Integrating temperature gradient-based 3T and resistance-based models for simulating evapotranspiration and its components

[Display omitted] •An new algorithm integrated 3T and resistance-based ET model performs well in 4 sites.•Reference temperatures in 3T were numerically resolved with the Newton-Rapson algorithm.•Canopy temperature gradient representing plant water stress was seasonally dynamic. The accurate estimation of evapotranspiration (ET) and its major components [e.g., transpiration (Tr) and soil evaporation (Ev)] requires an understanding of complex soil–plant-atmosphere interactions. The temperature gradient-based three temperature (3T) model for simulating the dynamics of Ev and Tr has no requirement for resistance terms and its use of simple input variables (i.e., temperature and radiation) allows its application to both field studies and remote sensing. However, reference temperature is a key requirement for the 3T model and approaches for obtaining accurate reference temperature and the implications of different reference temperature choices on ET estimates remain poorly understood. This study improved a resistance-based model applicable to a wide range of climatic/ecosystem conditions and integrated it with the 3T model to predict ET and Tr. A novel numerical approach was used to estimate the reference temperature under dry land surface scenarios. The two models provided similar predictions of Ev, Tr, and ET at the four sites provided that the available energy for the canopy (e.g., net radiation) and ground soil (e.g., net radiation and ground heat flux) was constrained within the framework of the energy balance. Alternatively, the accuracy of simulations of Ev and ET by the 3T model significantly declined due to systematic mismatching of input drivers (temperature and radiation) under dry and wet scenarios. We emphasize that 3T is working better when input temperature and radiation under dry and wet land conditions were systematically matched within the framework of energy balance. The results of this study provide a promising novel temperature-gradient method for detecting canopy water stress.