Estimativas populacionais a partir de dados orbitais de média resolução espacial: aplicações em municípios da Região Metropolitana de Belo Horizonte

• Ano de defesa: 2017
• Autor(a) principal: Jarvis Campos
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Estimativa para Pequenas Áreas; Sensoriamento Remoto; Mapeamento Dasimétrico; Sensoriamento remoto Minas Gerais; População Pesquisa Minas Gerais; Demografia
• Link de acesso: http://hdl.handle.net/1843/FACE-B27P7Y

In the last decades, there has been an increase in the search for more detailed information on population dynamics, given the growing demand for planning in the most different areas of the public sector, such as education, health, environment, economy, for the design of social programs, and, as well as, in the private sector, for the elaboration of market research. More recently, the dissemination of Geographic Information Systems, together with the increasing availability of Remote Sensing images, has contributed to the development of methodologies for the field of population estimates for small areas, with emphasis on a set of methods that use orbital images. In this context, the present work aims to evaluate the capacity and contribution of médium spatial resolution Landsat 7 ETM + orbital images, for the production of post-census population estimates at the municipal level, based on the construction of models based on 2000 and 2010 census tracts, and the statistical grid of IBGE, 2010. Estimates were calculated for four municipalities in the Metropolitan Region of Belo Horizonte: Contagem, Vespasiano, Sarzedo and Baldim. Firstly, we constructed models that use as explanatory variables the average of the reflectances of the spectral bands of the Landsat 7 ETM + images of the census tracts and statistical grids and the population density of the census tracts as a dependent variable. In a second moment, we constructed models that use the reflectances and the distributed population at the level of the pixels of the images, from an iterated regression procedure. All models were tested through internal validation procedures (for the year of construction of the models), external validation (for the year 2010), and comparative analyzes of post-census estimates (for the year 2015). The internal validation results, obtained through the comparison with 2010 Census data, were analyzed taking into account the undercoverage problem, in addition to the comparison of the estimates by Remote Sensing with IBGE projections for the year 2010, disaggregated to the municipal level (using the AiBi method). For the year 2015, estimates calculated using orbital images were compared to official IBGE post-census estimates, in addition to a set of demographic estimates (by simple extrapolation, vital rates and census ratio). For all municipalities, the internal validation presented excellent results (always below 6%, for all models), while in the external validation the method at the pixel level presented consistent results for municipalities with high population density (-0.7%), while the method at the sector level presented satisfactory results for municipalities with low population density (1,9%). In addition, it was possible to identify the potential of Landsat 7 ETM + images in the production of estimates that indicate a population decrease. The objective of these comparisons was to analyze the efficacy of Landsat ETM + medium resolution orbital images in the production of post-census estimates at the municipal level, as well as, more broadly, to analyze the contribution of this type of application to the field of estimation for small areas and Demography.