Sugarcane cultivation requires correct fertilizer rates. However, when nutrients are not available, or there is over-fertilization, the yields are significantly reduced and the environmental burden increase. In this study, it is proposed a decision support system (DSS) for the correct NPK (nitrogen, phosphorus and potassium) fertilization. The DSS consists of two fuzzy models; the edaphic condition model (EDC-M) and the NPK fertilization model (NPK-M). The DSS using parameters from soil analysis and is based on the experience of two groups of experts to avoid the bias to the reality of a single group of professionals. The results of the DSS are compared with the results of soil analysis and those of the group of experts. One hundred and sixty tests were developed in the NPK-M. The N rate shows R 2=0.981 for the DSS and R 2=0.963 for soil analyzes. The P rate shows R 2=0.9702 for the DSS and R 2=0.9183 for the soil analyzes. The K rate shows R 2=0.9691 for the DSS and R 2=0.9663 for the soil analyzes. Environmental results indicate that the estimated rates with the DSS do reduce the environmental impact on the tests performed.
Keywords: Confidence intervals; NPK fertilization; decision support system; life cycle assessment; response surface analysis.