Cellular automata predictive model for man-made environment growth in a Brazilian semi-arid watershed

The current study implements a cellular automata-based model for the development of land use/land cover (LULC) future scenarios using a Remote Sensing (RS) Imagery series (1985 to 2018) as data input and focusing on human activities drivers in a 6700-km² watershed vital for the water security of Paraiba state, Brazil. The methodology has three stages: the first stage is the pre-processing of images and preparing them as data input for the cellular automata land use model built in the R software environment (SIMLANDER); the stage of calibration establishes the variables and verifies the influence of each one on the LULC of the region; the last step corresponds to the validation procedures. After model calibration, land use maps for future scenarios (2019 to 2045) were simulated. The results estimate a reduction of 737 km² of natural land cover between the years 2019 and 2045. The spatial distribution of anthropogenic interference predicted a more significant degradation in the central region of the basin. This fact can be potentially attributed by the water availability increasing from the São Francisco River diversion. It is possible to identify an ascending trend of anthropogenic actions in the semi-arid region, which host the exclusively Brazilian biome-Caatinga-and contains biodiversity that cannot be found anywhere else on the Earth. The model helps large-scale LULC modelling based on RS products and expands the possibilities of hydrological, urban and social modelling in the Brazilian context.