Abstract: To clarify the mechanism of evapotranspiration in desert steppe, we investigated the evapotranspiration and its components in the shallow mountainous area of the Qilian Mountains with five methods, including eddy covariance, lysimeter, and Priestley-Taylor (P-T), Penman-Monteith (P-M), and Shuttleworth-Wallace (S-W) models. We further analyzed the multi-timescale characteristics of evapotranspiration and its components in the Qilian Mountains, and the influencing factors. The results showed that the eddy covariance method was more accurate than the lysimeter. The applicability of the P-T model in arid regions was significantly improved by improving the α coefficient. The S-W model had the best simulation accuracy among the three models, with an R² as high as 0.74. Based on the simulation results of the S-W model, we found that soil evaporation accounted for 55.1% of the evapotranspiration (E/ET), and that the E/ET during the growing season was mainly affected by the soil water content. ET and plant transpiration on the daily scale were mainly controlled by net radiation, while soil evaporation was regulated by soil water content. ET was most sensitive to aerodynamic drag between the canopy and the reference height. Soil evaporation was controlled by soil surface drag. Plant transpiration was dominated by canopy boundary layer drag.

Key words: Priestley-Taylor model, Shuttleworth-Wallace model, Qilian Mountains, evapotranspiration