Abstract
Combined with multivariate calibration methods, near-infrared (NIR) spectroscopy is a non-destructive, rapid, precise and inexpensive analytical method to predict chemical contents of organic products. Nevertheless, one practical limitation of this approach is that performance of the calibration model may decrease when the data are acquired with different spectrometers. To overcome this limitation, standardization methods exist, such as the piecewise direct standardization (PDS) algorithm. The dataset presented in this article consists of 332 manure samples from poultry and cattle, sampled from farms located in major regions of livestock production in mainland France and Reunion Island. The samples were analysed for seven chemical properties following conventional laboratory methods. NIR spectra were acquired with three spectrometers from fresh homogenized and dried ground samples and then standardized using the PDS algorithm. This important dataset can be used to train and test chemometric models and is of particular interest to NIR spectroscopists and agronomists who assess the agronomic value of animal waste.