2021 Volume 90 Issue 4 Pages 430-443
The productivity of commercial rice cultivation can be improved by efficiently applying fertilizer according to the growth of the rice plant. Remote sensing can be a useful tool to capture the growth of rice plants, but a system to translate this information for application in fertilizer management needs to be developed. This study aims to develop a simple growth model for fertilizer management using sensing information. A simple growth and yield prediction model was developed based on field trials conducted with different soil types and nitrogen management for two years by using a high-yielding cultivar Yumiazusa and Akitakomachi. We developed a simplified growth model to determine the nitrogen fertilization rate. Dry matter weight, nitrogen uptake, and yield were affected by year, soil, and nitrogen fertilizer application in all years. Yumiazusa showed high yields under all conditions, which could be attributed to preferable traits for light interception. A simple growth model based on the relationship between leaf area index (LAI) and nitrogen uptake was developed. Cross-validation showed that the accuracy of the model's prediction was high with a coefficient of determination of 0.98 for dry weight in both years, allowing for accurate predictions of dry matter production based on LAI and nitrogen uptake in different soils, varieties, and years. However, the coefficient of determination of grain yield varied from 0.91 in 2017 to 0.83 in 2018, and the treatment of the harvest index in the model remains an issue.
