Predicting grain yield and protein content in winter wheat at different N supply levels using canopy reflectance spectra

A field experiment using a split-plot randomized complete block design with three replications was carried out to determine relationships between spectral indices and wheat grain yield (GY), to compare the performance of four vegetation indices (VIs) for GY prediction, and to study the feasibility of VI to estimate grain protein content (GPC) in winter wheat. Two typical winter wheat (Triticum aestivum L.) cultivars 'Xuzhou 26' (high protein content) and 'Huaimai 18' (low protein content) were used as the main plot treatments and four N rates, i.e., 0, 120, 210, and 300 kg N ha(-1), as the sub-plot treatments. Increasing soil N supply significantly increased GY and GPC (P <= 0.05). For the two cultivars combined, significant and positive correlations were found between four VIs and GY, with the strongest relationship observed when using the green ratio vegetation index (GRVI) at mid-filling. Cumulative VI estimates improved yield predictions substantially, with the best interval being heading to maturity stage. Similar results were found between VI and grain protein yield. However, when using cumulative VI, GPC showed no significant improvement. The strong relationship between leaf N status and GPC (R-2 = 0.9144 for 'Xuzhou 26' and R-2 = 0.8285 for 'Huaimai 18') indicated that canopy spectra could be used to predict GPC. The strong fit between estimated and observed GPC (R-2 = 0.7939) indicated that remote sensing techniques were potentially useful predictors of grain protein content and quality in wheat.