講演情報
[ACG64-01]Development of an AGCM for High-Resolution Climate Simulation Combining MIROC and NICAM
*松岸 修平1、大野 知紀1、小玉 知央2、宮川 知己1、八代 尚3、建部 洋晶2、道端 拓朗4、渡辺 真吾2、廣田 渚郎1、鈴木 健太郎1、渡部 雅浩1 (1.東京大学 大気海洋研究所、2.国立研究開発法人海洋研究開発機構 、3.国立研究開発法人国立環境研究所、4.九州大学)
キーワード：
大気モデル、モデル開発、物理過程
In recent years, many leading modeling groups have been developing high-resolution climate models that couple atmospheric models with horizontal resolutions of around 25 km to eddy-resolving ocean models with horizontal resolutions of approximately 10 km. Our group has also initiated the development of a high-resolution climate model that will serve as a flagship model over the next 5–10 years, while building upon Japan’s existing climate modeling assets. Specifically, we are developing an atmospheric general circulation model by incorporating the physical parameterizations of the MIROC climate model into the dynamical core of NICAM (Tomita and Satoh, 2004), which employs highly efficient parallelization performances.
An initial evaluation was conducted to assess the implementation of the CMIP6 version of MIROC physical parameterizations (Tatebe et al. 2019) within the NICAM dynamical core. Although its official name has not yet been finalized, we refer to this model as NICAM-M now. In NICAM-M, the physical parameterizations are executed sequentially in the same order as in the original MIROC6, allowing the model state to evolve consistently with the MIROC6 framework. For comparison purposes, NICAM-M was configured at a horizontal resolution equivalent to that of MIROC6 (GL06; 112-km mesh/T85). Using the same parameter settings as MIROC6, NICAM-M produced more abundant low-level clouds, leading to excessive cloudiness and a substantial bias in the TOA radiative budget. To improve the TOA radiative budget, tuning was performed to reduce low-level cloud amounts by strengthening the turbulence and shallow convection parameterizations. Subsequently, a realistic radiation balance was achieved by adjusting the minimum threshold for cloud droplet number concentration.
The tuned NICAM-M reproduces climatological features and biases comparable to those of MIROC6. In addition, NICAM-M can be efficiently executed at higher resolutions. For example, a 1-year integration at a 28-km mesh resolution can be completed in approximately 12 wall-clock hours. However, further performance optimization is required to enable efficient long-term climate simulations.
The model developed in this study provides a seamless modeling framework spanning resolutions from conventional coarse grids to kilometer-scale resolutions. Further investigation is needed to fully exploit the potential of such seamless models.
An initial evaluation was conducted to assess the implementation of the CMIP6 version of MIROC physical parameterizations (Tatebe et al. 2019) within the NICAM dynamical core. Although its official name has not yet been finalized, we refer to this model as NICAM-M now. In NICAM-M, the physical parameterizations are executed sequentially in the same order as in the original MIROC6, allowing the model state to evolve consistently with the MIROC6 framework. For comparison purposes, NICAM-M was configured at a horizontal resolution equivalent to that of MIROC6 (GL06; 112-km mesh/T85). Using the same parameter settings as MIROC6, NICAM-M produced more abundant low-level clouds, leading to excessive cloudiness and a substantial bias in the TOA radiative budget. To improve the TOA radiative budget, tuning was performed to reduce low-level cloud amounts by strengthening the turbulence and shallow convection parameterizations. Subsequently, a realistic radiation balance was achieved by adjusting the minimum threshold for cloud droplet number concentration.
The tuned NICAM-M reproduces climatological features and biases comparable to those of MIROC6. In addition, NICAM-M can be efficiently executed at higher resolutions. For example, a 1-year integration at a 28-km mesh resolution can be completed in approximately 12 wall-clock hours. However, further performance optimization is required to enable efficient long-term climate simulations.
The model developed in this study provides a seamless modeling framework spanning resolutions from conventional coarse grids to kilometer-scale resolutions. Further investigation is needed to fully exploit the potential of such seamless models.