Variation and Transformation of Evapotranspiration at Different Scales in a Desert Steppe

Abstract

Water resources are severely scarce in desert steppes, and precipitation rarely collects in rivers or is transformed into groundwater. Evapotranspiration (ET) is the primary “export” of precipitation conversion and is the main mechanism for water vapor exchange between the underlying surface and atmosphere. ET changes have certain scale effects. This study focused on the natural grasslands in the Xilamuren Desert Steppe and analyzed and estimated the ET patterns at different scales, including micro-, point-, and surface scales, using observational data from instruments such as a photosynthetic meter, Eddy-covariance system (EC), and large-aperture scintillometer (LAS) from the Ecological Hydrology National Field Science Observation Station in the northern foothills of the Yin Mountains, Inner Mongolia. The spatial scale was extended based on this analysis. The results showed that at the microscale, the diurnal variation in the photosynthetic and transpiration rates of Leymus chinensis followed a bimodal curve. In July and August (high-temperature months), photosynthetic and transpiration rates were almost synchronous. In May and October, when the temperature was moderate, the transpiration rate was delayed compared to the photosynthetic rate at the first peak, and the second peak was significantly smaller than the first peak. At the point-scale, the daily average ET during the growing season was 1.37 mm·d−1 and the total cumulative ET was 251 mm. Transpiration levels exhibited significant seasonal variation in the following order: July > August > June > September > May > October. At the surface-scale, the daily average ET during the growing season was 1.60 mm·d−1 and the total cumulative ET was 294 mm, which was 17% higher than that of the point-scale. The surface-scale ET was estimated using the observed values of the EC and the scale relationship formula and was optimized using different spatial scales of crop coefficients. This well reflected the ET patterns at the surface-scale. Therefore, this study proposes a spatial scale expansion method for a homogeneous underlying surface, verifies its value, and provides methodological support for estimating ET in cases of data scarcity.