Improved representation of the Kuroshio extension through high-resolution dynamical downscaling: implications for future climate projections

Abstract

The Kuroshio extension (KE) plays a key role in the North Pacific climate through its strong heat transport and wintertime air–sea fluxes. Low-resolution Coupled Model Intercomparison Project Phase 6 (CMIP6) global climate models, however, continue to show structural biases such as unrealistic Kuroshio overshooting. To investigate how these biases compromise North Pacific climate projections, we perform high-resolution (1/8°) dynamical downscaling ensemble simulations using the Regional Ocean Modeling System (ROMS). The simulations are driven by seven CMIP6 models under four Shared Socioeconomic Pathway scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) from 1972 to 2100. During the historical period (1995–2014), the CMIP6 ensemble mean (EM) shows pronounced warm sea surface temperature (SST) and excessive surface net heat flux (NHF) biases in the overshooting KE region (OKE). The ROMS EM markedly mitigates both biases, reducing them by about 94% and 72%, respectively. In the late twenty-first-century projections (2081–2100), the CMIP6 EM exhibits amplified and spatially expanded overestimations of SST and NHF in the OKE compared with the ROMS EM. These overestimations are particularly pronounced under higher-emission scenarios and during winter. This scenario-dependent intensification is closely linked to overestimated ocean heat transport resulting from the persistent Kuroshio overshooting bias, which is absent in the high-resolution ROMS simulations. Our findings demonstrate that coarse-resolution Kuroshio pathway biases can distort future regional climate projections, highlighting the importance of high-resolution dynamical downscaling for more physically consistent North Pacific climate projections.