Different Types of Surface Chlorophyll Patterns of Oceanic Mesoscale Eddies Identified by AI Framework.

Oceanic mesoscale eddies (with scale 101–102 km) and their submesoscale fine structures (with scale 100–101 km) can effectively induce vertical motions and bring nutrients into the oceanic euphotic layer, which leaves abundant footprints on the ocean surface chlorophyll distributions and have the potential to promote primary productivity of oceanic ecosystem. In return, these surface chlorophyll footprints observed by ocean color satellites can serve as a useful tool to reveal the spatial structures of mesoscale eddies and their submesoscale fine structures. By combining artificial intelligence (AI) algorithms to develop a series of identification strategies for typical surface chlorophyll patterns around mesoscale eddies, we find that over 20% of mesoscale eddy observations exhibit identifiable typical chlorophyll patterns, which tends to regulate an increase of the surface chlorophyll concentration within the corresponding eddies, especially enhancing by about 30% in nutrient‐restricted subtropical regions compared with the background values. Based on their geometric features, typical chlorophyll patterns are primarily classified as Core, Spiral, Tail, Ring, Loop, and Eye respectively by clustering algorithm. Further spatial‐spectral analysis found that the typical patterns on eddies exhibit a much steeper wave‐number spectral slope about −3, compared to the non‐typical distributions on eddies and the non‐eddy background distribution (about −2.7–−2.2). This implies that the occurrence of different typical chlorophyll patterns may correspond to specific mesoscale and submesoscale dynamic processes. Plain Language Summary: Mesoscale eddies and their submesoscale fine structures collectively play non‐negligible roles in sustaining nutrient supply and promoting productivity in the oceanic euphotic layer. Although submesoscale processes are believed to result in ageostrophic motions and allow forward oceanic energy cascade, the resolutions of traditional in‐situ ocean measurements and satellite altimetry observations are typically too coarse to resolve them. In this paper, we elaborate on artificial intelligence (AI) techniques to develop a series of typical surface chlorophyll pattern identification strategies. The AI framework automatically identified six types of typical chlorophyll patterns of mesoscale eddy, in the context of the generally comparable geometric characteristics that are potentially reflected from their underlined oceanic dynamics. Over 20% mesoscale eddy observations tend to exhibit identifiable typical chlorophyll patterns, which leads to a remarkable increase of surface chlorophyll concentration. The wave‐number spectrum of eddy‐induced typical patterns exhibit a significantly steeper slope, in comparison of the non‐typical distributions on eddies and the non‐eddy background distribution. Our results may help to establish parameterizations for the submesoscale processes of mesoscale eddies, or provide an observational baseline to verify high‐resolution physical‐biogeochemical coupled oceanic numerical models, or even improve the model simulations of oceanic primary productivity and carbon fixation capacity. Key Points: Six types of typical chlorophyll patterns on mesoscale eddies and their submesoscale fine‐structures are identified globally by artificial intelligence (AI) methodThe typical patterns tend to substantially enhance chlorophyll concentration of the corresponding eddiesThe wave‐number spectral slope of the typical chlorophyll pattern is significantly different from the background value [ABSTRACT FROM AUTHOR]