SatRed: New classification land use/land cover model based on multi-spectral satellite images and neural networks applied to a semiarid valley of Patagonia

TRUJILLO-JIMÉNEZ, MAGDA ALEXANDRA; LIBEROFF, ANA LAURA; PESSACG, NATALIA; PACHECO, CRISTIAN; DÍAZ, LUCAS; FLAHERTY, SILVIA

Remote Sensing Applications: Society and Environment

Elsevier

Año: 2022 vol. 26

2352-9385

In this article we describe a new model, SatRed, which classifies land use and land cover (LULC)from Sentinel-2 imagery and data acquired in the field. SatRed performs pixel-level classificationand is based on a densely-connected neural network. The study site is the lower Chubut rivervalley which has an extension of 225 km2 and is located in estern semiarid Patagonia. SatRedshowed a 0.909 ± 0.009% (mean ± sd, n = 7) overall accuracy and outperformed the seven mosttraditional Machine Learning methods, including Random Forest. Our model accurately predictedbuildings, shrublands, pastures and water and yielded the best results with classes harder toclassify by all methods considered (Fruit crops and Horticulture). Further improvementsinvolving textural information and multi-temporal images are proposed. Our model proved to beeasy to run and use, fast to execute and flexible. We highlight the capacity of SatRed to classifyLULC in small study areas as compared to large data sets usually needed for state-of-the-art DeepLearning models suggested in literature.