Estimating ecosystem evaporation and transpiration using a soil moisture coupled two-source energy balance model across FLUXNET sites

The two-source energy balance model coupled with soil moisture (TSEB-SM) was evaluated against observations from a global set of 57 eddy covariance (EC) sites, part of the FLUXNET2015 dataset. In addition, modeled soil evaporation (E) and transpiration (T) were compared with the values obtained from the Transpiration Estimation Algorithm (TEA) and underlying water use efficiency (uWUE) approaches. The TSEB-SM model framework using near-surface soil moisture improved the agreement to EC-observed sensible and latent heat fluxes, reducing mean absolute percentage error (MAPE) by about 30% and root mean square error (RMSE) by about 44 W/m2 across all sites. The results show that the advantage of the TSEB-SM model, with respect to the original TSEB, becomes more evident as the ratio of actual to potential evapotranspiration (AET/PET) decreases. The E and T produced by TSEB-SM has better correlation with the results of uWUE partitioning than TSEB, especially under low soil water content condition. Likewise, TSEB-SM is superior to TSEB in simulating T when compared with sap flow measurements derived from the SAPFLUXNET database. These results imply that the development and application of TSEB-SM has made significant advances in modeling surface water fluxes, even though uncertainties remain. The approach used in TSEB-SM, driving the model with an extensive remotely sensed parameter set, gives valuable information on water use and provides an alternative to Global Climate Models where complex interactions of ecosystems are parametrized. Thus, TSEB-SM provides a unique insight into the flow of energy and the role of surface fluxes in the global water cycle.