Intégration de données topographiques et hydrographiques en vue de la localisation des zones humides potentielles de fond de vallée : cas d'un périmètre de la région du Bélier en Côte d'Ivoire

In the region of Belier (Côte d'Ivoire), there are wide areas of wetlands, with some of which began to be exploited in rice. In this study, the work consists in developing a prediction model to localize the potential wetlands of valley bottom in an area of the region of Belier. This prediction model was calibrated with topographic and hydrographic criterions of 2500 pixels which were determined using a machine learning procedure. The overall accuracy and the Kappa index obtained after the evaluation of the model are, respectively, 82.08 % and 82.05 %. Applied to whole the study area, the prediction model permits to mapping wetlands as well as it's reflected in the field reality described by 2425 punctual data of observation (overall accuracy = 69.23 % and Kappa index = 69.19 %) and a land use map.