Phenology Models for Heading Date Including Seedling Emergence Process in Dry Direct-Seeded Rice

Abstract

The phenological development model for transplanting rice as a function of temperature and photoperiod from emergence date (referred to as standard model) could not predict heading date accurately in direct-seeded rice which starts cultivation at sowing. Therefore, we developed a novel phenological development model for dry direct-seeded rice using phenology data of cultivars ‘Fusakogane’, ‘Koshihikari’ and ‘Akidawara’. First, from sowing to emergence, we compared the accuracy of emergence date predicted using six developmental rate (DVR) equations with different structures. A logistic model which showed relatively high DVRs at a low temperature could provide better prediction in all three cultivars. Second, in the prediction of heading date using the training data, the models from the predicted emergence date obtained from the above DVR equations and from the date reached at 100 and 120°C day on cumulative daily mean temperature assuming the water absorption process showed higher accuracy than the standard model calculated from the sowing date. From the prediction of heading date by these developed models and test data, the model divided at the predicted emergence date showed higher accuracy than the standard model calculated from the sowing date in ‘Koshihikari’. In ‘Fusakogane’ and ‘Akidawara’, the model considering the water absorption process showed relatively high accuracy. Since the model using the observed emergence date gave the highest accuracy in all three cultivars, improvement of prediction accuracy of emergence date could improve the accuracy of the prediction of heading date.