Evapotranspiration (ET, or latent heat flux) is the most essential and uncertain factor in water resource management.
Remote sensing is a promising tool for estimation of spatial distribution of ET at regional scale with limited ground
observations. We developed an algorithm for estimating regional evapotranspiration from MODIS 1b data and ancillary
meteorological data. The algorithm is an integration of Penman-Monteith equation and SEBS (Surface Energy Balance
System) model. The former is a combination of the energy balance theory and the mass transfer method to compute the
evaporation from cropped surfaces from standard climatological records of sunshine, temperature, humidity and wind
speed by introducing resistance factors, and the latter determines the spatio-temporal variability of regional evaporative
condition. First, we characterized key land surface parameters on satellite over passing days, including fractional
vegetation cover (fc), roughness height for momentum (z0m), net radiation (Rn) and soil heat flux (G0); Second, SEBS
was applied to partition the sensible heat (H) from latent heat (LE) in combination with Planetary Boundary Layer (PBL)
information from seven meteorological stations. A parameterization of surface roughness was applied at mountainous
area considering topographic influence; third, we chose available surface resistance (RS) as the temporal-scaling factor.
With bulk surface resistance is properly defined, P-M methods is valid for both soil and vegetation canopy. We validated
ET from this algorithm with limited actual observations of ET including 2 eddy covariance system dataset and 1
lysimeter sites. Water balance equation is used as a trend-analysis tool to show the consistency between rainfall and ET
on four drainage area. As a result, the prototype products showed different accuracy and applicability on different
underlying and time scale, which demonstrates the potential of this approach for estimating ET from 1-km to regional
spatial scale in North China Plain.