2019 Volume 88 Issue 2 Pages 132-142
This study aimed to develop a simple model to predict the culm length of rice cultivar “Koshihikari” supplied the standard level of nitrogen top-dressing in Chiba prefecture, Japan. The prediction would be useful to decrease the risk of promoting the culm extension and lodging by modifying the amount and timing of nitrogen topdressing. We analyzed the data set of “Koshihikari”, which were obtained from field experiments with three or four cropping seasons in a year for 12 years in Chiba prefecture, Japan. By the multiple regression analysis based on the results obtained by the forward stepwise method, a simple model to predict the culm length was developed (RMSE = 3.01 cm). In this model, plant length, tiller number, and leaf color (SPAD value) at the panicle formation stage, the daily minimum temperature and global solar radiation were selected as the explanatory variables. The model estimated accurately the culm length of independent experimental data sets at Chiba-city, Sanmu-city, Katori-city in Chiba prefecture (RMSE = 2.68～2.76 cm). The risk of lodging increased above the threshold of 90 cm in culm length. The accuracy rates regarding the diagnostic estimation of culm length, that is, whether culm length exceeds the threshold value or not, were 77～91%. These accuracy rates were higher than those by the conventional estimation method, which have been used in Chiba prefecture. This model would be more useful for deciding the optimum amount and timing of nitrogen fertilizer for topdressing than the conventional method.
