Movemricomwnp

MOVE/MRI. COM-WNP (Multivariate Ocean Variational Estimation system/Meteorological Research Institute Community Ocean Model—western North Pacific version) uses MRI.COM as ocean model. The layer thickness near the surface follows surface topography in this z-coordinate model (Hasumi 2006). For the nonlinear momentum advection, the generalized entropy-preserving scheme (Ara-kawa 1972) is used, which is based on the concept of diagonally upward/downward mass momentum fluxes along the sloping bottom. A biharmonic operator is used for horizontal turbulent mixing. A biharmonic friction with a Smagorinsky-like viscosity (Griffies and Hallberg 2000) is used for momentum. The vertical viscosity and diffusivity are determined by the turbulent closure scheme of Mellor and Blumberg (2004). The model domain spans from 117E to 160°W zonally and from 15°N to 65°N meridionally. The horizontal resolution is variable: it is 1/10° from 117°E to 160°E and 1/6° from 160°E to 160°W, and 1/10° from 15°N to 50°N and 1/6° from 50°N to 65°N. There are 54 levels in the vertical direction with thickness increasing 1 m at the surface to 600 m near the bottom. Oceanic states at the open boundaries are replaced by those from a North Pacific model (MOVE-NP) with a horizontal resolution of 1/2° (one-way nesting). A sea ice model with the thermodynamics of Mellor and Kantha (1989) and the elastic-viscous-plastic rheology of Hunke and Ducowicz (2002) is also applied. For more detailed information on the model see Tsujino et al. (2006). The analysis scheme adopted in MOVE is a multivariate three-dimensional variational (3DVAR) analysis scheme with vertical coupled Temperature-Salinity (T-S) Empirical Orthogonal Function (EOF) modal decomposition (Fujii and Kamachi 2003). In this system, the model domain is divided into 13 subregions and vertical T-S EOF modes are calculated from the observed T-S profiles for each subregion. The 3DVAR results are inserted into the model temperature and salinity fields above 1,500 m by the incremental analysis updates (Bloom et al. 1996). For more details see Usui et al. (2006). In situ temperature and salinity profiles, satellite sea surface height (SSH) anomaly and sea surface temperature (SST) are assimilated. The temperature and salinity data including ARGO data are obtained from Global Temperature-Salinity Profile Program (GTSPP: http://www.nodc.noaa.gov/GTSPP/). The SSHA data is the near-real time along-track data of Jason-1 and ENVISAT obtained from Archiving, Validation and Interpretation of Satellite Oceanographic data (AVISO: http://www. jason.oceanobs.com/). The SST data is the Merged Satellite and In-situ Data Global Daily SST (MGDSST) produced by JMA. The assimilation run is implemented every five days, and the forecasting period is one month. The model is driven by wind stress and heat flux from the JMA's Climate Data Assimilation System (JC-DAS) in assimilation run and forced by the result of the climate forecasting model in forecasting run.

17.2.3 JCOPE1,2

The ocean model in the JCOPE1,2 systems is based on the Princeton Ocean Model with a generalized sigma coordinate (POMgcs) (Mellor et al. 2002). A high-resolution regional model with a spatial grid of 1/12° and 47 vertical levels is embedded in a low-resolution model covering the North Pacific region (30S-62°N, 100E-90W) with a spatial grid of approximately 1/4° and 21 sigma levels. The inner model domain covers the western North Pacific (10.5-62N, 108-180E) and its lateral boundary conditions are determined from the basin-wide model using a one-way nesting method (Guo et al. 2003). The wind stress and heat flux fields are calculated from output of 6-hourly NCEP Global Forecast System (NCEPGFS) or NCEP/NCAR reanalysis (Kalnay et al. 1996). Salinity at the ocean surface is restored to the monthly mean climatology data (Levitus et al. 1994) with a time scale of 30 days. The following observation data are assimilated into the model: the sea surface height anomaly (SSHA) obtained from the TOPEX/Poseidon and ERS-1 satellites during September 1999 to June 2002 and from the Jason-1 and the Geosat Follow-On during June 2002 to present; sea surface temperature (SST) obtained from the Advanced Very High Resolution Radiometer/Multi-Channel Sea Surface Temperature (AVHRR/MCSST) Level 2 products; and vertical profile data of temperature and salinity obtained from GTSPP. The JCOPE1 system (Mi-yazawa et al. 2008a) uses a combination of optimum interpolation for horizontal gridding of SSHA, SST, subsurface temperature and salinity data and multivariate optimum interpolation for creation of three-dimensional grid data of temperature and salinity. The temperature and salinity data are introduced into the model using the Incremental Analysis Update (IAU) (Bloom et al. 1996). The JCOPE2 system (Miyazawa et al. 2008b) adopts the 3DVAR scheme with vertical coupled Temperature-Salinity Empirical Orthogonal Function modal decomposition (Fujii and Kamachi 2003).

17.2.4 Kyoto University System

The numerical model used is the ocean general circulation model developed at Kyoto University (e.g., Toyoda et al. 2004), which employs the hybrid o-z vertical coordinate to better simulate free surface motion of the ocean. To further enhance the representation of the upper ocean circulation, this model adopts some sophisticated parameterizations such as the Takano-Onishi scheme for the momentum equation (Ishizaki and Motoi 1999), the turbulence closure scheme for the mixed layer parameterization (Noh et al. 2005), the 3rd order scheme (Hasumi 2000) based on QUICKEST (Leonard 1979) for vertical advection, UTOPIA (Leonard et al. 1993) for horizontal advection and the isopycnal mixing scheme (Gent and McWilliams 1990; Griffies 1998). The model basin covers the northwestern North Pacific which includes the Sea of Japan and other marginal seas (see Fig. 17.1). The basic model resolution is 1/6° and 1/8° in longitude and latitude, respectively, and 78 vertical levels spaced from 4 m near the sea surface to 500 m at the bottom. Note that 67 vertical levels are set above 1,000 m depth. For the downscal-ing experiment focusing on the coastal circulation off Shimokita Peninsula, they employ the triple nesting approach from the above basic resolution (nest-1) to the finest resolution of 1/54° and 1/72° (nest-3) in longitude and latitude, respectively, via a medium resolution of 1/18° and 1/24° (nest-2). The nesting technique employed here is based on Oey and Chen (1992). The assimilated elements in this study are satellite-derived SST and sea surface height (SSH) data, and in-situ observation data of temperature and salinity. In detail, the SST data are from the New Generation Sea Surface Temperature (NGSST) produced at Tohoku University with a horizontal resolution of 1/20° and daily coverage. The SSH data are from the SSalto/Duacs gridded absolute dynamic topography, which is the sum of the sea level anomaly derived from multi altimeter data and mean dynamic topography (Rio and Hernandez 2004) provided by AVISO in time intervals of 3.5 days. The horizontal resolution is nearly 1/3° and the SSH data are interpolated onto the model grids for assimilation. The in-situ observation data are obtained from GTSPP. A 4D-VAR approach, optimized 4-dimensional datasets are sought by minimizing a cost function, in which the initial condition of model variables is chosen as the control variable. The cost function is composed of the background and the observational term. For more details, see Awaji et al. (2003) and Ishikawa et al. (2009).

17.2.5 NMEFC System for Western North Pacific

The NMEFC's operational forecast system for the western North Pacific (99°-150°E; 2°-45°N) uses POM with a horizontal resolution of 1/4° by 1/4° and 15 vertical levels. The model is driven by NMEFC's forecasts of a mesoscale atmospheric model. The data assimilation system utilizes OI for in situ observations (ship reports and Argo profiles) and a nudging scheme for MGDSST. The system produces a daily 3-day forecast of SST of the western North Pacific.

17.2.6 CAS Preoperational System

A Chinese shelf/coastal seas model based on a three-dimensional hybrid-coordinate ocean model (HYCOM, Bleck 2002; Chassignet et al. 2003) is used to simulate ocean circulation around China (Xiao and Zhu 2007). A curvilinear horizontal grid is utilized with an average horizontal resolution of 13 km. There are 22 layers in the vertical coordinate. Using the real topography, the model domain includes the whole CSCS and part of the West Pacific Ocean (see Fig. 17.1). The model was forced by ECMWF 6-hourly reanalysis dataset (Uppala et al. 2005). And the lateral open boundary conditions were provided by an India-Pacific domain HYCOM simulation (1/4° resolution). The data assimilation system has two alternatives: an EnOI scheme and the ensemble Kalman filter (Wan et al. 2008). The observations used for assimilation include GHRSST products, Argo profiles from GTSPP and SSHA from AVISO. Xie et al. (2008) compared several GHRSST products in the studied area.

17.2.7 YEOS

The Yellow Sea 3D ocean-ice prediction system is based on DMI (Danish Meteorological Institute) BSHcmod, which is a coupled hydrostatic 3D ocean-ice model based on primitive ocean equations, and a Hibler-type sea ice model (Dick et al. 2001). The baseline set-up operates on two nested, coupled grids, making possible to zoom in on an area of interest with high resolution. Multi-sensor SST data are assimilated into the model daily by using a simplified Kalman-Filter Scheme (Larsen et al. 2007). The ocean model is driven by atmospheric forcing, river runoff and lateral boundary conditions. Along open boundaries, sea level is specified as a combination of the tidal variation and wind/pressure driven surge. Temperature and salinity are obtained from monthly climatology. Outflowing water is stored in a buffer zone, which must be emptied before water of climatological properties are advected into the model domain. The hourly meteorological products in 7.5 km horizontal resolution are derived from DMI operational NWP model HIRLAM. The 3-D ocean model domain covers the Yellow Sea and part of the East China sea, north of 33^°N

and west of 127°E, with horizontal resolution of 1/20° (lat.) by 1/15° (lon.) and 30 vertical levels. This is then embedded in a coarse-grid, 2-dimensional external surge model of larger extent. The undisturbed top layer thickness is 6 m in order to avoid tidal drying of the top layer, in which case the model would not be stable. The layers below are of 2 m thickness, with increasing layer thickness towards the sea bed.

17.2.8 ESROM

ESROM (East/Japan Sea Regional Ocean Model), based on the GFDL MOM3 (GFDL Modular Ocean Model version 3; Pacanowski and Griffies 1999), has been driven by monthly mean wind stress from the ECMWF reanalysis and monthly mean heat flux calculated by bulk air-sea flux formulation using ECMWF reanalysis meteorological variables. The surface salt of the model has been restored to that of the WOA hydrographic data. For the open boundary condition, a radiation condition with a nudging term for inward boundary fluxes is applied for the tracers and barotropic currents (Marchesiello et al. 2001). The barotropic velocity through the Korea Strait is given by the volume transport monitored by the submarine cable. Using the 3-dimensional variational assimilation routine (Weaver and Courtier 2001), the satellite-borne sea surface temperature, sea surface height (SSH) anomaly and temperature profiles have been assimilated. Kim et al. (2009) could verify the performance of the ESROM using an independent measurement dataset by the Pressure-equipped Inverted Echo Sounder in the Ulleung Basin located in the western side of the East/Japan Sea and suggested that the ESROM well reproduced the mesoscale variability as well as the general circulation.

17.3 Highlighted Examples

17.3.1 Successful Prediction of the Kuroshio Large Meander

Some Japanese systems successfully predicted the Korushio large meander in 2004. Figure 17.2 shows the observations and forecasts of the Korushio main path in summer of 2004. The forecasts were made by MRI system.

17.3.2 Operational Forecast for 2008 Olympic Games

An operational demonstration of weather-ocean-wave forecast has been made during August-September 2008 for Beijing Olympic Games. The weather and ocean forecast were made by DMI and wave forecast was made by the First Institute of Oceanography (FIO) by using FIO wave model WAM forced by DMI on-line weather data. The general results are shown on YEOS website. An interesting event

ASSIM (01 JUN 2004) ASSIM (15 JUL 2004)

ASSIM (01 JUN 2004) ASSIM (15 JUL 2004)

100 cm/s

FCST (15 JUN 2004)

ASSIM (25 AUG 2004)

ASSIM (25 AUG 2004)

FCST (25 AUG 2004)

forecast

FCST (15 JUN 2004)

FCST (25 AUG 2004)

130E 134E 138E ^^Mlll II 111 11II111 Ml Mill

20 40 60 80 100 120 140

Fig. 17.2 The Kuroshio large meander formation in 2004. Top (bottom) panel shows the time sequence of the assimilated (predicted) current fields at 100 m depth. Vector and shade denote the current vector and its magnitude. The prediction stars on 1 June 2004 using the assimilated field as the initial condition happened on the 49er Medal course, at 16 in 17 August 2008. There is an abrupt change of the states of weather and sea. The mast of Danish sailors' sailing boat was broken due to a gust and rough sea. By borrowing a boat from the Croatian team, the Danes still won the gold medal. Figure 17.3 shows the forecast given by a fine resolution model (in a horizontal resolution of 2.5 km resolution) downscaled from YEOS forecast.

17.3.3 Prediction of Giant Jellyfishes

Recently large amount of the giant jellyfish came over from the East China Sea to the Japan Sea through the Tsushima Strait, and they migrated to the Japanese coast along the coastal and offshore branches of the Tsushima Warm Current. Especially in 2005 season, excessively large number of jellyfish was observed for a long period (from early July until March), and serious damages to fisheries occurred. To reproduce and predict the distribution of the jellyfish in the Japan Sea a numerical simulation for the migration of the Jellyfish in 2005 was carried out by using a data assimilation model. The assimilation model was the Japan Sea Forecasting System

Fig. 17.3 High resolution Wind forecast on 17 August 2008. Upper: winds time series (in Qingdao local time) at the race area of 49er Medal course of OL2008 in Qingdao; bottom: surface winds distribution at 2008-08-17-08GMT in Qingdao waters

Wind speed and direction at sation: race A

-5m/s

NW

W SW

SF

r

E NE N

™—■

08117 0BM7 08/18 08118 03j1B 08/18 12:00 18:00 00:00 00:00 12:00 18:00

Wind at 10m (m/s) 2008 AUG 17 OSgmt

Wind at 10m (m/s) 2008 AUG 17 OSgmt

developed by RIAM, Kyushu University (Hirose et al. 2007). Passive tracers were released from the both east and west channels of the Tsushima Strait at the depth of 0-22.5 m as artificial jellyfishes. Based on the sightings report around the Tsushima Island, the beginning of the tracer input was set on June 23. After August 16, the calculation was carried out by using predicted current data. The reproduction of the northward migration of the jellyfish was very successful; the northward front position of the jellyfish was in good agreement with sightings data within a few days, as shown in Fig. 17.4.

17.3.4 Reproducing the Model Transition of Coastal Waters off Shimokita Peninsula

Coastal waters off Shimokita Peninsula (see Fig. 17.5) with a complicated geographic feature well reflects the major features of short-term (a few days) to seasonal-

32 N

40 N

32 N

40 N

130E

136E

142E

130E

135E

140E

Fig. 17.4 Sightseeing observations of giant jellyfish (left) and tracer forecasts (right)

130E

136E

142E

130E

135E

140E

Fig. 17.4 Sightseeing observations of giant jellyfish (left) and tracer forecasts (right)

long circulations. A notable feature in this region is the vigorous interactions with the marginal seas, particularly with the Sea of Japan through the energetic Tsugaru Warm Current (TWC). In fact, coastal currents off Shimokita Peninsula exhibit a seasonally-varying circulation characterized by a transition of the TWC between two distinct patterns: a straight path along the east coast of Honshu Island in the cold season (hereafter "straight-path mode") and a swirl-like circulation in the warm season (hereafter "gyre-mode"). In et al. (2008) conducted a hindcast experiment for the year 2003 using a triple nesting approach. As a result, the system successfully reproduced the observed transition from the costal mode to the gyre mode and the subsequent opposite one. Figure 17.5 displays the time series of temperature distributions at 200 m depth, in which warm water of over 7°C that corresponds to the TWC water can be seen near Shimokita Peninsula in both the observation and the model. Interestingly, the warm water begins to extend offshore to form a gyre-like distribution.

17.3.5 Non-tidal Coastal Sea Level for Flood Warnings and Port Management in Australia

Accurate and timely warnings of anomalous coastal sea level are critical for the coordination of emergency response and port management. Extremes in coastal sea level frequently occur along the Australian coastline through the combined effects of high tides, storm surge and many other local and non local coastal processes across a range of time and space scales. During the Austral winter-spring period, cold fronts from the Southern Ocean propagate over the Great Australian Bight (GAB) and other southern regions of Australia producing large coastal surges which

Model Observation

142E143E

Fig. 17.5 Time series of temperature distributions at 200 m depth. (left): model and (right) observation. (From In et al. 2008)

142E143E

Fig. 17.5 Time series of temperature distributions at 200 m depth. (left): model and (right) observation. (From In et al. 2008)

can be of O(1) m. The resulting surge can propagate along the southern coastline as a free or forced coastally trapped wave. An example in October 2007 is shown in Fig. 17.6 where the surge formed in GAB has propagated eastward to record a 0.7 m non-tidal sea level in the Gulf of St Vincent. The coastal trapped wave represented in BLUElink in this example has a wavelength of approximately 500 km and a period of approximately two days.

Fig. 17.6 Sea level anomaly in the Great Australian Bight off South Australia at 21:00, 29th October 2007 from the BLUElink 0ceanMAPSv1.0b operational system

Fig. 17.6 Sea level anomaly in the Great Australian Bight off South Australia at 21:00, 29th October 2007 from the BLUElink 0ceanMAPSv1.0b operational system

Fig. 17.7 The root mean square error of 24 h averaged non-tital sea level from BLUElink 0ceanMAPSv1.0b compared Australian coastal tide gauges for a 6 day forecast period

The non-tidal sea level in coastal and shelf regions are frequently governed by a wide range of processes that are modelled in the BLUElink system including storm-surge, coastally trapped waves, boundary currents and eddies. BLUElink does not currently resolve tides, wind-waves, swell or river discharge. Evaluations from both the BLUElink reanalysis (Oke et al. 2008; Schiller et al. 2008) and the operational BLUElink 0ceanMAPSv1.0b (Brassington et al. 2007) indicate the non tidal sea level of these systems has low RMS errors and good correlations with coastal tide gauge (CTG) time series around the Australian coastline. Figure 17.7 shows the RMSE scores for the forecast system over the 6 day forecast period compared with Australian tide gauges. During the period where the atmospheric forecasts have skill 0(48) h the non-tidal sea level forecasts have a RMSE range from 4 to 9 cm. At longer lead times the RMSE grows indicating that the error has not reached saturation. Longer periods of skill can be obtained from the coastal ocean processes.

For example, coastally trapped waves generated in the GAB continue to propagate along the east coast of Australia reaching northern Queensland approximately 5 days after the initial disturbance. Coastal boundary currents and coastal eddies also contribute to anomalous coastal sea level over longer periods, particularly where the continental shelf is narrow. BLUElink has demonstrated useful forecast skill at combining the leading processes influencing non tidal coastal sea level.

17.4 SST Predictability and Forecast Error Growth in China Marginal Seas

17.4.1 Circulation in China Marginal Seas and Model Bias

The seasonal circulation and dynamic processes of the marginal seas around China are mainly controlled by the monsoonal wind force and the impact of Western boundary currents. The monthly mean SST of the two monsoon seasons simulated by the model is illustrated in Fig. 17.8 for the BYS and ECS. The climatologic SST observations (China Ocean Press 1992) are also shown. As one of largest shelf seas of the world, the BYS and ECS has a total area of 1.25 million km2, where most water depths are less than 100 m. Flowing on the shelf break in the ECS, Kuroshio carries warm, saline water from the tropics. In winter this warm water meets with the cold shelf water that is formed due to a huge amount of heat loss to the atmosphere as the northerly monsoonal wind brings in cold and dry air from the continent. A sharp SST front (the Kuroshio Front) therefore forms between the warm Kuroshio and the cold shelf water. The SST distribution in the shelf closely follows the bathymetry. Two mechanisms have been proposed to explain this winter SST distribution pattern. One theory believes that the horizontal advection plays an important role. One evidence is that the northward warm tongue in the Yellow Sea is consistent with the location of the Yellow Sea Warm Current which carries the warm water from west of Cheju Island. For more discussion, see Ichikawa and Beardsley (2002) for a review. An alternative mechanism proposed by Xie et al. (2002) is that water properties are well mixed up to 100 m deep due to intense surface cooling in winter, ocean depth thus has a strong influence on SST of the continental shelf, leading to a remarkable collocation of warm tongues and deep channels. In winter the model simulation can capture the main SST distribution features such the Kuroshio Front and the Yellow Sea Warm Tongue because the driving factors mentioned above is well presented in the model. In summer, the SST distribution is much more homogenous than that in winter. The Kuroshio Front is still there but its intensity is much weaker than in winter. Most SSTs in the BYS are between 24-26°C. SST is generally getting warmer towards the coast. Some local cold waters appear, especially near the southwest and the northwest coast of Korea, the seaward Changjiang runoff, the coastal waters of China and cold water in the north Yellow Sea. In summer,

Fig. 17.8 Observed and simulated mean SST in the winter and the summer over the Yellow Sea and the East China Sea

it is known that tidal fronts are formed in a boundary between well-mixed and stratified regions by the tide-induced mixing and are found along the west coast of Korea. Moon (2005) showed that the tide-induced mixing is to large extent responsible for the SST fronts along the west coast of Korea. Since our model does not contain the tides currently, the simulated SST missed the observed tidal fronts along the west coast of Korea. Generally there are about 2°C of warm bias in the simulation while in the summer the warm bias is about 1°C. And the bias can cause serious problems when using the model to produce a forecast without assimilating observational data.

17.4.2 Hindcast Experiment

The aims of performing a hindcast experiment are two-fold. First we can evaluate the forecast system and identify problems for following trouble-shooting work. Factors influencing the accuracy of an ocean forecast include the errors in the forecasted atmospheric forcing, the ocean initial condition and the model errors. Using the atmospheric reanalysis as atmospheric forcing allows us to minimize the impact of errors in the atmospheric forcing and to examine the performance of the ocean model and its data assimilation scheme. It is a necessary step before the system's applications to operational forecasting. On another hand, the results from such a hindcast experiment can help us to understand more about the ocean predictability. The hindcast experiment started at the beginning of 2006 and ended on the last day of 2006. A 7-day forecast was conducted every 3 days with assimilation of the thinned FSTIA SST data once a day and the along-track SLA data once every 3 days. Figure 17.9 shows a schematic of the setups of the hindcast experiment. The forecasted SST is evaluated against the same FSTIA SST data by calculating the room mean square errors (here we made an assumption that the FSTIA SST data is the truth) over the whole model domain and three interested sub-domains: BYS (Bohai/Yellow Sea), ECS (East China Sea) and SCS (South China Sea). To compare with the data assimilative model forecast, we also made a simple statistical forecast using persistence a trivial predictor.

The SST hindcast errors are compared against posterior SST data, with respect to the skill of persistence. The SST root mean square errors (RMSEs) over BYS, ECS and SCS are displayed in Fig. 17.10 along with that from the persistence. At

1 23456789

Time (day)

Fig. 17.9 The hindcast experiment design and setup. FSTIA daily SST product is assimilated every day at 00:00 time. Along track Jason SLA data is assimilated every 3 days at 00:00 time. The hindcast experiment is run over the whole year of 2006

1 23456789

Time (day)

Fig. 17.9 The hindcast experiment design and setup. FSTIA daily SST product is assimilated every day at 00:00 time. Along track Jason SLA data is assimilated every 3 days at 00:00 time. The hindcast experiment is run over the whole year of 2006

Bohai/Yellow Sea East China Sea South China Sea

Bohai/Yellow Sea East China Sea South China Sea

Fig. 17.10 The spatially averaged root mean square errors (unit: C) of the SST hindcast as a function of the forecast lead time over the four seasons and the whole year of 2006, respectively

the beginning of hindcasting, the RMSEs are zero for the persistence while the SST RMSEs in the model initial conditions are also small. As expected it demonstrates the benefit of data assimilation on the accuracy of the initial conditions comparing to model simulations without data assimilation. However the misfit of the initial SST field to the data shows there is room for further improvement on the data assimilation results. Over the whole model domain, the hindcast errors grow as the lead time increases. During the first two days the hindcast errors increases sharply from 0.3°C to 0.6°C. This indicates there are initial shocks on the model forecast that follow the assimilation by possible creation of gravity waves or other reasons. Then the growth slows down and the RMSE gradually reaches around 0.7°C. After 2-3 days, the hindcast beats the persistence. This shows that the model has an added value to the hindcast. The hindcast skills show seasonal dependence with the best skill in the fall season. The low hindcast skill in BYS during the summer is also obvious. The reasons will be discussed in the next section. Model hindcasts also show different skills in different regions. The order according to the skill from high to low is SCS > BYS > ECS in general. The lowest hindcast skill is in BYS during the summer. This region- and season-dependence is also supported by the skills of persistence. This fact indicates that to some extend the model hindcast skills are influenced by underlying dynamical processes in different regions and seasons and reflect the seasonal and regional short-term SST predictabilities.

The results from two hindcast cases are shown in Figs. 17.11 and 17.12. During the period from June 21 to 27, SST over ECS increases sharply as indicated by the northward movement of the 24°C isoline near the coast. A cooling event and its

SST Observations

SST Observations

June 21 June 23 June 25 June 27, 2006

Fig. 17.11 An example of SST hindcast in June, 2006. The scale of the color bar is °C

June 21 June 23 June 25 June 27, 2006

Fig. 17.11 An example of SST hindcast in June, 2006. The scale of the color bar is °C

SST Observations

SST Observations

Septa SeptlO Sept 12 Sept14.2006

Fig. 17.12 The same as Fig. 17.10, but for another example in September, 2006

Septa SeptlO Sept 12 Sept14.2006

Fig. 17.12 The same as Fig. 17.10, but for another example in September, 2006

hindcast are shown in Fig. 17.11. The 28°C isoline moves southward during September 8 to 14 and the water with SST less than 28°C covers all the shelf area on September 14. The model hindcasts capture the two events successfully while the persistence cannot.

17.4.3 Hindcast Error Distributions

We further explore the predictability of SST in the marginal seas around China. It is beyond the scope of this paper to investigate all aspects of predictability. We only focus on identifying where the SST forecast errors grow fast in summer and winter. Figure 17.13 shows the spatial distribution of the averaged RMSE of a 6-day hind-casts in winter and summer. The large errors in winter mainly locate in the Kuroshio path in the ECS, in the Luzon Strait, off the Vietnam coast along 110E and in the Taiwan Strait. This explains that the hindcast skills are lower in ECS than that in SCS and BYS as shown in Fig. 17.9. In the summer, the large errors mainly locate at coast and shelf regions and northeast of Taiwan. And thus cause the low hindcast skill in BYS during the summer. Because the mixed layer is very shallow in BYS (about 8 m according to Chu et al. 1997) and a strong stratification exist beneath the mixed layer in summer, the SST is very sensitive to the atmospheric forcing and has large day-to-day changes there. The atmospheric forcing errors could be one of the main factors that cause the large hindcast errors in BYS during summer via the

Fig. 17.13 The spatial distributions of the averaged RMSE of 6-day hindcasts. The scale of the color bar is °C. The errors in the water shallower than 40 m are marked out due to relative large errors in the validation data

Fig. 17.13 The spatial distributions of the averaged RMSE of 6-day hindcasts. The scale of the color bar is °C. The errors in the water shallower than 40 m are marked out due to relative large errors in the validation data

100E 110E 120E 13QE

thermal effect. The errors in wind have strong impact on the coast SST hindcast due to the upwelling and downwelling processes. Another reason may be the lack of tides in the model. As mentioned before, the tidal mixing has large impact on summer SST distribution in BYS.

Apart from the errors in the atmospheric forcing and the ocean mixing, the errors in horizontal advections also have large impacts on the hindcast skills. Since the horizontal advection is determined by the inner product of the surface current vector and the SST spatial gradient. Figure 17.14 shows the spatial distribution of absolute

100E 110E 120E 130E

Fig. 17.14 The spatial distributions of the temporally averaged absolute values of the local SST spatial gradients over the winter and the summer of 2006. The SST analysis is used to calculate the gradient. The scale of color bar is 10-4°C/m

100E 110E 120E 130E

Fig. 17.14 The spatial distributions of the temporally averaged absolute values of the local SST spatial gradients over the winter and the summer of 2006. The SST analysis is used to calculate the gradient. The scale of color bar is 10-4°C/m values of the temporally averaged local SST spatial gradients over the winter and the summer of 2006. Instead in winter a strong gradient appears in the shelf between Hainan Island and Taiwan, Kuroshio path associated with the shelf break in ECS, Changjiang River estuary, and the area off the southwest coast of Korea. In summer the gradient is weaker than that in winter and shows its strength mainly around the Korea coast. The front patterns agree well with the previous studies based on multi-year remote sensing SST data (Hickox et al. 2000; Wang et al. 2001). The strong SST gradients agree well with the large hindcast error distributions in Fig. 17.12 especially in the Kuroshio path and in the shelf between Hainan Island and Taiwan. However, off the east coast of Vietnam and the Luzon Strait large hindcast errors exist but the SST gradient is only moderate. Considering the strong current there (e.g., Fig. 17.2 of Li et al. 2010), the horizontal advection errors may also cause a large hindcast error.

17.5 Summary and Outlook

During the GODAE period, several regional operational and preoperational systems have been developed in the Asia-Oceania region. These systems have demonstrated their usefulness via providing routine service for public, government and commerce users or by successful forecasting/hindcasting high impact (socially, economically and scientifically) events.

All these regional systems have strong connections with the GODAE products. The Argo and GHRSST datasets are essential inputs for initialization of these forecast systems. The large scale GODAE products are also used to provide side boundary conditions (e.g., MOVE-NP). A 15 year ocean reanalysis from BLUElink has proved to be useful for engineering design. For example the modelling of internal waves through downscaling assisted the planning of the successful search of HMAS Sydney which was sunk during World War II. The BLUElink operational forecasts have also demonstrated to have good skills at forecasting a wide range of phenomena including: extreme coastal sea level, anomalous currents impacting offshore oil and gas operations, anomalous heat content over the North West Shelf influencing continental rainfall and many other processes. Most of these systems have been developed in GODAE related projects. For example, the 3D ocean forecasting system DMI BSHcmod has been continuously developed in the projects like in the MER-SEA and ECOOP projects. The MERSEA Baltic Sea forecasting system is based on the same model.

It is still a challenging task to further develop the existing systems from science perspective (De Mey et al. 2009). Establishing more observation networks, increasing model resolution, adding sea ice model, using more advanced data assimilation and coupling with atmospheric models are among the near future activities. For example, in the near future, JMA will introduce an assimilation scheme of sea ice concentration to MOVE/MRI.COM-WNP, which would yield to some improvements not only in the sea ice extent but also in the ocean state of the subarctic region, especially in the Okhotsk Sea. JMA is also planning to develop a coastal ocean modeling/assimilation system using a high-resolution model with a horizontal resolution of a few kilometers that is intended for a possible operational use in JMA's forecasting and warning systems for the coastal region of Japan. BLUElink through a follow-on research project will also introduce an upgraded reanalysis and operational prediction system (mid-2010) and introduce a new coupled regional ocean-atmosphere system for tropical cyclone prediction. The global prediction system will enhance the eddy-resolving region to include the Indian Ocean and South Pacific and will also include advances in data assimilation, initialization scheme and atmospheric fluxes.

Another challenge is how to further apply the achievements in Asia-Oceania operational activities comes from high-level decision-makers. The Indian Ocean is relatively less covered by these regional systems. GOOS-CLIVAR's effort in establishing the Indian Ocean Observing system (IndOOS) will improve the situation. Research Moored Array for African-Asian-Australian Monsoon Analysis and Prediction, a new observational Multi-national network designed for Indian Ocean, a subset of IndOOS, similar to TAO (Pacific) and PIRATA (Atlantic), aims to address outstanding scientific questions related to Indian Ocean variability and the monsoons. 22 out of 46 moored buoys are already occupied (McPhaden et al. 2009). On the other hand, Southeast Asian countries also urgently requires such a service for storm-surge forecast, coastal engineering and disaster prevention etc, and some existing systems such as YEOS system are ready to be extended to cover the entire NW Pacific coastal/shelf seas. There are encouraging signs of more Asia-Oceania countries plan to develop operational systems. A study on pre-operational oceano-graphic system will be funded by the Ministry of Land Transport and Ocean Affairs of Korea from next year. They will start to produce data products of coastal and environmental forecasts for the coastal waters around Korea.

From a series of SST hindcast experiments in China marginal sea, we found that several following-up works are necessary. The initial conditions provided by data assimilation seem to have room to further reduce the analysis misfit to data. The causes of the visible initial shocks after the assimilation should be further investigated. Apart from the generation of gravity waves, other reasons should also be thought. Counillon and Bertino (2009) found a data assimilation set-up that produces little noise that is dampened within two days, when the model is pulled strongly towards observations. Part of it is caused by density perturbations in the isopycnal layers, or artificial caballing. Because the model used by them is also HYCOM, their results are very suggestive. The tide-induced mixing has strong impact on the thermal field in BYS. A higher resolution model setup than the present one, with tides is now running and will be used to perform new hindcast experiments.

Acknowledgements Part of this lecture note comes from Zhu et al. (2008) to which co-authors: Toshiyuki Awaji, Gary B. Brassington, Norihisa Usuii, Naoki Hirose, Young Ho Kim, Qinzheng Liu, Jun She, Yasumasa Miyazawa, Tatsuro Watanabe and M. Ravichandran have contributed greatly.

References

Arakawa A (1972) Design of the UCLA general circulation model. Numerical simulation weather and climate, technical report. No. 7. Department of Meteorology, University of California, Los Angeles, p 116

Awaji T, Masuda S, Ishikawa Y, Sugiura N, Toyoda T, Nakamura T (2003) State estimation of the North Pacific Ocean by a four-dimensional variational data assimilation experiment. J Ocean-ogr59:931-943

Bleck R (2002) An oceanic general circulation model framed in hybrid isopycnic-Cartesian coordinates. Ocean Model 4:55-88 Bloom SC, Takacs LL, da Silva AM, Ledvina D (1996) Data assimilation using increment analysis updates. Mon Weather Rev 124:1256-1271 Brassington GB, Pugh T, Spillman C, Schulz E, Beggs H, Schiller A, Oke PR (2007) BLUElink> Development of operational oceanography and servicing in Australia. J Res Pract Inf Technol 39:151-164

Chassignet EP, Smith LT, Halliwell GR, Bleck R (2003) North Atlantic simulations with the Hybrid Coordinate Ocean Model (HYCOM): impact of the vertical coordinate choice, reference pressure, and thermobaricity. J Phys Oceanogr 33:2504-2526 Chen D, Busalacchi AJ, Rothstein LM (1994) The roles of vertical mixing, solar radiation, and wind stress in a model simulation of the sea surface temperature seasonal cycle in the tropical Pacific Ocean. J Geophys Res 99:20345-20359 China Ocean Press (1992) Marine Atlas of Bohai Sea, Yellow Sea, and East China Sea, hydrology.

China Ocean Press, Beijing, p 524 Chu PC, Fralick CR Jr, Haeger SD, Carron MJ (1997) A parametric model for the Yellow Sea thermal variability. J Geophys Res 102(C5):10499-10507 Counillon F, Bertino L (2009) Ensemble optimal interpolation: multivariate properties in the Gulf of Mexico. Tellus-A 61:296-308 De Mey P, Craig P, Kindle J, Ishikawa Y, Proctor R, Thompson K, Zhu J, CSSWG (2009) Applications in coastal modeling and forecasting. Oceanography 22(3):198-205 Dick S, Kleine E, Müller-Navarra S, Klein H, Komo H (2001) The Operational Circulation Model of BSH (BSHcmod). Model description and validation. Bericte des Bundesamtes für Seeschiffahrt und Hydrographie. Nr. 29/2001. Hamburg, Germany, p 48 Fujii Y, Kamachi M (2003) Three-dimensional analysis of temperature and salinity in the equatorial Pacific using a variational method with vertical coupled temperature-salinity empirical orthogonal function modes. J Geophys Res 108(C9):3297. doi:10.1029/2002JC001745 Gent PR, McWilliams JC (1990) Isopycnal mixing in ocean circulation models. J Phys Oceanogr 20:150-155

Griffies S, Hallberg RW (2000) Biharmonic friction with a Smagorinsky-like viscosity for use in large-scale Eddy-permitting ocean models. Mon Weather Rev 128:2935-2946 Griffies SM (1998) The Gent-McWilliams skew flux. J Phys Oceanogr 28:831-841 Griffies SM, Harrison MJ, Pacanowski RC, Rosati A (2004) A technical guide to MOM4 GFDL

Ocean Group technical report No. 5, NOAA/Geophysical Fluid Dynamics Laboratory, p 339 Guo X, Hukuda H, Miyazawa Y, Yamagata T (2003) A triply nested ocean model simulating the

Kuroshio—roles of horizontal resolution on JEBAR-. J Phys Oceanogr 33:146-169 Hasumi H (2000) CCSR Ocean Component Model (COCO) Version 2.1. CCSR Report No.13 Hasumi H (2006) CCSR ocean component model (COCO) version 4.0. Report No. 25. Center for

Climate System Research, The University of Tokyo Hickox R, Belkin I, Cornillon P, Shan Z (2000) Climatology and seasonal variability of ocean fronts in the East China, Yellow and Bohai Seas from satellite SST data. Geophys Res Lett 27(18):2945-2948

Hirose N, Kawamura H, Lee HJ, Yoon JH (2007) Sequential forecasting of the surface and subsurface conditions in the Japan Sea. J Oceanogr 63:467-481

Hunke EC, Ducowicz JK (2002) The elastic-viscous-plastic model for sea ice dynamics. Mon

Weather Rev 130:1848-1865 Ichikawa H, Beardsley RC (2002) The current system in the Yellow and East China Seas. J Ocean-ogr 58:77-92

In T, Ishikawa Y, Shima S, Nakayama T, Kobayashi T, Togawa T, Awaji T (2008) A triple nesting approach toward the improved nowcast/forecast of coastal circulation off Shimokita Peninsula (to be submitted)

Ishikawa Y, Awaji T, Toyoda T, In T, Nishina K, Nakayama T, Shima S, Masuda S (2009) Highresolution synthetic monitoring by a 4-dimensional variational data assimilation system in the northwestern North Pacific. J Mar Syst 78(2):237-248 Ishizaki H, Motoi T (1999) Reevaluation of the Takano-Oonishi scheme for momentum advection on bottom relief in ocean models. J Atmos Ocean Technol 16:1994-2010 Kalnay E et al (1996) The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorologic Soc 77:437-471

Kim YH, Chang KI, Park JJ, Park SK, Lee SH, Kim YG, Jung KT, Kim K (2009) Comparison between a reanalyzed product by the 3-dimensional variational assimilation technique and observations in the Ulleung Basin of the East/Japan Sea. J Mar Syst 78:249-264 Larsen J, H0yer JL, She J (2007) Validation of a hybrid optimal interpolation and Kalman filter scheme for sea surface temperature assimilation. J Mar Syst 65:122-133 Leonard A (1979) A stable and accurate convective modeling procedure based on quadratic upstream interpolation. Comput Methods Appl Mech Eng 19:59-98 Leonard BP, MacVean MK, Lock AP (1993) Positivity-preserving numerical schemes for multidimensional advection. NASA Technical Memorandum 106055, ICOMP-93-05 Levitus S, Burgett R, Boyer TP (1994) World Ocean Atlas, vol 3, salinity. NOAA Atlas NESDIS,

3, United States Department Of Commerce, Washington Li XC, Zhu J, Xiao YG, Wang RW (2010) A model-based observation thinning scheme for assimilation of high resolution SST in the shelf and coastal seas around China. J Atmos Ocean Technol 27:1044-1058

Marchesiello P, McWilliams JC, Shchepetkin A (2001) Open boundary conditions for long-term integration of regional oceanic models. Ocean Model 3:1-20 McPhaden MJ, Mayers G, Ando K, Masumoto Y, Murty VSN, Ravichandran M, Syamsudin F, Vialard J, Yu L, Yu W (2009) RAMA: the research moored array for African-Asian-Australian monsoon analysis and prediction. Bull Am Metrologic Soc. 90: 459-480 Mellor GL, Blumberg A (2004) Wave breaking and ocean surface layer thermal response. J Phys Oceanogr 34:693-698

Mellor G, Kantha L (1989) An ice-ocean coupled model. J Geophys Res 94:10937-10954 Mellor GL, Hakkinen S, Ezer T, Patchen R (2002) A generalization of a sigma coordinate ocean model and an intercomparison of model vertical grids. In: Pinardi N, Woods JD (eds) Ocean forecasting: conceptual basis and applications. Springer, New York, pp 55-72 Miyazawa Y, Kagimoto T, Guo X, Sakuma H (2008a) The Kuroshio large meander formation in 2004 analyzed by an eddy-resolving ocean forecast system. J Geophys Res 113:C10015. doi: 10.1029/2007JC004226

Miyazawa Y, Komatsu K, Setou T (2008b) Nowcast skill of the JCOPE2 ocean forecast system in the Kuroshio-Oyashio mixed water region (in Japanese with English abstract and figure captions). J Mar Meteorol Soc (Umi to Sora) 84(2):85-91 Moon I-J (2005) Impact of a coupled ocean wave-tide-circulation system on coastal modeling.

Ocean Model 8:203-236 Noh Y, Kang YJ, Matsuura T, Iizuka S (2005) Effect of the Prandtl number in the parameterization of vertical mixing in an OGCM of the tropical Pacific. Geophys Res Lett 32:L23609. doi: 10.1029/2005GL024540

Uppala SM, Kallberg PW et al (2005) The ERA-40 re-analysis. Q J R Meteorol Soc 131:2961-3012 Oey LY, Chen P (1992) A Nested-Gris Ocean model—with application to the simulation of Meanders and Eddies in the Norwegian Coastal Current. J Goephys Res 97:20063-20086

Oke PR, Schiller A, Griffin DA, Brassington GB (2005) Ensemble data assimilation for an eddyre-

solving ocean model of the Australian region. Q J R Meteorol Soc 131:3301-3311 Oke PR, Brassington GB, Griffin DA, Schiller A (2008) The Bluelink Ocean Data Assimilation

System (BODAS). Ocean Model 21:46-70 Pacanowski RC, Griffies SM (1999) MOM3.0 manual. WWW Page, http://www.gfdl.noaa.

gov/~smg/MOM/web/guide_parent/guide_parent.html Rio MH, Hernandez F (2004) A mean dynamic topography computed over the world ocean from altimetry, in situ measurements, and a geoid model. J Geophys Res 109:C12032. doi:10.1029/2003JC002226 Schiller A, Oke PR, Brassington GB, Entel M, Fiedler R, Griffin DA, Mansbridge J (2008) Eddy-resolving ocean circulation in the Asian-Australian region inferred from an ocean reanalysis effort. Prog Oceanogr 76:334-365 Toyoda T, Awaji T, Ishikawa Y, Nakamura T (2004) Preconditioning of winter mixed layer in the formation of North Pacific eastern subtropical mode water. Geophys Res Lett 31:L17206. doi: 10.1029/2004GL020677

Tsujino H, Usui N, Nakano H (2006) Dynamics of Kuroshio path variations in a high-resolution general circulation model. J Geophys Res 111:C11001. doi:10.1029/2005JC003118 Usui N, Tsujino H, Fujii Y, Kamachi M (2006) Short-range prediction experiments of the Kuroshio path variabilities south of Japan. Ocean Dyn 56:1616-7341 Wan L, Zhu J, Bertino L, Wang H (2008) Initial ensemble generation and validation for ocean data assimilation using HYCOM in the Pacific. Ocean Dyn 58:81-99. doi:10.1007/s10236-008-0133-x

Wang D, Liu Y, Qi Y, Shi P (2001) Seasonal variability of thermal fronts in the northern South

China Sea from satellite data. Geophys Res Lett 28(20):3963-3966 Weaver A, Courtier P (2001) Correlation modeling on the sphere using a generalizing diffusion equation. Q J R Meteorol Soc 127:1815-1846 Xiao Y, Zhu J (2007) Numerical simulation of circulations in coastal and shelf sea around China using a hybrid coordinate ocean model. Technical report (in Chinese) Xie S, Hafner J, Tanimoto Y, Liu WT, Tokinaga H, Xu H (2002) Bathymetric effect on the winter sea surface temperature and climate of the Yellow and East China Seas. Geophys Res Lett 29(24):2228. doi:10.1029/2002GL015884 Xie JP, Zhu J, Yan L (2008) Assessment and inter-comparison of five high resolution sea surface temperature products in the shelf and coastal seas around China. Cont Shelf Res 28:1286-1293 Zhu J, Awaji T, Brassington GB, Usuii N, Hirose N, Kim YH, Liu Q, She J, Miyazawa Y, Watanabe T, Ravichandran M (2008) Asia and oceania applications. Proceedings of the final GODAE symposium. Available from the GODAE website, pp 359-372

0 0

Post a comment