Evapotranspiration models for a maize agro-ecosystem in irrigated and rainfed conditions

A high level of accuracy in the estimation of crop evapotranspiration (ET) may lead to significant savings of economic and water resources in irrigated agriculture. Although ET is a fundamental process in many applications, it cannot be directly measured but it has to be estimated by monitoring the exchange of energy/water above the vegetated surface (micrometeorological methods), or as a residual term of the hydrological balance (lysimeters, soil water budget). The techniques to be adopted are often complex, costly and require specific equipment. Thus, since the ‘50s, many researchers have devoted their activity to the development of models for its estimation. The available approaches can be classified in “direct†methods, based on the original Penman-Monteith (PM) equation, in which the canopy resistance rc is modelled, and “indirect†methods, based on the preliminary calculation of ET for a well-watered reference grass (ET0) with a constant rc, which is then multiplied by a crop coefficient Kc and, in case, by a stress coefficient Ks to obtain ET. Even if the latter approaches are more widely adopted for their practical simplicity, many authors show that the former often provide better ET estimates in absence of calibration of crop parameters. In this study the performances of different direct and indirect methods were evaluated in the case of a surface irrigated and of a rainfed maize grown in the Padana Plain (Northern Italy). The following models were considered: the “one-layer†original PM equation with three different models for rc (Monteith, Jarvis, Katerji-Perrier), the “two-layers†PM model proposed by Shuttleworth and Wallace, the “single†and “double†crop coefficient models illustrated in the Paper FAO-56. Latent heat fluxes measured in 2006 and 2011 in an experimental maize field by eddy-covariance were used to evaluate the models accuracy. Crop, soil and meteo data monitored contextually were used for the implementation of the different models. Results confirm that direct methods are more performing for both irrigated (2006) and rainfed (2011) conditions, with the SW model providing the best results and the FAO-56 models with generalized crop coefficients overestimating ET, especially during the middle growth stage.