14:00 〜 14:15
[AAS08-02] NWP model development for mitigation of heavy rain and typhoon disasters using the supercomputer “Fugaku”
★Invited Papers
キーワード：数値予報
Disasters caused by heavy rains and typhoons occur almost every year in Japan. For the mitigation of such disasters, it needs the better forecast with the enough lead times. And for the better forecast, it needs to develop better NWP system, which provides the basis of the official forecast. The Japan Meteorological Agency (JMA) compiled “JMA’s NWP Strategic Plan Toward 2030” in 2018 and makes effort for the better NWP forecast especially for heavy rains and typhoons. To accelerate such efforts, the JMA conducts the program which utilizes the supercomputer “Fugaku”. In the program, there are several activities such as (1) real time simulation tests for the higher resolution (1km) Local Forecast System (LFM), (2) assessment for the configurations of the future Local Ensemble Prediction System (LEPS), (3) development for the higher resolution Global Spectral Model (GSM), and the other relating activities.
For the better heavy rain predictions, it needs to consider improving the resolution of the NWP model to resolve the clouds which cause the heavy rain and increasing the ensemble size of the system to estimate of the forecast uncertainty for heavy rains.
Since the resolution of the JMA’s official LFM system is 2 km with the forecast range of 10 hours, the higher resolution (1km) LFM was operated on a trial in the summer season in 2022 by using “Fugaku” up to 18 hours forecast. With such pilot operations, it was found that the 1km LFM can predict more realistic heavy rains while it predicted heavy rains too much. Optimizations of the cloud related processes are necessary.
Although JMA is currently operating Meso-scale Ensemble Prediction Systems with the resolution of 5 km, LEPS systems are under development. To assess the future configuration of LEPS is very important tasks now. With considering available computer resource for the operation, it needs to consider how much to increase the model resolution and/or how much to increase the ensemble size. JMA is conducting the feasibility study for it.
To predict the atmospheric phenomena up to several days especially for typhoon behaviors, JMA uses GSM to simulate the atmosphere for the whole globe. The current resolution is 20 km, and it is not enough for simulating the inner core of typhoons. Therefore, it needs to improve the resolution toward under 10 km. But there are several issues to be solved for the future high-resolution GSM in terms of computational accuracy and efficiency.
To accelerate the NWP improvement, the collaboration with the research sectors is very important activity. Relating to this issue, JMA is constructing the experimental system simulating the JMA’s official NWP system on “Fugaku”. The experimental system is composed by the various systems such the observation data decode system, observation data pre-process (quality control) systems, data assimilation system and NWP forecast system. These are almost same as the system operated in JMA every day. Therefore, the experimental system enables external researchers to conduct various study using official NWP like system such as observation system experiments, NWP model scheme upgrade experiments or the other experiments toward improving the official NWP system. It is expected that the collaborative development will be conducted using this experimental system.
Such activities utilizing the supercomputer “Fugaku” are introduced in this presentation.
For the better heavy rain predictions, it needs to consider improving the resolution of the NWP model to resolve the clouds which cause the heavy rain and increasing the ensemble size of the system to estimate of the forecast uncertainty for heavy rains.
Since the resolution of the JMA’s official LFM system is 2 km with the forecast range of 10 hours, the higher resolution (1km) LFM was operated on a trial in the summer season in 2022 by using “Fugaku” up to 18 hours forecast. With such pilot operations, it was found that the 1km LFM can predict more realistic heavy rains while it predicted heavy rains too much. Optimizations of the cloud related processes are necessary.
Although JMA is currently operating Meso-scale Ensemble Prediction Systems with the resolution of 5 km, LEPS systems are under development. To assess the future configuration of LEPS is very important tasks now. With considering available computer resource for the operation, it needs to consider how much to increase the model resolution and/or how much to increase the ensemble size. JMA is conducting the feasibility study for it.
To predict the atmospheric phenomena up to several days especially for typhoon behaviors, JMA uses GSM to simulate the atmosphere for the whole globe. The current resolution is 20 km, and it is not enough for simulating the inner core of typhoons. Therefore, it needs to improve the resolution toward under 10 km. But there are several issues to be solved for the future high-resolution GSM in terms of computational accuracy and efficiency.
To accelerate the NWP improvement, the collaboration with the research sectors is very important activity. Relating to this issue, JMA is constructing the experimental system simulating the JMA’s official NWP system on “Fugaku”. The experimental system is composed by the various systems such the observation data decode system, observation data pre-process (quality control) systems, data assimilation system and NWP forecast system. These are almost same as the system operated in JMA every day. Therefore, the experimental system enables external researchers to conduct various study using official NWP like system such as observation system experiments, NWP model scheme upgrade experiments or the other experiments toward improving the official NWP system. It is expected that the collaborative development will be conducted using this experimental system.
Such activities utilizing the supercomputer “Fugaku” are introduced in this presentation.