Delineation of Upwelling along the Northwest African coast using Attention-Coastal_up-Net.

Abstract

The area along the north-western coast of Africa, spanning from $12^{\circ} {N}$ to $28^{\circ} {N}$ latitude and from $12^{\circ} {E}$ to $24^{\circ} {E}$ in longitude, is distinguished by a consistent and fluctuating upwelling phenomenon that occurs nearly throughout the entire year. This upwelling is a critical oceanographic phenomenon that facilitates the movement of nutrients from the deep ocean to the surface, playing an essential role in boosting primary productivity and significantly impacting the health and sustainability of coastal ecosystems in this area. This study introduces Attention-Coastal<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">up</inf>-Net, a novel deep learning framework, designed to enhance the monitoring of upwelling systems along the Northwest African coastline. By leveraging sea surface temperature (SST) data and incorporating advanced attention mechanisms, this model provides precise segmentation and analysis of upwelling zones. Its effectiveness is demonstrated through improved accuracy and a higher upwelling validation index compared to traditional methods. This research provides a better understanding of ocean dynamics and offers valuable information for marine ecosystem management and environmental monitoring.