Evapotranspiration is not only an important component in hydrological cycle, but also a main part of energy balance. Estimation of the evapotranspiration is essential for understanding the large-scale energy and water balance. Many methods, including traditional methods, land surface process model and remote sensing, have been developed to estimate evapotranspiration. For gaining large-scale land surface characteristic information, remote sensing techniques make it possible to estimate evapotranspiration quite accurately.
There are a lot of remote sensing methods to estimate evapotranspiration, which can be summarized as follows: statistical (empirical) and half-statistical models, physical models and numerical models and each method has its own advantages and disadvantages.
According to the present studies, most remote sensing methods mainly use visible, near-infrared and thermal infrared wave bands. The former two bands are used to estimate canopy density and albedo and thermal infrared band can provide information about land surface temperature. In recent studies, a few models have been developed to estimate land surface water status and land surface temperature by using microwave data instead of optical remotely sensed data. A lot of retrieval algorithms have been developed in retrieving land surface parameters such as albedo, land surface temperature and emissivity. For example, land surface temperature, an important land surface parameter, can be estimated well by means of split window technique.
Great progresses have been achieved in the application of remote sensing technology to evapotranspiration research, but some problems must be resolved to improve the precision of these evapotranspiration models. ①The energy fluxes change of hourly, daily even longer time scale from observations of remote sensing are instantaneous. ②Air temperature of each pixel can not be obtained directly while most models are sensitive to difference of surface radiative temperature and air temperature. ③ Influences of atmospheric correct, radiation calibration and observing angle to the measurement of surface radiative temperature are not well known. ④Continuous calculation of surface fluxes is very important, but the exist of cloud makes satellite observation discontinuous. ⑤Satellites (such as NOAA and GOES) of big pixel scale (1～4 km) have enough observing frequency, but heterogeneous sub-pixel makes observation uncertainty. With the data of series of new EOS sensors, there will be new hope for evapotranspiration study.