Abstract
In recent years, sugar beet has been grown by the transplanting method in 95% of the sugar beet fields. However, this method is labor- and cost-intensive. Consequently, a drilled crop that could lead to cost-saving cultivation is now being targeted. To improve the adaptability of the drilling method, it is important to select heavier true seeds correlated with early growth. However, since the investigation depends on manual procedures, the processing of all the materials was associated with much labor and time. Therefore, a new method for estimating the true seed weight simply and efficiently using the spectra of the pulverized sugar beet fruits by Near-Infrared Diffuse Reflectance spectroscopic (NIR) analysis was developed. This method, enabled the estimation of the ratio of true seed in fruits by a quantitative analysis focusing on the spectrum of fat, namely 2310 nm–2346 nm. As a result of Partial Least Squares (PLS) regression analysis, it was considered that the calibration curve included five factors. Therefore, Standard Error for Calibration (SEC) and Standard Error for Prediction (SEP) were the same at 2%. Thus, the correlation coefficient between an actual measurement and an estimated measurement by the NIR method was 0.81. Furthermore, it became possible to compute the true seed weight from the product of this ratio and fruit weight. The actual measurement and the NIR measurement of the true seed weight was closely correlated (correlation coefficient of 0.91). Based on there data, it was shown that the true seed weight of sugar beet could be efficiently estimated by the NIR method. Since the true seed weight could be estimated accurately enough for practical use, this method was found to be useful for selecting lines with a high adaptability to drilled cultivation.
