Spectral analysis of wetlands in Newfoundland using Sentinel 2A and Landsat 8 imagery

Wetlands play a key role in providing food, water, and shelter to a multitude of plants and animals. While much of Island of Newfoundland, which is located within Canada, is covered by wetlands, they have not been studied sufficiently in the province. Therefore, the study of these wetlands' characteristics is an important task in the province. In this regard, remote sensing satellites provide useful and accurate information. In this study, the spectral characteristics of five wetland types, including bog, fen, marsh, swamp, and shallow water, in a pilot site in Newfoundland were analyzed. This study used the data acquired by the two satellites, Sentinel 2A and Landsat 8. According to the analyses, the best spectral bands for wetlands discrimination and classification were selected. Then, the optimum bands were inserted into an object-based Random Forest algorithm to classify wetlands in the study area. The overall classification accuracy was 84% with a Kappa Coefficient of 0.77.