11:45 〜 12:00
[ACG53-11] Downscaling ocean simulation of Japan coastal seas using an ocean reanalysis dataset (FORA-WNP30) in 2003-2012
★招待講演
キーワード：海洋ダウンスケーリング、海洋モデリング、海洋再解析データ、日本近海
We performed a high-resolution (2km) downscaling simulation of Japan coastal seas from an ocean reanalysis dataset (10km resolution) of western North Pacific (FORA-WNP30, Usui et al. 2016), and produced the corresponding downscaling simulation dataset in 2003-2012. We examined the basic performance of the downscaling model and the reproducibility of the main oceanic structures around Japan by comparing the model results with the FORA data.
The ocean model used is MRI.COM (Meteorological Research Institute Community Ocean Model, version 4), basically the same as one that used for creating the FORA-WNP30 data. The model domain is 122.6E-150E and 23.7N-47.5N. The zonal and meridional grid sizes are 1/30 and 1/50 degrees (about 2km), which are 1/3 and 1/5 of those for FORA-WNP30. The method of downscaling is based on the offline nesting tool in MRI.COM: values of the main variables (temperature, salinity, horizontal velocities, surface height, etc) at the lateral boundaries are given by those of the FORA-WNP30 data. The surface forcings (3 hourly) are from the JRA-55 data, which are basically common to those of FORA-WNP30. The initial conditions were made from FORA-WNP30 at 1 January 2003, and the model was integrated for 10 years (2003 to 2012) on the Earth Simulator.
From the comparison of the 10-year simulation results with the corresponding reanalysis data (FORA-WNP30), we confirmed that the averaged features of the main ocean structures around Japan (e.g., sea surface temperature and height, subsurface temperature and salinity, Kuroshio and Oyashio currents, and throuthflows of the main straits in the Japan Sea) are basically well reproduced. We also confirmed that, because of the higher horizontal resolution with the more realistic coastal topography, the model shows higher time variability of velocities and captures abrupt flow change events in coastal regions ("Kyucho" event), which are not clear in the coarser FORA-WNP30 data. On the other hand, some differences between the model and the reanalysis data were found. Warmer subsurface and less saltier surface tendencies in the Japan Sea of the model are the example. Absence of sea ice, river flow, and tidal effects in the present model may have affected some of these. We are planing to improve the model by introducing these processes.
The ocean model used is MRI.COM (Meteorological Research Institute Community Ocean Model, version 4), basically the same as one that used for creating the FORA-WNP30 data. The model domain is 122.6E-150E and 23.7N-47.5N. The zonal and meridional grid sizes are 1/30 and 1/50 degrees (about 2km), which are 1/3 and 1/5 of those for FORA-WNP30. The method of downscaling is based on the offline nesting tool in MRI.COM: values of the main variables (temperature, salinity, horizontal velocities, surface height, etc) at the lateral boundaries are given by those of the FORA-WNP30 data. The surface forcings (3 hourly) are from the JRA-55 data, which are basically common to those of FORA-WNP30. The initial conditions were made from FORA-WNP30 at 1 January 2003, and the model was integrated for 10 years (2003 to 2012) on the Earth Simulator.
From the comparison of the 10-year simulation results with the corresponding reanalysis data (FORA-WNP30), we confirmed that the averaged features of the main ocean structures around Japan (e.g., sea surface temperature and height, subsurface temperature and salinity, Kuroshio and Oyashio currents, and throuthflows of the main straits in the Japan Sea) are basically well reproduced. We also confirmed that, because of the higher horizontal resolution with the more realistic coastal topography, the model shows higher time variability of velocities and captures abrupt flow change events in coastal regions ("Kyucho" event), which are not clear in the coarser FORA-WNP30 data. On the other hand, some differences between the model and the reanalysis data were found. Warmer subsurface and less saltier surface tendencies in the Japan Sea of the model are the example. Absence of sea ice, river flow, and tidal effects in the present model may have affected some of these. We are planing to improve the model by introducing these processes.