Análise de cenários a partir de indicadores energéticos e ambientais para as emissões brasileiras de CO2

Abstract

Brazil is a developing country that emits high amounts of CO2 per year. Therefore, controlling these emissions is essential to achieve sustainable development. In this thesis, we tested six Artificial Neural Networks (4 feedback propagation and 2 cascade feedback propagation) and of these, one feedback propagation was able to quantitatively relating CO2 emissions, energy matrix and burning in Brazilian biomes, such as the Amazon Forest. The literature still does not have studies that quantitatively demonstrate the impact that changes in the Brazilian energy matrix have on CO2 emissions in the country. In addition, there are also no studies that use fires in Brazilian biomes as input in predictive models for emissions. Our results showed that Brazilian CO2 emissions will increase in the coming years. However, partial replacement of fossil energy resources with renewables associated with the reduction of fires in Brazilian biomes could significantly reduce these emissions. In our first scenario, in which there was a partial replacement of 30% of fossil resources by renewable ones and a 70% reduction in the burning of Brazilian biomes, CO2 emissions decreased by 13.58% for the year 2030. In the second scenario analyzed, we replaced fossil fuels by 90% with renewable ones, while burning in Brazilian biomes was reduced by 90%. In this situation, we observed a 28.45% reduction in Brazilian CO2 emissions. Thus, the model developed here can help Brazil to predict and control its CO2 emissions from changes in its energetic and environmental indicators to find a balance between development and sustainability. Our model can also be used by other developing countries. For this, it is necessary that the indicators are adapted to the reality of the country studied.