GlobCorine demonstrated an automatic service that can generate in a consistent way land cover / land use maps and land change indicators, based on a CLC-compatible legend. CLC is derived from a visual identification and classification of landscape objects using high resolution images. This methodology provides high thematic accuracy but limits the update rate since it is time-consuming. Therefore, the project evaluated the use of MERIS FR time series, processed automatically to provide a more frequent update of CLC-compatible maps. […]

Northeast Brazil (NEB), about three times the area of Spain, hosts >90% of Brazil’s drylands along with tropical rain- and dry forests. Climate variability, 45–60% cloud cover, and scarce reference data limit land use and land cover (LULC) mapping accuracy to ~80% across much of the region. Here, we introduce NEB’s first context-specific LULC framework, using phenologically timed annual MODIS mosaics (2000–2020) with 10,000 independent points, the 2018 NEB-wide map achieved 90.5% overall accuracy at Level-1. At Level-2, ecoregional accuracies were 95.9% (Amazon), 94.3% (Atlantic Forest), 89.4% (Cerrado), and 87.9% (Caatinga). Per-pixel spatial agreement with national and global datasets ranged from 29–70%. Between the 2000 and 2020 endpoints, ~540,000 km2 of NEB underwent LULC changes, based on pixel counts. Forest declined 22%, grasslands 68%, and agriculture expanded 140% – roughly 10 million soccer fields – mainly in the Cerrado MaToPiBa (Maranhão, Tocantins, Piauí, Bahia) frontier. Meanwhile, encroachment around protected areas intensified, particularly in the Amazon. This open-access product (http://www.dsr.inpe.br/DSR/laboratorios/LAF) sets a benchmark for LULC mapping in global dryland-forest mosaics, positioning NEB as a model for data-driven land management.