Eddies are vortices in the ocean formed by baroclinic instability, causing vertical circulation of materials and energy. Thus, it is important to understand their the scales, movements, and life cycles. In order to understand quantitatively, three-dimensional (3D) temperature and salinity profiles. Specifically, this study aims to analyze the Ulleung Warm Eddy (UWE) in the East Sea based on 3D reconstructed sutructures. For that, we used the 3-D subsurface data based on a Convolutional Neural Network (CNN) model, specialized in estimating output values by analyzing spatial relationships between input images. Inputs are sea surface temperature(SST), sea surface salinity(SSS), sea surface height(SSH), sea surface wind(SSW), and geographical information(longitude and latitude) from The Operational Sea Surface Temperature and Ice Analysis(OSTIA), the Copernicus Marine Environment Monitoring Service(CMEMS), and the European Center for Medium-range Weather Forecasts(ECMWF). The input data have a spatial resolution of 0.25° × 0.25° with a daily temporal resolution. The output is CTD temperature and salinity data from the National Institute of Fisheries Science (NIFS). As a result, a CNN model was employed to produce three-dimensional subsurface data from sea surface data with a spatial resolution of 0.25° × 0.25° with daily bases. In the test set, the model showed a root mean square error (RMSE) of 1.11 °C and 0.30 psu at a depth of 10 meters, with the highest RMSE of 1.69 °C and 0.20 psu in the thermocline at depths of 50–200 meters. Accordingly, the result allowed us to define the daily UWE's three-dimensional structure by extracting its boundary and employed wobbling ratio mathematically. The spatial extraction method involved two approaches: using a Lagrangian Particle Tracking Experiment (PTE) based on altimetry data and applying an empirical method to the derived subsurface data, based on the 10°C isotherm. Using the UWE's three-dimensional structure, we then analyzed changes in its scale, movement, and stratification to understand UWE’s life cycle.
 

暂无