@article{fdi:010085407,
  title = {{D}iffuse reflectance spectroscopy for estimating soil properties : a technology for the 21st century},
  author = {{R}ossel, {R}. {A}. {V}. and {B}ehrens, {T}. and {B}en-{D}or, {E}. and {C}habrillat, {S}. and {D}ematt{\^e}, {J}. {A}. {M}. and {G}e, {Y}. {F}. and {G}omez, {C}{\'e}cile and {G}uerrero, {C}. and {P}eng, {Y}. and {R}amirez-{L}opez, {L}. and {S}hi, {Z}. and {S}tenberg, {B}. and {W}ebster, {R}. and {W}inowiecki, {L}. and {S}hen, {Z}. {F}.},
  editor = {},
  language = {{ENG}},
  abstract = {{S}pectroscopic measurements of soil samples are reliable because they are highly repeatable and reproducible. {T}hey characterise the samples' mineral-organic composition. {E}stimates of concentrations of soil constituents are inevitably less precise than estimates obtained conventionally by chemical analysis. {B}ut the cost of each spectroscopic estimate is at most one-tenth of the cost of a chemical determination. {S}pectroscopy is cost-effective when we need many data, despite the costs and errors of calibration. {S}oil spectroscopists understand the risks of over-fitting models to highly dimensional multivariate spectra and have command of the mathematical and statistical methods to avoid them. {M}achine learning has fast become an algorithmic alternative to statistical analysis for estimating concentrations of soil constituents from reflectance spectra. {A}s with any modelling, we need judicious implementation of machine learning as it also carries the risk of over-fitting predictions to irrelevant elements of the spectra. {T}o use the methods confidently, we need to validate the outcomes with appropriately sampled, independent data sets. {N}ot all machine learning should be considered 'black boxes'. {T}heir interpretability depends on the algorithm, and some are highly interpretable and explainable. {S}ome are difficult to interpret because of complex transformations or their huge and complicated network of parameters. {B}ut there is rapidly advancing research on explainable machine learning, and these methods are finding applications in soil science and spectroscopy. {I}n many parts of the world, soil and environmental scientists recognise the merits of soil spectroscopy. {T}hey are building spectral libraries on which they can draw to localise the modelling and derive soil information for new projects within their domains. {W}e hope our article gives readers a more balanced and optimistic perspective of soil spectroscopy and its future. {H}ighlights {S}pectroscopy is reliable because it is a highly repeatable and reproducible analytical technique. {S}pectra are calibrated to estimate concentrations of soil properties with known error. {S}pectroscopy is cost-effective for estimating soil properties. {M}achine learning is becoming ever more powerful for extracting accurate information from spectra, and methods for interpreting the models exist. {L}arge libraries of soil spectra provide information that can be used locally to aid estimates from new samples.},
  keywords = {calibration ; machine learning ; model localization ; reflectance ; spectroscopy ; regression ; soil constituents ; spectral libraries ; validation},
  booktitle = {},
  journal = {{E}uropean {J}ournal of {S}oil {S}cience},
  volume = {73},
  numero = {4},
  pages = {e13271 [9 ]},
  ISSN = {1351-0754},
  year = {2022},
  DOI = {10.1111/ejss.13271},
  URL = {https://www.documentation.ird.fr/hor/fdi:010085407},
}