Algorithm for mapping land uses and vegetable coverage, from the use of NDVI applied in the northeastern of Mato Grosso Do Sul

• Autor(a) principal: Mantovani, José Roberto
• Data de Publicação: 2023
• Outros Autores: Lelis , Leandro Reginaldo Maximino
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Revista Cerrados (Montes Claros. Online)
• Texto Completo: https://www.periodicos.unimontes.br/index.php/cerrados/article/view/5068
• topic: Cluster; K-Means; Comportamento espectral; Jenks; Cluster; K-means; Spectral behavior; Jenks; Cluster; K-means; Comportamiento espectral; Jenks

A methodology is proposed that uses mathematical concepts of the theory of cluster formation, the K-means, to automatically obtain the clusters of the NDVI values, to determine the use and occupation of the land. The implementations of the K- means method existing in specific software require the predefinition of the number of clusters, and the contribution of this methodology is the determination of the number of clusters automatically, without the need for interference of the decision maker, which may vary according to time and space from one image to another, as well as from one sensor to the next. Different sensors were selected to generalize this index classification: Thematic Mapper (TM) on board the Landsat-5 satellite; Operational Terra Imager (OLI) aboard the Landsat-8 satellite; MultiSpectral Instrument (MSI) aboard the Sentinel, level-2A satellite. The mapping and validation of algorithm are carried out in the northeast region of state of Mato Grosso do Sul, which, over 37 years (1984-2021) shows changes in its vegetation cover. The results obtained for the three periods, provided by the algorithm, better distinguished the spectral behavior of pixels referring to water classes, exposed soil and urban areas; on the other hand, the JENKS method generalized these classes, on the other hand, it better distinguished low-sized vegetation, natural vegetation and planted forests.