Land use and land cover mapping in wetlands one step closer to the ground: Sentinel-2 versus landsat 8.

Environmental studies with Landsat images have revealed many of the problems faced by wetland ecosystem, which are crucial for the conservation of biodiversity and the natural values of our planet. The study of LULC changes in wetlands through remote sensing constantly helps to identify and combat their main environmental threats improving the conservation of these natural habitats. Starting in mid-2015, the Sentinel-2 satellite opens new possibilities in the field of earth observation thanks to its higher spatial, spectral and temporal resolution becoming a powerful source of information for LULC monitoring in wetland areas. However, researchers may ask them selves to what extent Sentinel-2 is an improvement over Landsat 8 for general purposes. This research test if there is a real difference in the quality of the results delivered by both Sentinel-2 and Landsat 8 imagery when basic classification methods are applied. The study uses Sentinel-2 and Landsat 8 imagery to produce LULC maps in a Mediterranean wetland area applying an object based classification method in order to compare the accuracy and reliability in the surface detected by both satellites. The results show that an object based classification using only the Sentinel-2 and Landsat 8 image information, without band indexes or ancillary data, offers very similar results for most LULC classes, being the overall accuracy around 87-88% with slightly better results when using Sentinel-2. Although using Sentinel-2 leads to an increase in file size and processing times, the analysis of certain LULC classes presents an improvement compared to Landsat 8, detecting more linear and small size elements with a better delineation of image features in the classified map. However, these improvements should not underestimate the value of Landsat imagery in the future since both satellites provide high precision information, so they can and should coexist and be used together to increase data availability in order to have the best possible results in remote sensing research.