Abstract: Evapotr_anspir_ation (ET) is critical for energy and water balance in agricultur_al systems. Accur_ate estimation or measurement of ET is therefore important in improving water use efficiency and optimizing the structure of regional water use. Canopy resistance is one of the most important variables in the estimation of ET. The accur_acy of simulation of the response process of canopy resistance to environmental variables is critically important for crop ET research. A convenient approach to simulate the response process of canopy resistance to multiple factors is based on the relationship between measured latent heat, climatic variables and by using the modified Penman-Monteith (P-M) equation. However, this method has certain limitations in some pr_actical applications due to the lack of a few effective par_ameters. Another approach is to construct empirical and semi-empirical models using multiple factors (such as the Irmak model) based on measured data combined with the rearr_anged P-M equation. Based on canopy resistance values calculated by the rearr_anged P-M equation and on maize data (for the period May to September 2012) collected from the three eddy covariance observation stations in Heihe River Basin, this study constructed Irmak model taking into account the effect of atmospheric CO₂ concentr_ation of half-hourly and daily time-steps to simulate the response processes of maize to environmental variables such as net r_adiation (R_n), air temper_ature (T_a), leaf area index (LAI), relative humidity (RH), wind speed (U₃), aerodynamic resistance (r_a), effective soil water content (θ) and atmospheric CO₂ concentr_ation. In the study, the performance of the two Irmak models were tested with measured values of latent heat from the eddy covariance systems of the other two verification points. Besides, the sensitivity of environmental variables was analyzed. The results indicated that the improved Irmak model which took into account the effect of atmospheric CO₂ concentr_ation well estimated canopy resistance and ET. The coefficients of determination (R²) for canopy resistance and ET were respectively 0.76 and 0.95 for the calibr_ation phase, with root mean square errors (RMSE) of 33.1 s·m^-1 and 34.5 W·m^-2. Meanwhile, R² for canopy resistance and ET were respectively 0.68 and 0.90 for the validation phase, with RMSE of 63.2 s·m^-1 and 49.0 W·m^-2. The two verification points showed that the improved Irmak model had a good performance and strong regional applicability and spatial portability. The model also simulated the response processes of canopy resistance to environmental variables and reflected the effect of the variations in atmospheric CO₂ concentr_ation on ET. Sensitivity analysis of the improved Irmak model showed that canopy and ET were the most sensitive to net r_adiation and relative humidity, followed by air temper_ature, leaf area index and atmospheric CO₂ concentr_ation. The improved Irmak model used in this study was applicable in estimating crop water consumption and the accur_acy of ET of maize, in providing scientific basis for improvements in water use efficiency and in optimizing the structure of regional water use under increased future atmospheric CO₂ concentr_ation.