Automatic environmental zoning with self-organizing maps.

• Autor(a) principal: SILVA, M. A. S. da
• Data de Publicação: 2018
• Outros Autores: MACIEL, R. J. S., MATOS, L. N., DOMPIERI, M. H. G.
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Repositório Institucional da EMBRAPA (Repository Open Access to Scientific Information from EMBRAPA - Alice)
• Texto Completo: http://www.alice.cnptia.embrapa.br/alice/handle/doc/1103941
• topic: Artificial neural network; Exploratory spatial analysis; Similarity coefficients; Alto Taquari river; Correspondence analysis

This article presents the application of the Self-Organizing Maps (SOM) as an exploratory tool for automatic environmental zoning by combining the handle of categorical data and the other for automatic clustering. The SOM online learning algorithm had been chosen to treat categorical data by using the dot product method and the Sorense-Dice binary similarity coefficient. To automatically perform a spatial clustering, an adaptation of the automatic clustering Costa-Netto algorithm had been also proposed. The correspondence analysis had been used to examine the profiles of each homogeneous zones. To explore the approach it has been performed the environmental zoning of the Alto Taquari River Basin, Brazil, using as input data a set of thematic maps. The results indicate the applicability of the approach to perform the exploratory environmental zoning.