Some Lessons Learned from Comparisons of Numerical Simulations and Observations of the

The Japan/East Sea (JES) is a large, multi-ported, semi-enclosed sea situated between the subtropical and subpolar zones. It exhibits most oceanic phenomena (e.g., wind-driven and buoyancydriven boundary currents, a subpolar jet and front, and mesoscale eddies) and processes (e.g., intense air-sea interaction, subduction and deep convection, and topographic trapping). JES circulation is driven by wind and thermohaline forcing, tides, and throughfl ow; however, this circulation is controlled largely by its bottom topography, especially by the large Yamato Rise in the center of the southern half, the large Japan Basin to the north, Ulleung Basin to the west, Yamato Basin to the east, and numerous seamounts. Infl ow is primarily through the Korea/Tsushima Strait in the south and the outfl ow is primarily through the Tsugaru and Soya Straits in the east; however, there is weak infl ow seasonally through the shallow Tatar/Mamiya Strait in the north. For these reasons, and due to its small size compared to an ocean basin, the JES is a convenient natural laboratory for numerical modeling and observation of ocean circulation phenomena and processes. Here, we summarize the implementation of a well-resolved (eddy-admitting) state-of-the-science community ocean model (Princeton Ocean Model [POM]; Mellor, 1998) and evaluate its results against ocean measurements. The model’s performance then encourages us to use it to break new ground by exploring the role of synoptic (weather-scale) atmospheric forcing, especially Siberian cold-air outbreaks, in producing wintertime deep convection.

(2) the infl owing Nearshore Branch The JES-POM used for our simulations has a meridional (north-south) grid size of 10 km and a zonal (east-west) grid size that changes from 10 km in the south to 7.5 km in the north. We used 26 sigma (terrain-following) levels in the vertical with fi nest resolution in the surface and bottom turbulent-boundary layers (Kang, 1997  Rotary spectra (diagrams representing clockwise-and counterclockwise-turning current energy as a function of frequency) for observed currents (red lines) and model outputs (black lines) with three diff erent model forcing fi elds. Th e "mont" and "empm" fi elds are monthly averages, and the "syn" fi eld has time resolution good enough to resolve individual storm passages. Th e important result here is that the model only does a good job for periods of 10 days or less if the forcing fi elds resolve weather events. Th e currents were measured from August 1993-July 1996 at 1-km depth in 3.5 km of water at 41.495°N, where the inertial frequency is 1.33 cycles per day. spheric variables and monthly MCSST (mont model run). Detailed procedures for preparation of these air-sea fl uxes can be found in Kang (2001).
Model/Data Comparison: CREAMS I Current Spectra (August 1993to July 1996 Japanese colleagues    Senjyu and Sudo (1994;red lines), Isobe and Isoda (1997; blue lines), and Senjyu (1999;green lines). Th e other three panels show modeled (driven by weatherresolving winds) currents at 800-m depth for February 15 of 1993, 1995, and 1997. Th e model repeatedly reproduces important features such as the southward fl ow along the western boundary and the eastward fl ow near 40°N.
Comparisons to simulated instantaneous fl ows at 800-m depth for February 15, 1993February 15, , 1995February 15, , and 1997  Area V corresponds to the so-called "fl ux center" (Kawamura and Wu, 1998), the "subduction region" (Senju andSudo, 1994, 1996;Yoshikawa et al., 1999), and the "wintertime convection location" (Seung and Yoon, 1995). In addition to Area V, our simulations identifi ed two other areas as possible convection/ventilation sites: Area K (36-39°N, west of 132°E) and Area KB (near Korea Bay) (Kang and Mooers, 2005). Area K has not been previously considered as a potential source region for JES intermediate water, yet it is a ventilation region in the three atmospheric-forcing cases run in all years. This result led us to seek an explanation for the ventilation other than air-sea fl ux conditions. Because this is known to be an upwelling area, the formation of denser water is possible when this upwelled cool water meets the warm, saline water advected by EKWC. A possible explanation for . Simulated 1997 annual mean subduction rate (m yr -1 ) for the monthly mean (mont, empm) and weather-resolving (syn) wind-forcing cases. Red shading indicates subduction greater than 500 m yr -1 . Contour intervals are 100 m yr -1 for values less than 500 m yr -1 and 500 m yr -1 for values greater than 500 m yr -1 . Th e labeled boxes represent the Vladivostok area (V), the East Coast Korea area (K), and Korea Bay (KB). Th e "K" area had not previously been considered a location for active subduction, but these model results suggest that it deserves more consideration.
convection in Area KB is that subduction may be a consequence of a strong EKWC dominated by lateral induction without a steep horizontal MLD gradient (Kang and Mooers, 2005). Also, Area KB has been previously suggested as a possible convection area (Ryabov, 1994)  were specifi ed from 1/6-degree monthly transects interpolated from bimonthly data acquired at standard depths by NFRDI (Kim, 1996). The atmospheric forcing consisted of six-hourly wind stress and surface heat fl ux records (total heat fl ux and short wave radiation) from NOGAPS (Navy Operational Global Atmospheric Prediction System) on a onedegree grid for 1999 through 2001. The simulated temperature and salinity were relaxed to initial values with depth-dependent weighting .   Simulated Salinity Figure 8. Observed (upper panels) and simulated (lower panels) temperature (left panels) and salinity (right panels) obtained along the red path in Figure 7. Note how the model replicates the general seasonal patterns in both temperature and salinity.
four straits. It will be valuable to run models with ca. htm). New model studies are now able to tackle, for example, the strong cyclonic subpolar gyre, which is evidenced in both observations and simulations over the Japan Basin, but interestingly, this gyre has yet to be studied in its own right. Such studies may be key to future understanding of the dynamics of the JES and developing predictive models.