Use of thermal units to estimate corn crop coefficients under semiarid climatic conditions

Two crop coefficient equations were derived as a function of fraction of thermal units from lysimeter measured corn evapotranspiration (ETc-lys) during 1997 and 1998, and reference evapotranspiration obtained from: (a) lysimeter measurements (Kc(mes)) or FAO Penman-Monteith (ETo-PM) estimates (Kc(est-PM)). For validation, corn evapotranspiration (ETc-est) was estimated in 2005 and 2006 from ETo-PM and: (a) the equation for Kc(mes) with (ETc-est-lyslc) or without (ETc-est-lys) locally calibrated ETo-PM; (b) the equation for Kc(est-PM); and (c) the FAO approach (ETc-est-FAO). The ETc-est_lys estimates showed the lowest bias (0.09 mm day(-1)); the ETc-est-PM and ETc-est-FAO, the highest (0.50-0.51 mm day(-1)). However, the root mean square error (RMSE, 1.23-1.27 mm day(-1)) and the index of agreement (IA, around 0.94) of the ETc-est-lys, ETc-est-lyslc and ETc-est-PM were similar. Therefore, ETc-est-lys is recommended although the ETc-est-lyslc was similarly accurate. The ETc-est-PM is less recommended due to poorer bias and systematic mean square error, and a general underestimation except for low corn ET values. For real time irrigation scheduling, the ETc-est-FAO should be avoided as RMSE (1.35 mm day(-1)), IA (0.93) and bias were slightly worse, corn ET was overestimated but for high values, and the length of the four phenological stages must be known in advance.