Evapotranspiration estimation by combining soil moisture, land surface temperature/vegetation cover fraction data and surface-atmosphere exchange modeling

A precise estimate of the evapotranspiration (ET) is fundamental for determining the crop water needs and subsequently for optimizing water management practices and irrigation regimes. Surface soil moisture (SM) and land surface temperature (LST) are the essential components of the hydrological cycle, especially, for controlling soil evaporation and plant transpiration. The main goal of this work is to test a model of ET based on a combination of LST and SM data. For this purpose, SM data are integrated into an energy balance model (TSEB-SM) already constrained by LST and that explicitly represents the fluxes of soil and vegetation. TSEB-SM is derived from the TSEB formalism for the vegetation transpiration and represents the soil evaporation using empirical parameters (a(rss), b(rss)) of the relationship between soil surface resistance (r(ss)) and SM. For the vegetation cover fraction f(c)<= 0.4, the model calibration consists of inverting r(ss) every 30-min. When f(c)>0.4, the calibration consists in estimating the Priestly Taylor coefficient (alpha(PT)) at the daily time scale. The procedure is applied over an irrigated wheat field in the Tensift basin, central Morocco. The mean retrieved values of soil resistance calculated for the entire study period are (6.06, 2). The calibrated daily alpha(PT) ranges between 0 and 1.9. The high value of alpha(PT) is mainly justified by the intense irrigation along the agricultural season. Moreover, the ET simulated by the model fits well with observations. The obtained correlation coefficient R-2, root mean square RMSE, and mean bias error MBE are 0.75, 92 and 59 W m(-2), respectively.