The accurate estimation of evapotranspiration (ET) at regional scale is vital for agricultural ecosystem and hydrologic cycle. Remote sensing can overcome the limitation of spatial scale from traditional ET measurement and inversion approaches. However, remote sensing only obtains the instantaneous ET estimates at the overpassing time of satellite. In the practical application, temporal upscaling of instantaneous ET to daily or longer temporal scale is required. This study reviewed the methods for the temporal upscaling of ET, compared the applicability of different methods, gave quantifiable advice on how to choose the different methods, and assessed the uncertainties from model itself, model parameterization, along with model evaluation. Future research should address the following issues: 1) The precision of remote sensing for obtaining surface parameters needs to be further improved. Land surface temperature, surface albedo, emissivity, normalized difference vegetation index (NDVI), fractional vegetation cover, soil moisture and leaf area index (LAI) have significant impact on the partition of the 4 energy components in the energy balance models and consequently on the accuracy of the retrieval regional ET. More attention should be paid to the physical interpretation of these surface variables, and understanding the mechanism of remote sensing radiative transfer needs to deepen. 2) Now, more than 2000 terrestrial flux stations have been established all over the world, which almost cover the major ecosystem types of the world. However, the comparative research on existing methods of temporal scale expansion is less. Therefore, relying on the existing field data of long-term ground observation stations, further comparison of the applicability between different temporal upscaling methods is definitely required in different climate zones and crop types, which can provide reference on how to select the appropriate temporal scale expansion method based on the characteristics of study area. 3) The temporal upscaling method of remote sensing ET has its own advantages, disadvantages and suitable conditions, and hence it needs to be further improved and perfected. For example, although the constant reference evaporative fraction method has higher estimation precision than other methods, a variable canopy resistance would be introduced for the reference grass instead of using the fixed value. 4) In the future, we can introduce hyperspectral data to improve the accuracy of surface temperature inversion, use microwave data which is less affected by weather to estimate the ET of rainy days, and combine new technology and method such as large aperture scintillation detector (LAS) to carry out satellite-aircraft-ground simultaneous observation experiment; and integrated use of multi-source, multi-scale remote sensing data to expand temporal scale of ET based on the characteristics of study area, is a breakthrough for future research. 5) Based on data assimilation methods, land surface process and atmospheric general circulation model, relying on the existing field data of long-term ground observation stations, the land surface data assimilation system of ET can be constructed, which can integrate multi-source/scale direct and indirect observations data. Achieving dynamic monitoring and forecasting of ET in agro-ecosystem is one of the future trends.