Redes neurais artificiais na predição da produtividade de milho e definição de sítios de manejo diferenciado por meio de atributos do solo

Abstract

The understanding of the factors influencing yield is essential for the crop success and adoption of site-specific management. The use of artificial neural networks (ANN) is an alternative of corn yield prediction from soil properties attributes. Thus, this study aimed to evaluate the effectiveness of adoption of soil properties by interface of the regression analysis, and ANNs in the establishment of site-specific management zones and prediction of corn yield, second crop in Cerrado’s soil. Data were collected in an area of 41.76 ha cropped in 2010 and 2011. The adoption of ANNs allows better corn yield prediction despite of higher demand of construction time and processing when compared to linear regression. In consonance to the soil site-specific establishment from clay content, exchange cation capacity, organic matter and base soil saturation.