Assessment of the land use classification surrounding a dam in the semi arid of Pernamuco/Brazil from OLI, MUX images and WorldCover data

Abstract Detecting changes in land cover helps policymakers to understand the dynamics of environmental changes to ensure sustainable development in the Caatinga Biome. Thus, the identification of spatial features by Remote Sensing emerged as an important research aspect and, therefore, an adequate and accurate methodology for land cover mapping was necessary. In this study, Landsat-8 and CBERS-4 satellite data captured by the Operational Land Instrument (OLI) and the Regular Multispectral Camera (MUX) were used for classification and accuracy analysis for six land cover classes around the dam ‘Barra do Juá’, located in the semi-arid region of Pernambuco state. Recent high resolution images from the European program WorldCover were used as spatial and thematic reference images. The approach developed provided land cover maps for each dataset. After the comparative analysis with the reference product, an accuracy of the producer and average user of 62.44% and 71.74% for MUX and 60.88% and 62.38% for OLI were obtained, respectively. Spatial and spectral characteristics of the images were the main causes of the variability found in the thematic accuracy coefficients. The results obtained showed that the CBERS-4 MUX presented more satisfactory land cover maps than the Landsat-8 OLI data. However, for the classes of natural vegetation, OLI presented better results than MUX.