Abstract: During atmospheric precipitation, the evaporation of raindrops falling from the bottom of cloud layer to the ground and passing through unsaturated air, a process was called sub-cloud secondary evaporation, which will change the isotopic composition of precipitation. Using the hydrogen and oxygen stable isotope method to understand the temporal and spatial variation of secondary evaporation effect under clouds and its causes is important to understand regional water cycle process. Based on hourly meteorological data of 187 meteorological stations in Shaanxi-Gansu-Ningxia region from March 2018 to February 2019, the spatial and temporal variations of evaporation surplus ratio (f) and precipitation excess deuterium variation (Δd) were analyzed using the improved Ste-wart model, and the relationships between f and meteorological elements and Δd were examined. The results showed that, at the hourly scale, the minimum values of f and Δd in all provinces of the region appeared in the daytime, and the maximum values appeared in the night, indicating that the secondary evaporation effect under the cloud was more obvious in the daytime. At the monthly scale, the monthly variation trend of f and Δd in each province was relatively consistent, with the minimum value appearing in the summer half year, and the maximum value appearing in the winter half year, indicating that the second evaporation effect under cloud was more significant in the summer half year. From the spatial perspective, the spatial variation of f and Δd values in the region was consistent with that at the seasonal scale. In spring, the eastern and western regions were larger while the central part was smaller. In summer, the northwest region was smaller, and other regions were larger. In autumn, it decreased from south to north. In winter, the central and southern regions were smaller, and the western and northeast regions were larger. The spatial differences of secondary evaporation effects under clouds in different seasons was significant. The slopes of the linear relationship between f and Δd in Shaanxi, Gansu and Ningxia provinces were all less than 1‰·%^-1, which may be caused by the arid and semi-arid climate in this area. When air temperature was higher and the relative humidity, vapor pressure, precipitation and raindrop diameter were smaller, the value of Δd was smaller, and the secondary evaporation effect under the cloud was more obvious.

Key words: Shaanxi-Gansu-Ningxia region, precipitation, stable isotopes, sub-cloud secondary evaporation, Stewart model