Fig. 4. Scatter plots of predicted leaf weight (W leaf), leaf N content in the canopy (N leaf) and leaf area index (LAI) from hyperspectral image data with respect to their measurements in the training (a-c) and testing (d-f) set, using the Gaussian Process Regression (GPR) algorithm. The vertical bars denote the predicted standard deviations from the GPR. The grey lines indicate a 1:1 relationship.