01840naa a2200229 a 450000100080000000500110000800800410001902400550006010000180011524501390013326000090027252010540028165300310133565300270136665300230139365300360141670000220145270000270147470000240150170000230152577300620154821494582022-12-12       2023    bl uuuu     u00u1 u    #d7 ahttps://doi.org/10.1016/j.jappgeo.2022.1049002DOI1 aBASTOS, B. P.  aClustering airborne gamma-ray spectrometry data in Nova Friburgo, State of Rio de Janeiro, southeastern Brazil.h[electronic resource]  c2023  aThe goal of this study was to test different image clustering techniques, using airborne gamma-ray spectrometry data to optimize the interpretation of areas with similar properties in large scale. The methodology applied was based on the comparison of two techniques of data-driven classification (K-means and Gaussian Mixture Models) and a technique of knowledge-driven classification (Simplified RGB) to discriminate domains from the primary variables potassium (K), uranium (eU), and thorium (eTh) obtained from airborne gamma-ray spectrometry surveys. The performance of these three methods was evaluated through pre-processing techniques and post-processing including the best number of clusters/classes, visual interpretation, internal validation, and boxplot analysis. The clustering performance was considered satisfactory through the visual interpretation and comparison with the geological map and the DEM, for all three methods. For a quantitative analysis, the simplest model of unsupervised clustering Gaussian Mixture Models prevailed.  aData-driven classification  aGamma-ray spectrometry  aGeological mapping  aKnowledge-driven classification1 aPINHEIRO, H. S. K1 aCARVALHO JUNIOR, W. de1 aANJOS, L. H. C. dos1 aFERREIRA, F. J. F.  tJournal of Applied Geophysicsgv. 209, 104900, Feb. 2023.