Feasibility of Using NIR and PLSR Model for Prediction of Wheat Grains protein and moisture content and Mapping Quality Yield Map

Protein as an important ingredient in wheat plays main role in the production of wheat’s products. Because of the production of various products from wheat, fast and online measuring of wheat grain quality is very important   to control of flour production process and choosing an appropriate variety. Also in precision farming, combination of quantity and quality maps lets farmers to evaluate and control the plant production, well. Therefore, the purpose of this study was to evaluate the use of infrared spectroscopy in reflectance mode to predict protein and moisture content of wheat grain. In this study about 108 samples were collected from three varieties   namely Mihan, Gazkojhen and Pishgam in the region near Hamedan province in Iran. Grain proteins content were measured with a DA7200 near infrared spectroscopy apparatus. This spectroscopy collects reflectance over a wavelength range of 650-1650 nm in 5 nm increments. Results show that the best models were obtained using the PLSR method and its preprocessing SG+SNV+D1 and MA+D2+SNV for protein and moisture content, respectively. The correlation coefficient (R2), root mean square error of prediction (RMSEP) and Standard Deviation Ratio (SDR) were obtained 0.84, 0.835 and 2.54 for protein content, whereas 0.96, 0.994 and 5.34 for moisture content, respectively. Results showed that there are no significant differences among proteins of three varieties. But the sampling places have a significant effect on the protein content at the significant level of 5%. These results indicated that the infrared spectroscopy method is an efficient method and has a strong potential for rapid detection of protein and moisture content of wheat grains