Data from: Geographic Object-Based Image Analysis Framework for Mapping Vegetation Physiognomic Types at Fine Scales in Neotropical Savannas
DOI10.5281/zenodo.10035746Zenodo10035746MaRDI QIDQ6720569FDOQ6720569
Dataset published at Zenodo repository.
Frank W. Davis, Kelly K. Caylor, Laura L. Hess, Gabriel Antunes Daldegan, Fernanda F. Ribeiro, Dar A. Roberts
Publication date: 24 October 2023
Copyright license: Creative Commons Attribution 4.0 International
Spatially-explicit information featuring a wide range of detailed vegetation structural types are necessary to support conservation and ecological analyses. A systematic approach that accurately maps such detailed categories at regional scales is currently lacking for neotropical savannas. We developed a systematic Geographic Object-Based Image Analysis (GEOBIA) framework that accounts for spectral and spatial properties to map Cerrado vegetation structural types at 5-m resolution. This framework counts with a two-step mapping approach: (1) image segmentation and a Random Forest land cover classification based on spectral information (Level 1 classification), followed by (2) a GEOBIA knowledge-based classification that follows contextual and topological spatial rules developed in a systematicmanner for mapping Cerrado ecological classes (Level 2 classification).The framework was tested for two large study sites covering most major Cerrado physiognomic types: a control site in central Brazil (Taquara watershed site, located in Brasilia, Federal District, and embraces the IBGE Ecological Reserve) and another larger site that is located in an agricultural landscape in the western portion of Bahia State (encompasses parts of the municipalities of: Sao Desiderio, Luis Eduardo Magalhaes, Barreiras, and Riachao das Neves), which brings additional mapping challenges related to intra-class spectral similarity.Results demonstrate that our GEOBIA approach is effective for mapping 13 land cover classes with 87.6% overall accuracy, of which all 11 major vegetation classes were identified. For additional details, please check the associated publication with these datasets (Ribeiro et al. 2020).The datasets developed in this study are available in shapefile format with attributed metadata following ISO 19115 standards:wba_lcmap_geobiaL2_2011lucmask: final land cover map (Level 2 -- physiognomic types) for a section of the western bahia region in the Cerrado. This product uses a land use mask of 2011 derived from ancillary datawba_lcmap_geobiaL2_2013lucmask: finalland cover map (Level 2 -- physiognomic types) for a section of the western bahia region in the Cerrado. This product accounts for the land use mask of 2011 as well as an updated land use mask of 2013 derived from ancillary dataibge_taquara_luc_L1_L2_2013_original: land cover maps (Levels 1 and 2) developed for the Taquara/IBGE siteibge_taquara_detailed_luc_2013: final land cover map (Level 2 -- physiognomic types) developed for the Taquara/IBGE site with additional detailed land use categoriesDataMapped land cover classes: 1) cerrado woodland, 2) savanna, 3) open savanna, 4) shrubby grassland, 5) grassland (only present in the Taquara site), 6) cerrado scrub (only present in the western bahia site), 7) marsh, 8) shrub swamp, 9) palm swamp, 10) riparian forest, 11) semi-deciduous forest, 12) seasonally dry tropical forest (only present in the western bahia site), 13) invasive forbes (only present in the Taquara site), 14) non-natural/barren, 15) water, 16) shade, 17) cloudsThe land-use mask incorporated into our maps feature the following classes: main roads, farming/crops, silviculture, and urban areas.Coordinate Reference SystemDatasets were projected to South America Albers Equal Area Conic, with all areal calculations based on this projection.Data UsageThe datasets are publicly available and should be cited appropriately (main publication: Ribeiro et al. 2020).Additional informationFor any other data requests or questions, please contact Fernanda Ribeiro (fernanda.ffr@gmail.com).
This page was built for dataset: Data from: Geographic Object-Based Image Analysis Framework for Mapping Vegetation Physiognomic Types at Fine Scales in Neotropical Savannas
