### North Pacific Ocean model, NPO model

The North Pacific Ocean (NPO) model is based upon the Princeton Ocean Model (POM; Blumberg and Mellor, 1987). The model covers the North Pacific Ocean from 16$^{\circ}$S to 60$^{\circ}$N latitude and 99$^{\circ}$E to 77$^{\circ}$W longitude with a non-uniform resolution in horizontal (See the figure below). The grid size decreases from 40 km at the equator to 20 km on the northern boundary. There are 26 sigma levels in the vertical. The topography of the NPO model is blended from the digital bathymetry version-5 model (TaiDBMv5) of the National Center of Ocean Research in Taiwan and the 5-min elevation data (ETOPO5) of the National Geophysical Data Center [Edwards, 1988]. The lateral boundaries of this model are all closed.

The NPO model is driven by the monthly climatologic wind stress reanalyzed by the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) during spin-up. After spinning up from rest for 50 years, the model is continuously forced by the NCEP/NCAR reanalysis version 1 wind data from 1948 to 1978 and version 2 wind data from 1979 to 2005. The monthly climatologic SST data derived from NCEP/NCAR reanalysis are used as the surface temperature boundary condition during the spin-up period. The 6-h advanced very high resolution radiometer SST ($2.5^{\circ}\times2.5^{\circ}$) from NCEP/NCAR is used for the period between 1948 and 1981. From 1982 to 2005, the weekly SST data determined by optimal interpolation with a spatial resolution of $1^{\circ}\times1^{\circ}$ from the Data Support Section at the Computational and Information Systems Laboratory of NCAR (NCAR data are available at http://dss.ucar.edu/) are used. The purpose of this model is to provide proper boundary data to the small-scale East Asian Marginal Seas (EAMS) model.

### East Asian Marginal Seas model, EAMS model

The East Asian Marginal Seas (EAMS) model is based upon the Princeton Ocean Model (POM; Blumberg and Mellor, 1987). The EAMS model uses realistic bathymetry over an expanded domain including the upstream and downstream regions of the Kuroshio, the South China Sea, and the East China Sea but not the Kuroshio extension region. The bathymetry used in this model is TaiDBMv5/ ETOPO5 blend topography. The horizontal grid size is $1/8^{\circ}$ in the EAMS model, and there are 26 vertical sigma levels.

On open boundaries, the EAMS model derives its values from the NPO model using one-way coupling. From Flather [1976], the vertically-averaged normal velocity on the open boundaries of the EAMS model, , is derived from the corresponding velocity estimate from the NPO model, $\bar{u}_{n}^{0}$ , using $\bar{u}_{n}=\bar{u}_{n}^{0}+\sqrt{\frac{g}{H}}\left (\eta-\eta^{0}\right )$, where $\eta$ is the sea surface height in the EAMS model, $\eta^0$ is the NPO model-derived sea surface height, $H$ is the water depth on the open boundary, and $g$ is the gravitational constant. The sea surface height $\eta$ is located half a grid inside the open boundary in the EAMS model, while $\eta^{0}$ is located on the open boundary of the EAMS model. Baroclinic velocities on the open boundaries of the EAMS model are determined using the inflow scheme by Mellor [2004]: the daily baroclinic velocity from the NPO model is spatially interpolated and assigned to the open lateral boundary grids of the EAMS model. Temperature and salinity on the open boundaries are calculated using an upstream advection scheme [Mellor, 2004]: daily NPO profiles of temperature and salinity supply the upstream values in case of inflow.

During spin-up, the EAMS model is initialized by the temperature and salinity fields of the NPO model output in January 1980 and is under climatology forcing for two years. The EAMS model is then forced by the 6-hourly NCEP/NCAR reanalysis version 2 wind stress ($2.5^{\circ}\times2.5^{\circ}$) at the sea surface and by the open boundary values provided by the NPO model. This model has been used for the studies on the current in the Taiwan Strait and the Kuroshio east of Taiwan.

#### Reference

Hsin, Y.-C., C.-R. Wu, and P.-T. Shaw (2008): Spatial and Temporal Variations of the Kuroshio East of Taiwan, 1982-2005: A numerical study. Journal of Geophysical Research: Oceans, 113(C4), C04002, doi:10.1029/2007JC004485.

Wu, C.-R. and Y.-C. Hsin (2005): Volume transport through the Taiwan Strait: a numerical study. Terrestrial, Atmospheric and Oceanic Sciences, 16(2), 377-391.

### South China Sea model, SCS model

The South China Sea (SCS) model used here is the sigma-coordinate Princeton Ocean Model (POM; Blumberg and Mellor, 1987). The three-dimensional, free surface model solves the primitive equations for momentum, salt, and heat. It includes a 2.5-level turbulence closure by Mellor and Yamada (1982), and the horizontal mixing by Smagorinsky (Smagorinsky, 1963). Figure shows the SCS model domain with realistic bathymetry. The horizontal grid size is $1/16^{\circ}$, and there are 26 sigma levels in the vertical. At the open boundaries, the SCS model derives its boundary condition from a larger-scale East Asian Marginal Seas (EAMS; Wu and Hsin 2005) model. The EAMS model is also based on the POM, and has a horizontal resolution of $1/8^{\circ}$ and 26 sigma levels. The EAMS model domain extends from 99$^{\circ}$E to 140$^{\circ}$E in longitude, and from 0$^{\circ}$N to 42$^{\circ}$N in latitude. A detailed description of the EAMS model is given by Wu and Hsin (2005). The EAMS model has been validated with observed temperature and salinity data in the South China Sea, and with observed velocity data from ADCPs in the Taiwan Strait (Wu and Hsin, 2005).

The POM uses the mode splitting technique, in which the vertically integrated governing equations (barotropic, external mode) are separated from the equations governing vertical structure (baroclinic, internal mode) . The one-way coupling between the SCS and EAMS models is described below. The vertically averaged barotropic velocities on the open boundaries of the EAMS model are estimated by the Flather (1976) formulation: $\bar{u}_{n}=\bar{u}_{n}^{0}+\sqrt{\frac{g}{H}}\left (\eta-\eta^{0}\right )$.

$\bar{u}_{n}$ is the vertically averaged outward normal component of the velocity on the open boundary of the SCS model at time $t$, $\bar{u}_{n}^{0}$ is the vertically averaged normal component of the velocity on the open boundary at time $t$, estimated from the EAMS model. The model sea surface height $\eta$ is calculated from the continuity equation, and is located half of a grid inside the open boundary of the SCS model domain. The EAMS model sea surface height $\eta^{0}$ is located on the open boundary of the SCS model. The water depth of the open boundary is $H$, and $g$ is the gravitational acceleration. Baroclinic velocities on the open boundaries of the SCS model are determined using an inflow condition; daily baroclinic velocities from the EAMS model are spatially interpolated and assigned to the open lateral boundary grids of the SCS model. Temperature and salinity on the open boundaries are subject to upstream advection and, in case of inflow daily EAMS profiles of temperature and salinity supply the upstream values.

The SCS model was initialized by the temperature and salinity fields of the EAMS model outputs of January 1999, and thereafter was subject to climatological forcing for one year. After the spin-up period, the SCS model was forced with NASA Quick Scatterometer/NCEP (QSCAT/NCEP) wind data sets. The blended QSCAT/NCEP wind stress data set is one of the most up-to-date high-resolution data of ocean surface winds. We adopted six-hourly fields of zonal and meridional wind components, 10 m above sea level and with a resolution of $0.5^{\circ}\times0.5^{\circ}$. These fields are derived from a space and time blend of QSCAT-DIRTH satellite scatterometer observations and NCEP analyses (Milliff et al., 1999). The SCS model was subject to wind stress at the sea surface and forcing at the open ocean boundary (as described above) provided by the EAMS model which also was driven by the QSCAT/NCEP wind forcing. The simulation period is from 1999 to 2003.

#### Reference

Chiang T. -L., C.-R. Wu, and S.-Y. Chao (2008): Physical and Geographical Origins of the South China Sea Warm Current. Journal of Geophysical Research: Oceans, 113(C8), C08028, doi:10.1029/2008JC004794.

Wu, C.-R. and T.-L. Chiang (2007): Mesoscale eddies in the northern South China Sea. Deep Sea Research II, 54(14-15), 1575-1588, doi:10.1016/j.dsr2.2007.05.008.

### Seas Around Taiwan model, SAT model

The Seas Around Taiwan (SAT) model is derived from the sigma-coordinate Princeton Ocean Model (POM; Blumberg and Mellor, 1987). The three-dimensional, free surface model solves the primitive equations for momentum, salt and heat. It includes a 2.5-level turbulence closure submodel developed by Mellor and Yamada (1982), and the Smagorinsky (1963) formulation for horizontal mixing. Figure shows the SAT model domain, from 110.5$^{\circ}$E to 126$^{\circ}$E and from 13.5$^{\circ}$N to 28$^{\circ}$N, with realistic bathymetry. The horizontal grid size is $1/20^{\circ}$, and there are 26 sigma levels in the vertical. On open boundaries, the SAT model derives its boundary condition from a larger-scale East Asian Marginal Seas (EAMS) model. The EAMS model is also based on the POM, and has a horizontal resolution of $1/8^{\circ}$ and 26 sigma levels. The EAMS model domain extends from 99$^{\circ}$E to 140$^{\circ}$E in longitude, and from 0$^{\circ}$N to 42$^{\circ}$N in latitude. A detailed description of the EAMS model has been given by Wu and Hsin (2005). The EAMS model has been validated with observed volume transport of Kuroshio east of Taiwan (Hsin et al., 2008), and corroborated with observed velocity data from both bottom-mounted and shipboard ADCPs in the Taiwan Strait (Wu and Hsin, 2005).

The POM uses the mode splitting technique to solve the depth-integrated governing equations (barotropic, external mode) and the equations governing vertical structure (baroclinic, internal mode). Boundary conditions are separately formulated for the barotropic and baroclinic modes, and then adjusted to take into account the different truncation errors for those modes (Blumberg and Mellor, 1987). The one-way coupling between the TS and EAMS models is as follows. Following Flather (1976), the vertically averaged velocity on open boundaries of the TS model is $\bar{u}_{n}=\bar{u}_{n}^{0}+\sqrt{\frac{g}{H}}\left (\eta-\eta^{0}\right )$, where $\bar{u}_{n}$ is the vertically averaged velocity normal to open boundaries of the SAT model at time $t$, and $\bar{u}_{n}^{0}$ is the corresponding velocity estimated from the EAMS model. The model sea surface height $\eta$, derived from the continuity equation, is located half of a grid inside the open boundary of the SAT model domain. The EAMS model sea surface height $\eta^{0}$ is located on the open boundary of the SAT model. The water depth on the open boundary is $H$, and $g$ is the gravitational acceleration. Baroclinic velocities on open boundaries of the SAT model are determined using an inflow condition; daily baroclinic velocities from the EAMS model were spatially interpolated and assigned to the open lateral boundary grids of the SAT model. Temperature and salinity on the open boundaries are subject to advection from upstream; in the case of an inflow, daily EAMS profiles of temperature and salinity provide the upstream values.

The SAT model was initialized by temperature and salinity fields of the EAMS model outputs in January 1999, and thereafter subject to climatological forcing for 1 year. After the spin-up period, the SAT model was forced with NASA Quick Scatterometer (QSCAT/NCEP) wind data sets. The blended QSCAT/NCEP wind stress data set is one of the most up-to-date high-resolution data of ocean surface winds at the present time. We adopted 6 hourly maps of 10 m winds at a resolution of $0.5^{\circ}\times0.5^{\circ}$. These fields are derived from a space and time blend of QSCAT-DIRTH satellite scatterometer observations and NCEP analyses (Miluliff et al., 1999). The SAT model was subject to 6 hourly wind stresses at the sea surface and open ocean boundary forcing (as described above) provided by the EAMS model. The simulation period spans from 1999 to 2003.

#### Reference

Wu, C.-R., H.-F. Lu, and S.-Y. Chao. (2008): A numerical study on the formation of upwelling off northeast Taiwan. Journal of Geophysical Research: Oceans, 113(C8), C08025, doi:10.1029/2007JC004697.

### Taiwan Strait model, TS model

The Taiwan Strait (TS) model is formulated on the basis of the sigma-coordinate Princeton Ocean Model (POM; Blumberg and Mellor, 1987). The three-dimensional model solves the primitive equations for momentum, salt, and heat. It includes a 2.5-level turbulence closure sub-model developed by Mellor and Yamada (1982), and the Smagorinsky (1963) formulation for horizontal mixing. Figure a and b shows the orthogonal, curvilinear grid system and model domain. The horizontal grid size varies from 3 to 10 km, with finer resolution located near the center of the Strait. The model has 26 vertical sigma levels. On open boundaries, the TS model derives its boundary condition from a larger-scale East Asian Marginal Seas (EAMS) model, which also adopts the POM formulation with a horizontal resolution of $1/8^{\circ}$ and 26 sigma levels. The EAMS model domain extends from 99$^{\circ}$E to 140$^{\circ}$E in longitude, and from 0$^{\circ}$N to 42$^{\circ}$N in latitude. A detailed description of the EAMS model has been given by Wu and Hsin (2005). The EAMS model has been validated with observed temperature and salinity data in the South China Sea and corroborated with observed velocity data from both bottom-mounted and shipboard ADCPs in the Taiwan Strait.

The POM uses the mode splitting technique to solve the depth-integrated governing equations (barotropic, external mode) and the equations governing vertical structure (baroclinic, internal mode) separately. The one-way coupling between the TS and EAMS models is as follows. Following Flather (1976), the vertically averaged velocity on open boundaries of the TS model is $\bar{u}_{n}=\bar{u}_{n}^{0}+\sqrt{\frac{g}{H}}\left (\eta-\eta^{0}\right )$, where $\bar{u}_{n}$ is the depth-averaged velocity normal to open boundaries of the TS model at time $t$ and $\bar{u}_{n}^{0}$ is the corresponding velocity at time $t$, estimated from the EAMS model. The model sea surface height $\eta$ is located half of a grid inside the open boundary in the TS model domain. The EAMS model sea surface height $\eta^{0}$ is located on the open boundary of the TS model. The water depth on the open boundary is $H$, and $g$ is the gravitational acceleration. Further, the zero normal gradient condition for sea level is used on all open boundaries. Baroclinic velocities on open boundaries of the TS model are determined using an inflow condition; daily baroclinic velocities from the EAMS model are spatially interpolated and assigned to the open boundary grids of the TS model. Temperature and salinity on the open boundaries are subject to upstream advection and, in case of inflow, daily EAMS profiles of temperature and salinity supply the upstream values.

The TS model was initialized with the temperature and salinity fields of the EAMS model outputs in January 1999 and thereafter subject to climatological forcing for one year. After the spin-up period, the TS model was forced with QSCAT/NCEP wind data sets. Having a temporal resolution of 6 hours and a spatial resolution of $0.5^{\circ}\times0.5^{\circ}$, the blended QSCAT/NCEP wind stress data set is one of the most up-to-date, high-resolution datasets of ocean surface winds at the present time. We adopted 6-hourly fields of zonal and meridional wind components 10 m above sea level with a resolution of $0.5^{\circ}\times0.5^{\circ}$. These fields are derived from a space and time blend of QSCAT-DIRTH satellite scatterometer observations and NCEP analyses (Milliff et al., 1999). The TS model was subject to 6-hourly wind stress at the sea surface and open ocean boundary forcing (as described above) provided by the EAMS model. The AVHRR sea surface temperature is specified at the sea surface. The simulation period is from 1999 to 2003.

#### Reference

Wu, C.-R., S.-Y. Chao, and C. Hsu (2007): Transient, seasonal and interannual variability of the Taiwan Strait Current. Journal of Oceanography, 63(5), 821-833, doi:10.1007/s10872-007-0070-1.