Shelf water streamers: a potential pathway for ocean carbon sequestration in the Atlantic and Pacific Eastern Boundary Upwelling Systems

Ocean carbon uptake is essential to the global carbon cycle, as the ocean absorbs about one third of the atmospheric carbon released by human activities. Central to this process is the biological carbon pump, fueled by phytoplankton primary production. This pump transports organic carbon from surface waters to the ocean's deeper layers, where it can remain sequestered for hundreds to thousands of years. Coastal shelf regions, with their shallow, nutrient-rich waters, play a particularly active role in this cycle, supporting higher productivity levels compared to the deeper open ocean. However, in these shallow waters, organic carbon is more likely to be remineralized before it reaches the deep ocean. In these coastal zones, submesoscale filaments—or streamers—form through the instability of coastal currents, creating long, narrow structures that concentrate phytoplankton and chlorophyll. These streamers enhance long-distance transport of organic material, carrying chlorophyll-rich waters from the productive shelf regions into the open ocean, potentially increasing carbon flux to deeper ocean layers. A systematic estimation of how much carbon is transported by these structures is missing from the literature, due to the challenges in detecting and measuring streamers, which exhibit strong time variability over scales of several days. To address this gap, our work proposes a K-means based framework to detect streamers from chlorophyll and sea surface temperature satellite data in the Pacific and Atlantic Eastern Boundary Upwelling regions, and estimate their associated lateral carbon transport from satellite ocean color data. By estimating this cross-shelf export, we aim to deduce the potential sequestration rate, assuming that offshore high-chlorophyll streamers might increase the sinking fluxes of organic carbon to the deep ocean. This method relies solely on satellite products and can be operationalized for constant monitoring of the process in the years to come.