Centre de documentation
scientifique

Using machine learning to map Western Australian landscapes for mineral exploration / Thomas Albrecht in ISPRS International journal of geo-information, vol 10 n° 7 (July 2021)  

Public [article]  inISPRS International journal of geo-information > vol 10 n° 7 (July 2021) . - n° 459 Titre : Using machine learning to map Western Australian landscapes for mineral exploration Type de document : Article/Communication Auteurs : Thomas Albrecht, Auteur ; Ignacio Gonzalez-Alvarez, Auteur ; Jens Klump, Auteur Année de publication : 2021 Article en page(s) : n° 459 Note générale : bibliographie Langues : Anglais (eng) Descripteur : [Vedettes matières IGN] Applications de télédétection [Termes IGN] apprentissage automatique [Termes IGN] Australie occidentale (Australie) [Termes IGN] cartographie automatique [Termes IGN] classification dirigée [Termes IGN] classification par forêts d'arbres décisionnels [Termes IGN] classification par réseau neuronal convolutif [Termes IGN] classification par séparateurs à vaste marge [Termes IGN] géomorphologie [Termes IGN] modèle numérique de surface [Termes IGN] prospection minérale [Termes IGN] Python (langage de programmation) Résumé : (auteur) Landscapes evolve due to climatic conditions, tectonic activity, geological features, biological activity, and sedimentary dynamics. Geological processes at depth ultimately control and are linked to the resulting surface features. Large regions in Australia, West Africa, India, and China are blanketed by cover (intensely weathered surface material and/or later sediment deposition, both up to hundreds of metres thick). Mineral exploration through cover poses a significant technological challenge worldwide. Classifying and understanding landscape types and their variability is of key importance for mineral exploration in covered regions. Landscape variability expresses how near-surface geochemistry is linked to underlying lithologies. Therefore, landscape variability mapping should inform surface geochemical sampling strategies for mineral exploration. Advances in satellite imaging and computing power have enabled the creation of large geospatial data sets, the sheer size of which necessitates automated processing. In this study, we describe a methodology to enable the automated mapping of landscape pattern domains using machine learning (ML) algorithms. From a freely available digital elevation model, derived data, and sample landclass boundaries provided by domain experts, our algorithm produces a dense map of the model region in Western Australia. Both random forest and support vector machine classification achieve approximately 98% classification accuracy with a reasonable runtime of 48 minutes on a single Intel® Core™ i7-8550U CPU core. We discuss computational resources and study the effect of grid resolution. Larger tiles result in a more contiguous map, whereas smaller tiles result in a more detailed and, at some point, noisy map. Diversity and distribution of landscapes mapped in this study support previous results. In addition, our results are consistent with the geological trends and main basement features in the region. Mapping landscape variability at a large scale can be used globally as a fundamental tool for guiding more efficient mineral exploration programs in regions under cover. Numéro de notice : A2021-546 Affiliation des auteurs : non IGN Thématique : GEOMATIQUE/IMAGERIE Nature : Article nature-HAL : ArtAvecCL-RevueIntern DOI : 10.3390/ijgi10070459 Date de publication en ligne : 06/07/2021 En ligne : https://doi.org/10.3390/ijgi10070459 Format de la ressource électronique : URL article Permalink : https://documentation.ensg.eu/index.php?lvl=notice_display&id=98048 [article]