Identification of Priestley-Taylor transpiration Parameters used in TSEB model by Genetic Algorithm

The accuracy degree of extracted canopy latent heat from canopy net radiation is depending extremely to the proposed PriestleyTaylor approximation. This extracting canopy latent heat is an initial approximation to compute iteratively partitioned energy components to soil and vegetation using in Two Source Energy Balance (TSEB) Model. This approximation is using a PriestleyTaylor coefficient (αp) and fractional of Leaf Area Index (fg) that is green. The standard values are 1.26 and 1 for respectively (αp) and (fg). This study is focused to identify these two transpiration parameters (αp) and (fg) by Genetic Algorithm method to accurately predict patterns of turbulent energy fluxes by TSEB Model (Norman et al. 1995), over irrigated olive orchard in semi-arid area (Marrakech, Morocco). The (αp) and (fg) are depending on local climatic characteristics and data measurements accuracy for different periods of the year 2003. In summer 2003, the GA gives optimal values for (αp=0.93) and (fg=0.61). Ten runs of GA computing have been applied to guaranty stability of the optimization process. In fact, the simulation of latent heat becomes improved as presented as below, since comparison to ground measurements shows acceptable representativeness in summer 2003 with enhancement of TSEB Model performance assuming correlation to (0.45), bias is to (+15 W.m-2), and the root mean square have been improved to (63 W.m-2). Thus, the results obtained here show the most important support of Genetic Algorithm through the calibration and optimization processes.