Using the CERES-Maize model in a semi-arid Mediterranean environment. New modelling of leaf area and water stress functions

The CERES-Maize model does not simulate satisfactorily the productivity under soil water scarcity. In a previous paper by Ben Nouna et al. (Eur. J. Agron. 13 (2000) 309), this poor performance was attributed to an inaccurate leaf surface simulation and to the water stress function. This paper proposes the introduction of two new modules in the original version of CERES-Maize model. For the LAI module, an analytical approach is followed, consisting of simulating the leaf area of each individual leaf, by the calibration of different functions (leaf appearance, leaf area growth, end of leaf growth, leaf senescence). The new stress function expresses the ratio of actual to maximum evapotranspiration rate as a function of an indicator of the plant water status (pre-dawn leaf water potential). The two modules have been separately introduced into the CERES model and the output compared with the observed values and those simulated by the CERES model in its original version. The results of validation showed that the revised LAI module improved the LAI prediction compared with the one of the original model. Moreover, for the purpose of generalising the new LAI module, its predictions were also compared with the LAI simulated by the Muchow et al. model (Agron. J. 87 (1990) 338), which has been calibrated at several latitudes with different maize varieties. As compared with the original version of CERES, the new water stress function gave a different prediction in terms of the timing and intensity of soil water deficit during the cropping cycle. The stress intensity, as calculated by the new function, was generally less severe than that simulated by the original version.