*4.2. Topographic Contribution to Habitat Classification*

At the landscape scale, the contribution of the morphometric predictors surpassed the classification performance. Although the basic RGB performs well, as soon as morphometric predictors are added, the OA increases (from +1.28% to +13%). The positive effect of the slope raster and its by–products on the landscape classification were evidenced in several studies based on satellite sensors [30] as well as in UAV studies [31]. The classification test with the addition of the NIR band showed the relevance of a classification with morphometric variables. Indeed, when these variables were added to the RGB, the classification performance on the coastal fringe increased considerably contrary to the multispectral RGB + NIR combination without any other predictors.

At the ecosystem class level, the morphometric predictors produced different results depending on the coastal habitat type. For example, the topographic variables provided successful results for salt marshes. This particular ecosystem is located in a geographical area that is sheltered from strong prevailing marine currents. Coastal erosion has minimal or no impact on the meadows, allowing the different plant species to grow.
