Improving predictions of developmental stages in winter wheat: a modified Wang and Engel model ☆

Accurate simulations of plant development is an important component in crop simulation models and for use in managerial decisions, such as fertilizer applications or pest control. The three major environmental factors that control development in winter wheat (Triticum aestivum L.) are temperature, photoperiod, and vernalization. Results from a previous study showed that the prediction of several developmental stages from booting to physiological maturity were better with the Wang and Engel (WE) model compared to CERES-Wheat, but not for the prediction of terminal spikelet initiation (TS). In the WE model, the vernalization function [f(V)] is a three-stage linear function and the life cycle of the wheat crop is divided into two phases, vegetative (emergence–anthesis) and reproductive (anthesis–physiological maturity). The objective of this study was to modify the WE model by introducing a nonlinear f(V), and dividing the vegetative phase into two sub-phases (emergence–terminal spikelet initiation and terminal spikelet initiation-anthesis). A series of field experiments were carried out at Lincoln, NE, USA, to provide independent data on the date of developmental stages of two winter wheat cultivars (Arapahoe and Karl 92) for evaluating the original and the modified WE model. The root mean square error (RMSE) with the modified WE model was 5 days for Arapahoe and 6 days for Karl 92 for all developmental stages, which corresponds to a 45% decrease in the RMSE compared with the original WE model.