Elvina Alphna Mekui
Remote Sensing Research and Application Center (CARTEL)
University of Sherbrooke
Yacine Bouroubi, University of Sherbrooke
Mapping of riparian strip properties using earth observation data and deep learning
Riparian strips (RS) fulfill several ecological functions, including reducing the pollution of watercourses in agricultural environments. With the rise of remote sensing, fine spatial resolution has presented advantages for the characterization of RSs, allowing considerable reduction in the resources allocated to field campaigns. In Quebec, the River Bands Quality Index (IQBR) is made up of indicators that can be generated by geomatics techniques. However, these traditional IQBR classification and processing techniques are less suited to the very large amount of information available. Due to the opportunity offered by the new context of abundance of Earth observation (EO) data and advanced artificial intelligence (DL, Deep Learning) methods capable of exploiting both spatial dependencies -temporal data, this research proposes an innovative approach to processing earth observation data by deep learning for the mapping of riparian strip properties.