of 0-500 cm soil depths shows that during May through October (the principal rainy season), temperature and absolute humidity in upper soil layers are greater than in lower layers. According to the law of water vapor migration, moisture moves from regions of higher temperature and humidity to those with lower temperature and humidity. Thus, in the aforementioned season within an extremely arid area, conditions may be suitable for rainwater vapor to move downward to the subsoil. This may mean that some rainwater does not evaporate into the atmosphere but penetrates the earth in such areas, and precipitation is likely the main source of deep soil moisture. This suggests there is no phreatic water evaporation of deeply buried phreatic water in extremely arid areas. Based on this hypothesis, a closed greenhouse with air-conditioning system was constructed in the arid area of the Mogao Grottoes in the Gobi Desert. In this greenhouse, we conducted a simulation recycling experiment of 5-mm precipitation via an air-conditioning condensation method. In this way, we first traced where rainwater went. Temperature and humidity were controlled by refrigeration and air-conditioning condensation, which made the temperature and humidity inside the greenhouse approach that of the outside. To understand characteristics of rainfall evaporation in extremely arid areas, the quantity of condensation water was monitored daily. We analyzed temperature and humidity variations, which were monitored by HOBO monitors in the soil, to understand the rainfall infiltration and evaporation process. This confirmed the ultimate flow direction of precipitation, and whether it completely evaporated. This approach was used to determine if there was phreatic water evaporation. The results showed that 5 mm of precipitation, which represented 85% of the frequency of precipitation in this area, completely evaporated. The relative and absolute humidities of the 50 cm above ground within the greenhouse were 12.10% and 3.50 g/m³ greater than outside it, respectively. These results caused the experimental time to be longer than for actual evaporation. In addition, soil temperature and humidity monitoring indicated that after water was sprinkled in the greenhouse, soil temperature, relative humidity and absolute humidity at 30 cm depth were greater by 1.46℃, 4.17% and 2.50 g/m³ than the outside control, respectively. This means that during evaporation, a certain amount of rainfall reached the outside through the greenhouse side soil, under the greenhouse film. This was another reason why the precipitation evaporation collection time was greatly extended. The collection time of 90 days may be much longer than the actual time for natural evaporation, but it confirmed the effectiveness of the rainfall recycling experiment. After complete take-back of all the simulated rainwater, soil moisture did not decline, but increased. The phreatic evaporation speed also increased, which means that precipitation pulsation has a certain control on evaporation of the phreatic water. It was demonstrated that there is phreatic evaporation, and soil water was supplemented by this water. Groundwater-Soil-Plant-Atmosphere Continuum (GSPAC) movement was upward, and the precipitation could completely evaporate.