Estatística espacial para dados de área no estudo de indicadores econômicos e ambiental na Amazônia Legal

Abstract

The Legal Amazon shelters the largest tropical forest in the world and has suffered constant de- forestation due to land occupation with the practice of agriculture, livestock, and development

of urban centers, resulting in a decrease of the existing forest area. Despite the wealth generated

by the resources extracted from the Legal Amazon, its inhabitants live in a situation of under- development, with low-quality life and low-wage incoming. One of the indicators to quantify

the social and economic development of a region is the Municipal Human Development Index (MHDI) and the Gross Domestic Product (GDP). The entire Amazon region has been the target of development policies in an attempt to minimize deforestation and sustainably leverage the population’s development. Thus, analyzing whether such changes occurred due to the influence of neighboring municipalities or by chance raises interest. The objective of this research was to

apply spatial statistics methods to data obtained from the municipality areas of the Legal Ama- zon in order to identify differences between spatial distributions and spatial autocorrelation over

several years. The variables analyzed were the forest area (2000, 2005, 2010 and 2015), GDP (2000, 2005, 2010 and 2015), and MHDI (2000 and 2010). Different methods were used to explore the first-order effects on these variables, such as spatial moving average and the Syrjala test. Global and local Moran indices were used to quantify the spatial dependence among the Legal Amazon municipalities on the variables analyzed. The results of the analysis of the 1st order effects with the moving average showed that there was a trend towards larger forest areas in the municipalities of the states of Amazonas, Roraima, and Pará. For MHDI, the trend was detected in the municipalities of the states of Amazonas, Maranhão, and Mato Grosso. Such behavior was not observed for GDP, indicating complete randomness. The Syrjala test showed that there were statistically significant changes in the spatial distribution of the variables. The study of spatial autocorrelation, using global and local Moran indices, indicated the presence of positive spatial autocorrelation between forest area and MHDI. In general, there were changes in the spatial distributions of the variables associated with the social and economic development of the Legal Amazon municipalities from 2000 to 2015.