The work proposed by Liu et al. [8] involved samples that presented a certain similarity in their chemical composition, which can facilitate the performance of the models. In the present study, one large variability was sampled, with quite different biomasses grouped in one single calibration model.