Improvement of Ocean Circulation Prediction System: Data assimilation, Hybrid Vertical coordinate configuration, and Sub-Grid Scale Topography Parameterization Inseong Chang Division of Earth Environmental System Sciences

Abstract

Abstract

The Northwest Pacific, encompassing diverse marginal seas and major western boundary currents such as the Kuroshio and Oyashio, is a region of intense oceanic variability and complex circulation. The recent acceleration of sea surface warming and increased frequency of extreme marine events, such as heatwaves, has highlighted the urgent need for more robust and accurate ocean prediction systems. To response this needs, the Korea Institute of Ocean Science and Technology (KIOST) has developed the Korea Operational Oceanographic System–Ocean Predictability Experiment for the Marine Environment (KOOS-OPEM, which has demonstrated promising performance. However, recent evaluation (Jin et al., 2024) has revealed persistent limitations in KOOS-OPEM's predictive skills, particularly in the Kuroshio Extension, where strong mesoscale eddy activity and intricate air–sea interactions continue. In order to address these limitations and enhance predictive performance, this dissertation established four coordinated strategies: (1) assessing the impact of different observational datasets, including satellite and in-situ observation, on ocean state estimation; (2) producing and evaluating the high-resolution regional reanalysis KOOS-OPEM Reanalysis version 2022 (K-ORA22) against global reanalysis products to diagnose data assimilation system; (3) updating the dynamic core to MOM6 with advanced vertical coordinate systems; and (4) implementing a porous barrier scheme to better represent sub-grid scale topographic effects. In first strategy, Sensitivity experiments with observation data revealed that assimilating satellite-derived pseudo-profiles significantly reduced errors in the subsurface temperature and salinity as well as representation of the Kuroshio axis, while regional in-situ temperature profiles improved subsurface temperature and salinity fields in the East Sea and along the Kuroshio pathway. This result highlights the importance of assimilating satellite and regional in-situ observation data for improving ocean state estimation in NWP. In addition, as Jin et al. (2024) has not yet incorporated assimilation scheme for SSH data, this result suggests that doing so may help overcome persistent limitation in Kuroshio and its extension region. In second strategy, the K-ORA22 reanalysis demonstrated skill in capturing key regional features such as the Yellow Sea Cold Water Mass and the East Sea Intermediate Water, although challenges remain in representing large-scale sea surface height variability and mesoscale eddies in the Kuroshio Extension. These limitations stem from challenges in representing background error covariance arising from the use of a static ensemble in Ensemble Optimal Interpolation. These findings suggest that future improvements should consider adopting multiscale data assimilation approaches or time-varying ensemble-based methods to better capture background error covariance, particularly in the Kuroshio Extension region. In third strategy, the dynamical core of KOOS-OPEM was updated from MOM5 to MOM6 enabled the comparison of hybrid isopycnal–z* (HYBRID) and pure z* (ZSTAR) vertical coordinates. As a result of comparisons with reanalysis and observation data, the Kuroshio separation points in both configurations closely align with the reanalysis data without data assimilation. However, the EKWC tends to overshoot in both models compared to reanalysis data, with ZSTAR exhibiting a more pronounced northward separation. Notably, the HYBRID configuration suppressed spurious diapycnal mixing and enhanced stratification, resulting in more accurate simulations of water masses such as the North Pacific Intermediate Water, as well as improved representation of both barotropic and baroclinic tidal dynamics, particularly in the Yellow Sea. However, its performance declined in weakly stratified, high-latitude regions, highlighting the need for region-specific adjustments. Finally, the implementation of a porous barrier parameterization enabled more physically realistic representation of unresolved topographic constraints, particularly in narrow straits and along steep bathymetric slopes such as the Izu-Ogasawara Ridge. This scheme improved surface and subsurface tracer distributions, enhanced eddy kinetic energy in the Kuroshio Extension, and corrected volume transport through key straits by mitigating spurious flow blocking and steering. Notably, regionally targeted applications—especially in the Kuroshio region—successfully suppressed unrealistic meanders and restored stratification and potential vorticity balance in the recirculation zone. However, uniform application in shallow, frictionally dominated coastal areas, such as the Yellow Sea, led to excessive restriction of lateral exchange and degraded tidal amplitude simulations. These results highlight the importance of selective, depth-aware implementation and suggest that the porous barrier approach offers a computationally efficient strategy for improving the performance of both regional prediction systems and coarser-resolution global ocean and climate models. Together, these strategies establish a comprehensive foundation for next-generation ocean forecasting in the Northwest Pacific by integrating improved use of observational data, enhanced data assimilation systems, and model advancements including a transition to a modern dynamical core and the implementation of a sub-grid scale topography parameterization. The developments presented in this dissertation contribute toward more accurate simulations of regional circulation, thermohaline structure, and mesoscale dynamics, thereby enhancing the capacity of KOOS-OPEM to support operational oceanography and climate adaptation efforts in the Northwest Pacific.