EVALUATING THE CONSISTENCY AND ACCURACY OF WETLAND MAPPING IN GLOBAL AQUATIC LAND COVER PRODUCTS
Summary
Wetlands are valuable ecosystems that are under threat due to the impact of human activities and climate change. Assessing the distribution of wetlands and their extent is relevant to manage these environments. However, the lack of a universal definition and the use of different input data and methodologies to map wetlands can complicate the assessment. This study aims to evaluate the consistency and accuracy of three selected global products: GLWD-3, LC-CCI, and CW-TCI. Specifically, it defines the consistency on a thematic, spatial, and temporal level, while the accuracy assessment was completed with an independent integrated reference dataset. A comparison among the products was completed to determine their consistency to a global extent, and a regional analysis was included to assess the consistency of the global products in wetlands hotspots such as the Amazon and Congo Basins with regional datasets. The accuracy assessment showed an overrepresentation of wetlands for all maps at the global extent. Furthermore, the results reveal some definition inconsistencies that created disagreement for LC-CCI, while CW-TCI was considered unreliable for mapping wetlands due to the high disagreement and low classification accuracies. GLWD-3 had better definition consistency but missed mapping wetland areas in the Amazon and Congo Basins while having a high disagreement and overestimation for the class wetlands. As a result of the assessment, despite the overrepresentation of wetlands by all products, LC-CCI is more suitable for contemporary studies and GLWD-3 is an old dataset that did not perform better than the other products, while CW-TCI is not a reliable dataset. Based on these results, it is necessary for users to have a prior understanding of the wetland definition before choosing an appropriate product that meets their needs. The LCCS-based aquatic land cover characterization framework presented recently is recommended as a possible solution to improve the differences between aquatic land cover and generic land cover mapping. Lastly, researchers need to continue their assessments on products to highlight disagreements and determine their consistency over time.