A Simple guide to complex world of overtone and combination bands: Theoretical simulation and interpretation of NIR spectra – summary of the workshop at NIR-2021 Beijing Conference

Between 18 and 21 October 2021, the 20th International Conference on NIR spectroscopy in Beijing took place. Despite this time being held as a virtual event, it was a highly successful symposium met with high interest from the wide audience – as evidenced by many excellent presentations, around which numerous vivid discussions developed. During the conference, four workshops were offered, focused at discussing few areas essential for NIR spectroscopy and its applications. Excellent workshops were provided by Professors Heinz Siesler, Hui Yan, Dolores Pérez-Marín and Tom Fearn, in which invaluable knowledge was shared with the participants of the conference. Among these renowned experts, I had the honour to offer my contribution with the workshop aimed at the physicochemical foundations of NIR spectroscopy, an area that seldom is exhaustively presented in the textbooks. The workshop aimed at shedding light on the complex world of overtone and combination bands and was met with a considerable interest from the participants. As many questions have been asked both during the dedicated Q&A session, as well as through other channels and private correspondence, I would like to provide a short recapitulation of the workshop in the form of this brief article. Some of the most essential ‘take home messages’, such as the origin of the intensity variation of the overtone bands and the famous ‘selection rule’ of the harmonic oscillator, among others will be briefly outlined here.

Interpretable machine learning with an eye for the physics: Hyperspectral Vis/NIR “video” of drying wood analyzed by hybrid subspace modeling

Chemometric multivariate analysis based on low-dimensional linear and bilinear data modelling is presented as a fast and interpretable alternative to more fancy “AI” for practical use of Big Data streams from hyperspectral “video” cameras. The purpose of the present illustration is to find, quantify and understand the various known and unknown factors affecting the process of drying moist wood. It involves an “interpretable machine learning” that analyses more than 350 million absorbance spectra, requiring 418 GB of data storage, without the use of black box operations. The 159-channel high-resolution hyperspectral wood “video” in the 500–1005 nm range was reduced to five known and four unknown variation components of physical and chemical nature, each with its spectral, spatial and temporal parameters quantified. Together, this 9-dimensional linear model explained more than 99.98% of the total input variance.