Evapotranspiration（ET）

including evaporation from soil surface and vegetation transpiration

is an impor- tant parameter for water and energy balances on the Earth’s surface.Many applications in water resource

agriculture

and forest management require the knowledge of ET over a range of spatial and temporal scales.Satellite provides an unprece- dented spatial distribution of critical land surface parameters such as surface albedo

fractional vegetation cover

land sur- face temperature

and so on

so numerous physical and empirical remote sensing-based models in conjunction of ancillary surface and atmospheric data have been developed to estimate ET for clear sky days in recent years.The models for evapo- transpiration estimation using remotely sensed data are categorized as single-layer and two-layer model

Penman-Monteith model and empirical model in this paper.A review of algorithms and approaches of these models is presented

with a dis- cussion regarding the main merits and limitations of SEBAL（Surface Energy Balance Algorithm For Land）

SEBS（Sur- face Energy Balance System）and TSEB（Two Source Energy Balance）that are widely used.The problems of temporal ex- tension with its uncertainty

the temporal and spatial resolution with relative scale effect

the selection of models for appli- cation and the evaluation of model applicability

the effect of advection on energy balance principle

and the accuracy as- sessment are discussed respectively.To improve the accuracy of ET estimation using remotely sensed data at reginonal scale

several problems have to be overcome in the future.Suggestions are as follows:Firstly

the research on the mecha- nism of interaction between land surface and atmosphere with related land-surface process should be enhanced because the parameterization schemes of land-surface process are currently still too simple to depict the complicated mechanism pre- cisely

especially for the conditions of heterogeneous underlying surface and complex terrain.Secondly

the accuracy of land surface variables retrieved from remotely sensed data has not achieved perfect level and needs to be improved because it can significantly affect the accuracy of ET estimation.Enhancing the research on the interrelationship between remote sensing information of a variety of bands and characteristics of land surface and micrometeorology can enlarge the applica- ble range of two-layer model

which also conduces to the establishment of ET estimation model of more accurate and appli- cable.Thirdly

the use of hydrological methodology can provide effective methods for accuracy assessment of remote sens- ing-based models

underway to develop a new parameterization scheme incorporating the strengths of remote sensing-based and hydrological models.Fourthly

enhancing the research regarding scale effect and error propagation could improve the accuracy of ET estimate model using coarse spatial resolution data.