Press Releases

December 11, 2007
The Japan Agency for Marine-Earth Science and Technology
The University of Tokyo
The Japan Science and Technology Agency

Realistic Simulation of Madden-Julian Oscillation (MJO)
- Suggests the potential of high-accuracy seasonal prediction -


Hiroaki Miura, Postdoctoral fellow at Global Environment Modeling Research Program, Frontier Research Center for Global Change (FRCGC), the Japan Agency for Marine-Earth Science and Technology (JAMSTEC: Yasuhiro Kato, President) and Masaki Satoh, Associate professor at Center for Climate System Research (Prof. Teruyuki Nakajima, Director), the University of Tokyo, and their research team have succeeded detailed reproduction of Madden-Julian Oscillation (MJO) Event which is a prominent cloud activity in tropical area. Forming and vanishing of cloud were computed directly by a global atmosphere model (*1) on the Earth Simulator. The feasibility of prediction for month-long behavior of MJO associated cloud cluster from its generation was actually proved for the first time in the world. This succession is expected to improve weekly and seasonal prediction and shall affect development of world-scale atmosphere model. The experiment titled “Global Cloud Resolving Model Simulations toward Numerical Weather Forecasting in the Tropics” has been conducted as a simulation assignment for “High Performance Computing for Multi-Scale and Multi-Physics Phenomena” which is one of research areas of Core Research for Evolutional Science and Technology (CREST) (*2) of the Japan Science and Technology Agency (JST).

This achievement will be in the science magazine “Science” published by American Association for the Advancement of Science (AAAS) on December 14, 2007.


MJO had been discovered as pressure oscillation (of 40~50 days cycle), and that is actually an event that a cumulus ensemble develops to several thousand kilometers scale and propagates slowly eastward at the average speed of approximately 5km/s from the Indian Ocean to the Pacific Ocean (Fig.1). In tropical area, a cloud cluster accompanied with local torrential rain often causes disasters. Cloud cluster becomes a large-scale low pressure crossing the equator, with unique atmospheric circulation such as strong west wind blowing along the equator. It appears that the event affects greatly on world-wide climate system (*3) including tropical cyclone, monsoon activity and El Niño.

Therefore, accurate MJO prediction is expected in order to improve weekly or seasonal prediction in Tropical area as well as all over the world. However, existing atmosphere models including the atmosphere-ocean coupled model are not able to simulate MJO completely.

3.Outline of simulation experiment

FRCGC developed a high-resolution global atmosphere model which is able to directly compute interplay between atmospheric circulation and forming or vanishing of cloud. Simulation experiment was done for a large-scale cloud activity that accompanies MJO, developed in Indian Ocean in December 2006 and propagated to Pacific Ocean by January 2007.

Aiming to reproduce detail cloud distributions, 7 days computing experiment started from 0 time on December 25, 2006 (Coordinated Universal Time (UTC)) with 3.5km-horizontal mesh (*3). And to reproduce large-scale cloud cluster propagate from Indian Ocean to Pacific Ocean, 30 days experiment started from 0 time on December 15, 2006 (UTC) with 7 km-horizontal meshes.

4.Result of simulation experiment

Experiment with 3.5km-horizontal mesh realistically simulated large-scale cloud cluster distribution accompanied with MJO (Fig.1), it showed that Southeast Asian Islands were covered with many clusters of several hundreds kilometers horizontal scale. Experiments with 7km-horizontal mesh precisely simulated large-scale cloud cluster propagating from Indian Ocean to Pacific Ocean temporally and spatially (Fig.2). Center of convection above Indian Ocean on December 23, 2006 (left of Fig.2) traveled to Southeast Asian Islands on December 31, 2006 (middle of Fig.2) and Pacific Ocean on January 8, 2007 (right of Fig.2). Tropical cyclone in southward of Java, generated along with MJO, has been successfully simulated also on January 2 in 2007, 2 weeks after commence computing (Fig.3).

5.Future prospective

Predictability of month-long MJO was demonstrated by high-resolution atmosphere model which is capable to simulate forming and vanishing of cloud. Improvement of computational ability in the future would result in higher resolution of climate-forecasting model which can provide more accurate, longer period MJO prediction, so that climate forecast in tropical area is to be possible, weekly or seasonal forecast shall be improved, and it is expected that rainfall variability in the tropical area is going to be predictable which is a key for global warming projection. Therefore, developing global atmosphere model to compute cloud directly has become one of “Innovative Program of Climate Change Projection for the 21st Century” from this fiscal year (FY2007). Benefits of developing high-resolution atmosphere model were clearly demonstrated in this achievement which possibly will make great influence to the world trend of developing atmosphere model. It is expected that as atmosphere model become more elaborate and higher-resolution, computation will be more realistic, and mechanism of tropical cyclogenesis accompany MJO will be elucidated.

*1: It is called Global Cloud-resolving Model. Uncertainty had been caused because existing global atmosphere model required some assumption about interplay between large-scale atmospheric circulations such as high/low pressure and cloud. By responding directly from cloud to atmosphere circulation, more accurate computing has become possible.

*2: Team type study “CREST” is to promote study to create innovation with great impact by organizing research team and aiming strategic objectives indicated by the government (the Ministry of Education, Culture, Sports, Science and Technology) and ensure to meet the nations’ social and economic needs.

*3: A study showed that winter temperature rather warm in Japan when convective center of MJO is above Indian Ocean, rather cold when it is above Southeast Asian Islands.

*4: IPCC (Intergovernmental Panel on Climate Change) Forth Assessment Report (AR4), section 8 “Climate Models and their Evaluation” described “Simulation of the Madden-Julian Oscillation (MJO) remains unsatisfactory.”

*5: 40km-thick atmosphere around the earth is divided in grid with 16 hundred million points for computation with 3.5km-horizontal mesh and with 4 hundred million points for computation with 7km-horizontal mesh.

Fig.1: Left: Cloud image at 00:00 UTC on December 31 in 2006 reproduced by data acquired from Climate Satellite (MTSAT-1R). Right: Cloud image computed with 3.5km-horizontal mesh at the same time. (Hereinafter, each time is described with UTC and each model shows the result of 6days (144 hours) prediction.)

Fig.2: Time-variable images of cloud computed with 7km-horizontal mesh of initial condition as of 00:00 on December 15, 2006. Left: 00:00 on December 23, 2006 (8days later). Middle: 00:00 on December 31, 2006. Right: 00:00 time on January 8, 2007.

Fig.3: At 00:00 on January 2, 2007. Left: Observed tropical cyclone (precipitable water [mm/hr]) Right: Status of 18days after starting the tropical cyclone simulation with 7km-horizontal mesh (integrated cloud water [kg/m2]) Tropical cyclone is identified at longitude 120°east and latitude 15°north.


(For the study)
Hiroaki Miura, Postdoctoral fellow
Global Environment Modeling Research Program,
Katsuhiko Masuda, Manager, e-mail:
Research Promotion Office
Frontier Research Center for Global Change (FRCGC),
The Japan Agency for Marine-Earth Science and Technology
Masaki Satoh, Associate professor
Center for Climate System Research, the University of Tokyo
(For publication)
Shinji Oshima, e-mail:
Manager, Planning Department Press Office
The Japan Agency for Marine-Earth Science and Technology
Yukio Nakata
Kashiwa Campus Office
The University of Tokyo
Department of Public Relations and Science Portal
The Japan Science and Technology Agency