Canopy resistance (rc) is a critical parameter for estimating vegetation transpiration. The site-specific rc can be calculated using the inversed Penman-Monteith (PM) equation with the effective leaf area index (LAI), which requires meteorological and turbulent flux data. The spatial distribution of rc is difficult to characterize due to the harsh environment of the Tibetan Plateau. The Jarvis-type model for modeling rc, described as a multiplicative function of environmental variables, has been widely used. However, the differences and optimization of different Jarvis-type models for alpine meadows have not been fully addressed. Consequently, our overall objective was to determine the appropriate functions for rc estimation and improve its accuracy for the alpine meadow ecosystem. Twelve Jarvis-type models composed of different stress functions were examined and compared with the observed rc calculated using PM equation at the Arou site in the northeastern Tibetan Plateau. The results suggest that the proper air temperature function and vapor pressure deficit function could improve model performance obviously. There was no obvious difference between the two different stress functions of downward shortwave radiation. The best model (M10), which was composed of an asymptotical function of downward shortwave radiation, a linear function of air temperature, an exponential function of vapor pressure deficit and a piecewise function of soil water content, had best performance with coefficient of determination of 0.93, root mean square error of 60.2 s m-1 and Nash-Sutcliffe efficiency coefficient of 0.92. The selection of proper stress functions is important for rc modeling. Models that considered the air temperature for rc calculations produced better results than those without temperature. The sensitivity analysis of rc to environmental variables indicated that rc was most sensitive to vapor pressure deficit, followed by LAI and downward shortwave radiation, whereas rc was less sensitive to soil water content. For all optimized parameters, rc was the most sensitive to kT (a fitting parameter for temperature), followed by kD (a fitting parameter for vapor pressure deficit) and rcmin (minimum rc under the optimal physiological condition). This study addresses the selection of proper stress functions in modeling rc for the alpine meadow site, which can also provide a reference for other ecosystems.

来源平台：JOURNAL OF HYDROLOGY