Título:	Digital soil mapping for soil class prediction in a dry forest of Minas Gerais, Brazil.
Autoria:	DART, R. de O. VASQUES, G. M. COELHO, M. R. FERNANDES, N. F.
Afiliação:	RICARDO DE OLIVEIRA DART, CNPS; GUSTAVO DE MATTOS VASQUES, CNPS; MAURICIO RIZZATO COELHO, CNPS; NELSON FERREIRA FERNANDES, UFRJ.
Ano de publicação:	2015
Referência:	In: CONGRESSO BRASILEIRO DE CIÊNCIA DO SOLO, 35., 2015, Natal. O solo e suas múltiplas funções: anais. Natal: Sociedade Brasileira de Ciência do Solo, 2015.
Conteúdo:	Investment on soil survey has become scarce over the past decades. Digital Soil Mapping (DSM) techniques emerged as an economic alternative to produce soil maps. We applied a classification tree algorithm to predict soil suborders in a tropical dry forest area with 102 km2 in the north of Minas Gerais state, Brazil. We tested environmental covariates with different spatial resolutions as predictors, and used 361 observations to train the model and 64 independent observations to validate the map. Prediction models included three decision trees and one logistic regression model. The results showed that freely available environmental covariates with coarser spatial resolution can produce as good or better suborder predictions than more expensive covariates with finer resolution.
Palavras-chave:	Covariáveis ambientais Árvore de classificação Resolução espacial
Tipo do material:	Artigo em anais e proceedings
Acesso:	openAccess
Aparece nas coleções:	Artigo em anais de congresso (CNPS)