Land use and land cover maps of the municipality of Buritizeiro, Minas Gerais State, Brazil, for 2018 and 2022, and the corresponding farmland abandonment map (2018-2022)

Description

The land use and land cover (LULC) map of the municipality of Buritizeiro, Minas Gerais State, Brazil, for 2018 was produced from a Sentinel-2 Multispectral Imager (MSI) satellite image acquired on 15 August 2018 and processed by the Fully Connected Neural Network (FCNN) deep learning classifier. Training, test, and validation data were obtained during a field campaign conducted in June and August 2023, complemented by visual inspection of high-resolution satellite images available in the Google Earth platform. The 2022 LULC map was generated following the same procedure, using the Sentinel-2 image acquired on 19 August 2022. The abandoned farmland map for the period 2018–2022 was derived by selecting agricultural areas mapped in 2018 (annual croplands, cultivated pastures, eucalyptus plantations, and harvested eucalyptus) that were reclassified as grasslands or shrublands in 2022.

Steps to reproduce

The methodological approach involved the acquisition of two cloud-free (<5% cloud cover) Sentinel-2 Multispectral Imager (MSI) Level-2A surface reflectance scenes. We selected the bands 2 (blue), 4 (red), and 8 (near-infrared). The following spectral indices were derived from the Sentinel-2 images acquired in 2018 and 2022 and employed as input parameters for the deep learning classification: the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Bare Soil Index (BSI). Additionally, Sentinel-2 bands 3, 4, and 8 were processed using the Linear Spectral Mixture Model (LSMM) to generate vegetation, soil, and shade fraction images. The spectral bands, vegetation indices, and fraction images were used as input parameters for the Fully Connected Neural Network (FCNN) deep learning classifier. For the FCNN classification, eight representative LULC classes were considered: forestlands, shrublands, grasslands, cultivated pastures, annual croplands, eucalyptus plantations, harvested eucalyptus, and water bodies. Agricultural areas mapped in 2018 as annual croplands, cultivated pastures, eucalyptus plantations, or harvested eucalyptus, which were reclassified as grasslands or shrublands in 2022 were identified as abandoned farmlands. The resulting raster-based change map was smoothed using the Sieve tool in QGIS to remove isolated patches smaller than 100 pixels and was subsequently converted to vector format.

Institutions

• Empresa Brasileira de Pesquisa Agropecuaria

Categories

Funders

• Foundation for Research Support of the Federal District

  Brazil

  Grant ID: 00193.00002276/2022-90 and 00193.00002586/2022-12

Licence