Estimating municipal economic activity: an alternative data-based approach

Published: 15 December 2021| Version 1 | DOI: 10.17632/6s3548srk5.1
Matheus Libório


The data originates from two sources. The territorial grids, including the limits of the Belo Horizonte microregion and the municipalities were obtained from the Brazilian Institute of Geography and Statistics (Related link 1). Estimates of population size and Gross Domestic Product of municipalities were obtained from IBGE (Related links 2 and 3). Satellite images were obtained from the United States Geological Survey (USGS, Related link 4). Images from the Landsat-8 satellite, OLI Sensor of the Collection-2 Level-2, spatial resolution of 30 meters were used. Collection-2 Level-2 images are atmospheric-corrected surface reflectance images. The dates of the images are July 6, 2018 and August 15, 2021.


Steps to reproduce

The built-up area indices were obtained from eight steps performed in QGIS software. First, 214 points were randomly sampled in the study area. Points were classified as water, vegetation, bare land and construction. Second, the urban spectral indices BU, EBBI, IBI, NDBI and UI were calculated. Third, the pixel value of each of the 214 points sampled for the five urban spectral indices was extracted. These values ​​were used to calculate the Kappa coefficient of agreement. Fourth, we obtained the pixel classification thresholds as built-up and non-built-up areas that maximized the K value. To reduce the problem of confusion between bare land, a minimum value of 0.90 for a True negative classification was established. Fifth, the upper and lower thresholds that define an area as built-up or as an unbuilt-up area were used in the binarization process. Sixth, the built-up area index of the municipality was obtained from the sum of the pixels contained within each municipality using the QGIS Raster layer zonal statistics algorithm. Seventh, the built-up area indices of the twenty-four municipalities for the five urban spectral indices were correlated with population size and the GDP of the municipalities. Eighth, the urban spectral index that best portrays the built-up areas of the municipalities in the microregion of Belo Horizonte was defined taking into account the Spearman correlation coefficients obtained in step seven.


Pontificia Universidade Catolica de Minas Gerais


Economics, Remote Sensing, Big Data