Testing CERES-Maize versions to estimate maize production in a cool environment

CERES-Maize is a relatively simple deterministic crop model that simulates maize development, growth and yield. Since its first release in 1986, slight changes in the original model have been proposed by different researchers, generating a number of official or unofficial versions of CERES-Maize. Nevertheless, these versions have not been extensively tested under cool environments. The objective of this work is to evaluate the performance of three different recent versions of CERES-Maize ((i) CERES-Maize-2003 (called thereafter CERES-2003), the most recent version proposed by Kiniry; (ii) that included in DSSAT V3.5 official release or CERES-3.5 and (iii) the recently released version with DSSAT V4.0 or CERES-4.0) in a cool environment, where water and nutrients are fully available, and to document and discuss the equations causing differences in model predictions among versions. To achieve these objectives, these versions were tested against field data sets, obtained in northwest Spain between 1998 and 2002. CERES-4.0 simulated more closely the biomass and grain yield under this relatively cool environment. CERES-2003 showed the poorest performance, mainly due to the 64% dry weight loss programmed to occur with dry matter translocation from stem to grain. Reasons for CERES-4.0 advantage are related to the new look-up temperature functions affecting radiation use efficiency (PRFF) and grain filling rate (RGFILL) that in V4.0, create less sensitivity to temperature. Nevertheless, under these growing conditions CERES-4.0 predictions may benefit from slightly more temperature sensitive PRFT or RGFILL functions, but more extensive testing is needed to confirm this fact. (C) 2005 Elsevier B.V. All rights reserved.