Yield modeling for prediction of regional whole-crop barley productivity
Document Type
Article
Publication Date
7-1-2019
Abstract
Rice production has surpassed market demand in South Korea, and it is necessary to diversify the inputs of available agricultural resources from rice production to forage or other more demanded grains. This research was conducted to develop a yield predictive model of whole-crop barley (Hordeum vulgare L.) based on climatic data in South Korea. A data set (n = 290) containing dry matter yield (DMY), cultivated locations and climatic variables was developed based on collected research reports. With DMY as the response variable, three optimal climatic variables including spring accumulated temperature (SAT), period to accumulated temperature 150°C (PAT150) and spring duration of sunshine (SDS) were selected through the stepwise approach of multiple regression analysis. Subsequently, through the general linear model, the yield predictive model including the three climatic variables and cultivated locations was constructed. The adjusted R2 of the model was 51.6%. Meanwhile, the model was tested through residual diagnostics and cross-validation. The selection of SAT, PAT150 and SDS confirmed that the growth and development of autumn-seeded whole-crop barley were active in the next spring in South Korea, while the coefficient of determination of the model was lower than anticipated since only considering climatic variables. In the future research, a new model with higher interpretability might be developed via increasing data size, adopting more environmental variables or constructing the models for each location individually instead of pooling their data together.
Publication Source (Journal or Book title)
Grassland Science
First Page
179
Last Page
188
Recommended Citation
Guan, L., Peng, J., Han, K., & Sung, K. (2019). Yield modeling for prediction of regional whole-crop barley productivity. Grassland Science, 65 (3), 179-188. https://doi.org/10.1111/grs.12233