Abstract

The use of crop growth simulation models such as those incorporated into Decision Support System for Agro technology Transfer (DSSAT) are useful tools for assessing the impacts of crop productivity under various management systems. The maize growth model of DSSAT is CERES-Maize. To use it to predict fertilizer recommendation for maize (Zea mays L.) under Guinea savanna agro ecological conditions, data on maize growth, yield and development as well as data on soil and weather were collected from field experiment conducted during the 2010 growing season at Kpalesawgu in Ghana. The model was calibrated using various crop growth and development data observed at the field experiment at Kpalesawgu. Maize variety obatanpa was used in the experiment. The cultivar coefficient was calibrated with data collected from the field experiment. All measured data on phenology, grain yield and biomass from the field experiment were used for model validation and simulations. Validation results showed good agreement between predicted and measured yields with a NRMSE value of 0.181. Highest observed mean harvest maturity yield of 3831and 3795 kg/ha were obtained from plots which received 120-90-60 and 120-60-60 kg/ha N-P2O5-K2O respectively. However, the model under predicted weight per unit grain. The mean difference between observed and simulated by-product produced at maturity and top weight at maturity was significant (P ≤ 0.001).In general, maize yield simulation by DSSAT under Guinea savanna agro-ecological conditions was good. Average predicted harvest maturity yields were very close to measured values with MD of 336.0, RMSE of 498.77, NRSME of 0.181 and simulated and observed mean yields of 3096 and 2750 kg/ha for the entire treatments respectively. The mean difference between predicted and observed was not significant. The highest harvest maturity yield predicted and observed was achieved with 120-90-60 kg/ha N-P2O5-K2O. The predicted and observed average mean yield were 383