High-resolution North Pacific climate changes using dynamical downscaling: impact of improved Kuroshio

We conducted high-resolution (1/8°) dynamical downscaling over the North Pacific to produce ocean climate simulations for 1982–2100. The Regional Ocean Modeling System (ROMS) was forced by seven top-ranked CMIP6 global climate models (GCMs) under four SSP scenarios (SSP1–2.6, SSP2–4.5, SSP3–7.0, and SSP5–8.5), which provided the required forcing fields. Evaluation of sea surface temperature (SST) over the recent 20 years (1995–2014) shows that the ROMS ensemble mean (EM) substantially reduces the warm bias present in the GCM EM, improving the RMSE by 10.1%, with particularly strong improvement in the subpolar region (17.5%). These SST improvements primarily result from a more realistic representation of the Kuroshio, which alleviates the unrealistic overshooting in coarse-resolution GCM simulations, and are accompanied by improved wintertime net surface heat flux (NHF) near the Kuroshio path. Future projections (2081-2100) reveal pronounced differences between the GCM EM and ROMS EM in the subpolar region. Although both EMs project a strengthened and northward-shifted Kuroshio under higher-emission scenarios, the GCM EM exhibits an excessively large poleward shift. As a result, the GCM EM projects exaggerated, scenario-dependent wintertime changes in SST and NHF, which are substantially mitigated in the ROMS EM. These results highlight the importance of high-resolution regional ocean modeling for reducing biases in western boundary current systems and improving the reliability of future ocean climate projections.