The turbulence data observed by eddy covariance system combined with soil temperature and soil volumetric water content (VWC) data from Xilinhot National Climate Observatory were used to analyze respiration process and its key climatic influencing factors of the semi-arid stipa grassland ecosystem in Xilin Gol during the growing season from 2010 to 2012. The applicability of four different ecosystem respiration models over this ecosystem was compared, including three multiplication models and a Q₁₀ model. Based on this research, the interannual variability of net ecosystem exchange (NEE), ecosystem respiration (R_eco) and gross primary productivity (GPP) were discussed. The respiration was affected by soil temperature and soil water content, in which soil water content was an important limiting factor. The precipitation in 2010 and 2011 was less than normal, so this ecosystem suffered from different degrees of drought stress. The respiration rate increased significantly with soil water content in the range of 12%?20%, while it was not sensitive to the change of soil water content in the range of <12% and >20%. In 2012, when the precipitation was abundant, respiration rate was more correlated with the trend of soil temperature. The fitting results indicated that Q₁₀ model had better performance (R² = 0.64) than other three models, and the cumulative total ecosystem respiration during growing season in these three years simulated by the Q₁₀ model was 157.32, 138.75 and 246.32 gC/m². The total amount of NEE was −110.28, −68.79 and −310.05 gC/m², while the total amount of GPP was 267.52, 207.57 and 555.85 gC/m². The effect of drought stress on photosynthesis was greater than that of respiration. Therefore, the inter-annual difference of net carbon exchange due to drought stress was more obvious than that of total primary productivity and respiration.

Data of EBEX-2000 was used to analyze the influences of Schotanus correction and Webb correction on turbulent heat fluxes and the governing factors of the two corrections. Furthermore, considering turbulent heat flux corrections, impact factors of energy balance ratio (EBR) were discussed and reasonable interpretations were provided according to actual weather and topographic condition. Results show that the magnitude of Schotanus correction varies from ?40 to 2 W/m², with a significant diurnal variation, and the mean value of which is ?8 W/m². The two dominant factors of Schotanus correction are atmospheric stability and vertical gradient of water vapor. The magnitude of Webb correction varies from ?5 W/m² to 14 W/m² with a mean value of 1.8 W/m², which is much smaller than that of Schotanus correction. Webb correction is positive for daytime and negative for nighttime, and it is mainly affected by Bowen ratio and specific humidity, having little probability to reach a great value. EBR calculated using a 30 minutes average is about 0.63 with large diurnal variations and dispersion, and decreases rapidly with a sudden rise of soil water content.

Data of turbulence, net radiation and temperature profiles obtained in International Energy Balance Experiment (EBEX-2000) are used to study characteristics of turbulence cospectra, temperature profiles and turbulent fluxes at 8.7 m and 2.7 m in the atmospheric surface layer. Interaction of large eddies with local turbulence in a disturbed surface layer induced by patch-to-patch irrigation is emphasized, as well as their influence on turbulent fluxes. Results indicate that peak frequencies of cospectra in the low frequency range at the two levels are well consistent, revealing that low frequency turbulence obeys the OLS law. High frequency turbulence obeys the local isotropy theory. Spectral energy in low frequency range is greatly enhanced by large eddies, and this enhancement is more notable at 8.7 m than that at 2.7 m. Multiple peak frequencies are observed at low frequency range of cospectra, and their scales correspond well with thermal heterogeneous scale of the underlying surface. Three dominant scales of these large eddies are 800, 400 and 200 m. Furthermore, large eddies have greater influence on turbulence at higher levels than lower levels and in unstable atmospheric surface layer than stable atmospheric surface layer.