A k-means-based-approach to analyze the emissions of GHG in the municipalities of MATOPIBA region, Brazil

Brazil is the sixth largest emitter of GHG, with land use change and agriculture being the main source of these emissions. The recent expansion of agriculture in Brazil has been occurring mainly in the MATOPIBA region, a territorial division inserted on Cerrado. Clustering methods are useful for driving hyper-local GHG emission reduction strategies, but they have yet to be applied at the municipal level, nor from the emissions of various economic sectors. In order to contribute to the identification of the most critical areas in relation to GHG emissions for MATOPIBA, this study proposes an approach of municipal clustering according to the percentage contribution of Agriculture, Land Use Change, Energy and Waste sectors in total emissions. The clustering was performed with the k-means algorithm, using the elbow method and the silhouette score to define the number of clusters. In addition, statistical and geostatistical analyses were conducted to assess the consistency and spatial autocorrelation of the groups formed. The approach was able to generate six clusters with distinct characteristics, showing the heterogeneous profile of GHG emissions from MATOPIBA. At the same time, the clustering of similar municipalities can help in making decisions about the best pro-environmental measures to reduce/remove GHG to contain global warming.