关于植被用水量对环境因子响应的问题研究。网上可见后附加链接和校正后的文章，现在是提交的修改稿。欢迎指教！

Wang et al. 2016. Agr_For_Meteo.pdf

Brief background and objectives

Tree water use (E_c) models used in this studyinclude (1) Penman-Monteith equation with the canopy conductance (g_c) simulated from theJarvis-Stewart (JS) approach, denoted as PMJS; (2)models modified from the JS approach that correlate E_c directly withenvironmental variables, omitting the calculation of g_c, denoted as MJS; and (3) a simplified process basedmodel by Buckley et al., (2012), denoted as BTA. Tree water use is constrained bythe root-zone water supply and atmospheric demand (solar radiation, airtemperature and humidity, and wind speed) factors. Many studies estimated E_c or g_c considering both supply and demand factors, whilesome only included the demand factors. In addition, soil water availability hasseasonal variations in most cases in correspondence with precipitation, and airtemperature is often neglected in E_cmodels when humidity is used as one variable in previous studies.

Therefore, thisstudy compared selected E_cmodels, with the emphasis on the following questions for a better understandingof the responses of E_c toenvironmental variables:

(1) Which type ofE_c models performs betterat different temporal scales (hourly and daily)?

(2) Are soilwater availability and air temperature functions critical for E_c simulation?

(3) At which timescale and in which season do soil water condition and air temperature functionspose a stronger influence on E_cmodeling; and

(4) Are parametervalues transferable across different temporal scales (daily and hourly) for thesame E_c model?

Major conclusions

Themodified JS and process based models gave generally better simulations than thePM models at both daily and hourly scales. The models constructed with fourenvironmental variables outperform the ones with a reduced number of variables.At the daily scale, the best PM model performs comparable to the best MJSmodel. The major advantage of the MJS and BTA models is the simplicity in termsof inputs and number of parameters, thus have potential for wide applicationsin land surface models for evapotranspiration estimation. BTA model has theleast number of parameters and sound physical interpretations of plantphysiology. However, BTA failed on the tree under water stress at the dailyscale due to its treatment of soil water availability and other factors as anintegrated parameter.

Soilwater availability function is important for E_c simulation at both temporal scales, particularly atthe daily scale. For hourly E_cmodeling the soil water function can be omitted in spring time in this studywhen there was sufficient water in the root-zone soil for vegetation wateruptake. The influence of an air temperature function on model performancevaries. Parameter values showed divergence across models and temporal scales,calling for attention to model application across temporal scales. At thehourly scale, parameters are better to be calibrated for each season ratherthan calibrated for all seasons for the improvement of long-term total treewater use modeling.