New values of the accumulated development rate for detailed the spring wheat phenological stage model calibrated with multilocal and multiyear documentation in subtropical Brazil

Abstract The model for simulating wheat phenological development proposed by Wang and Engel (WE) represents a major advance in establishing wheat phenological phases based on air temperature and photoperiod information. However, the WE model was developed for winter wheat and may have flaws in the phenological prediction of spring wheat. In this sense, the study aims to predict in detail the phenological stages of spring wheat, with new values of the accumulated development rate and with no effect of vernalization (f(V) = 1.0) cultivated in the subtropical region from Brazil. Climatic and phenological information of three genotypes were collected at five locations and eleven sowing dates for calibration and validation of the WE model adjusted for spring wheat. The WE model multilocal and multiyear documentation of spring wheat growth were partitioned in El Niño, La Niña, and neutral conditions to calibrate and validate the model in five sites for five growing seasons. The original model was improved from three phenological events to ten phenological stages and the model coefficient was calibrated to predict the Julian date of occurrence of each stage. A validation dataset that contained 344 observations in different weather conditions was compared with the model outputs and it resulted in R2 and Willmott agreement index greater than 0.98, and a root mean square error (RMSE) of 4 days. The new values of the accumulated development rate predicted the dates of the developmental stage for spring wheat more precisely and with a low RMSE. The cumulative RMSE of the model along the growth period reached 6 days. The present model is a sharp tool to schedule spring wheat management.