Extend NIR calibrations (re-calibration)

How to extend a NIR calibration with new measured sample spectra and optimize the calibration again? The important thing is not stubbornly cling to the existing chemometric model settings, but to evaluate the whole data set and newly re-modeling, so as to enable better performance. In practice, it is so that new NIR data fill the gaps in the base model, bring in concentration range extensions with them, the raw materials or the process has changed somewhat, and thus a new calibration optimization is certainly useful. Because the added new variations in the data pool generates a totally new picture, so that a different pre-processing and a modified wavelength selection can provide a much better and robust compensation for unwanted effects and increase the overall calibration performance and improve the models accuracy.