|Report of the "Earth System Models: The Next Generation" by Dr. Michio Kawamiya
A workshop took place in Aspen, Colorado, US with the theme "Earth System Models: The Next Generation" for July 30 - August 5,
2006. There were 22 participants from across the world, including G. Meehl (US), P. Cox (UK), M. Kimoto (Japan, Univ. of Tokyo) and
M. Kawamiya (Japan, P6/FRCGC). The workshop has been held by Aspen Global Change Institute since 1989 once a year (or more)
with a different theme every time. Although small in size, the workshop always obtains attendance of many key persons from the
related fields, and its result has been influential. Under the above theme, the participants discussed experimental designs with Earth
System Models (climate models with biogeochemical processes included) for the IPCC 5th Assessment Report, which is expected to
be published around 2013. The venue was a tent set up in the yard of the institute, which created friendly atmosphere throughout the
workshop. The main topics included incorporation of carbon cycle, enhancement of model resolution and treatment of
aerosols/chemistry. The Japanese participants made a significant contribution to the discussion based on their rich experiences on
these topics (although it is regrettable that Japanese scientists, including me, often lack the vigor to "dominate" this sort of
discussion). The summary of the discussion will shortly be published on "EOS" issued by American Geophysical Union.
The current status of the next generation model development
group is reported. |
The first version of a global cloud resolving model has been developed
at the atmosphere modeling branch. The world's first global cloud
resolving simulations were performed for an aqua planet. Results of
these simulations have being analyzed especially about large-scale
organized cloud systems in the tropics. The model was also applied
for a climate sensitivity test. Recently, simulations with realistic
condition are being performed using the model to investigate cloud-radiation
interaction and to evaluate accuracy of the model. The ocean modeling
branch aims at understanding how mesoscale eddies affect climate systems.
The Southern Ocean and the Labrador Sea are the targets of regional
model simulations. Effects of mesoscale eddies in the Southern Ocean
were quantitatively evaluated in the simulations. In addition to the
approach using regional models, a new global ocean model has being
developed with cubed sphere grids to achieve high computational performance.
It is expected that quantitative discussions about mesoscale eddies
in the global ocean will become possible by using the model.
|Introduction of the research of "Kyousei Project 7".
||A realistic simulation of the global climate system using a coupled
atmosphere-ocean-land surface model is generally moredifficult than that using a single component model due to
various kinds of air-sea feedback processes. To improve a conventionalcoupled model, we developed a new low-level
stratocumulus scheme, implemented a canopy-type land surface model, and tunedthe model parameters relating to
ocean mixed layer processes. We also have applied a variational dataassimilation scheme to optimize the coupled
model climatology by controlling the bulk coefficients at thesea surface. The adjusted bulk coefficients
considerably improve the results of a free-run of the coupledmodel not only in the climatological state but
also in the ENSO behavior, especially its spectral features.These results were presented at the WMO International
Symposium (Apr 2005, Prague) and at the JointSymposium on data assimilation (May 2005, University of Oklahoma).
We are planning to apply this dataassimilation system to the reanalysis experiments which reproduce various
coupled phenomena, to thesensitivity analyses of a climate dynamics, and to the initialization of
seasonal/interannual forecasts. Wehope that our achievements will be of help to elucidate various climate
An international project concerning the Earth System Modeling, the
task force meeting of IGBP-GAIM was held from 25, October, 2004 to
27 at JAMSTEC Yokohama Institute, and FRCGC served as a local host.
Almost 30 people participated in the meeting including Dr. Ayako Abe
and Dr. Michio Kawamiya from FRCGC, Dr. Colin Prentice, the co-chair
of Global Analysis, Integration and Modeling (GAIM), Dr. Dork Sahagian,
the director of GAIM, Dr. Guy Brasseur, a chairman of International
Geosphere-Biosphere Programme (IGBP) and Dr. Kevin Noone, an Executive
director of IGBP, as well as many executives involving in IGBP programs.
The meeting had proceeded taking enough time for the Q and A and discussion
sessions after having presentations by participants. In the sessions,
some were about the iron and the nitrogen cycles, which can be considered
as relatively "traditional" issues for IGBP. Yet there were
other kinds of reports trying to find subjects for which social science
and global environmental science can collaborate. Such presentations
include, one titled "Economic Science of Shifting for Sustainability",
and the report of the project, Integrated History and Future of the
Planet Earth (IHOPE), which is trying to understand the interaction
between the human history and the global environment during this past
We would like to introduce recent activities
of the Next-Generation Model Development Group.
The Atmospheric Modeling subgroup develops a nonhydrostatic
and regular icosahedral grid model called NICAM. NICAM, which
was installed cloud physical processes and radiation processes
already, is scheduled for aqua-planet experiments at a resolution
of about 3.5km horizontal grid. The group's goal of this fiscal
year is to install an aerosol model to perform experiments
such as AMIP-like ones and climate sensibility ones. This
group also works on examining non-negative advection, and
on examining and improving physical processes (convection
parameterization, cloud physical processes, and mixed-layer
schemes) on a regular icosahedral grid. The Ocean Modeling
subgroup studies the roles of mesoscale eddies, which fill
the ocean and rang from several km to several 10 km, in climate
system. To identify the mesoscale eddies, it is necessary
to cover the whole globe with a fine grid. Therefore, this
group works on developing a new mathematically efficient ocean
model by using a cubic grid. Now, this group conducts test
runs at relatively low-resolution for preparation of future
high-resolution climate calculation. Moreover, in cooperation
with Center for Climate System Research, University of Tokyo,
this group develops local models of the Southern Oceans and
the Labrador Sea, where the mesoscale eddies play important
roles, to study mesoscale eddies' roles in these areas.
Introduction of the research of "Kyousei Project 7".
We have been adding many modifications to the Coupled
Atmosphere-Ocean-Sea Ice model for the Earth Simulator (CFES) to improve
its performance. The simulation of the climate in the CFES composed
by the Earth Simulator Center improved a lot by our intensive tunings
of the parameterizations of the ocean and atmosphere parts, which
were necessary for exploiting its full potential. Furthermore, we
introduce newly developed land parameterization scheme MATSIRO (Minimal
Advanced Treatments of Surface Interaction and Runoff) to the CFES
instead of the current one layer model so called "bucket model".
On the top of this coupled model, we are constructing the variational
data assimilation system. We have started the design of the experiment
to produce climatological reanalysis data. We are working hard for
our final goal to provide the integrated reanalysis datasets of 1990s.
Annual meeting of the GAIM taskforce was held with more than 20 participants
from various nations on October 26-29, 2003 in Cambridge, UK. GAIM
stands for Global Analysis, Interpretation and Modelling and is a
core project of the International Geosphere Biosphere Program (IGBP).
From FRCGC, Ayako Abe-Ouchi of Global Warming Research Program, and
Michio Kawamiya of Global Environment Modeling Research Program attended the
meeting. The discussion put emphases on the role that IGBP/GAIM could
play in the partnership among current international research projects
on global change including IGBP, and modeling human activities for
incorporation into earth system models. Some may consider that the
second topic above, in particular, is a dubious one, but it will be
necessary in the future to deal with this sort of issues with a well-balanced
manner, since the 4th assessment report of IPCC is also expected to
stress the importance of cross-cutting researches among its three
working groups. It was proposed that the next meeting would be held
in Japan, which could provide an excellent opportunity to present
contributions from Japan to international projects on global change.
Global Environment Modeling Research Program
We have started a research collaboration with a team from the Hadley Centre in the U.K. (FUJI, First UK-Japan Initiative of climate model intercomparison) as a part of Subject 1 of the Ministry of Education, Culture, Sports, Science and Technology (MEXT)'s project for Sustainable Co-existence of Humans, Nature and the Earth. The team visited us three times; in September, February and May-June. Visiting team members included Dr. Dave Griggs (the Director), Dr. Richard Wood and Dr. Malcolm Roberts from the Hadley Centre, and Prof. Julia Slingo, Dr. Lois Steenman-Clark, Dr. Jeff Cole and Dr. Mat Collins from the University of Reading. Some of us involved in this project also visited the U.K. last February. Through this collaboration we hope to facilitate information exchange on the development of high-resolution climate models, detailed comparison of our model results, and collaboration utilizing the Earth Simulator.
The next generation atmospheric modeling group is comparing two dynamical cores; our icosahedral gridpoint model and the spectral model (AFES in Earth Simulator Center). It has been generally pointed out that gridpoint models have computational advantage over spectral models in high resolutions. Inspecting the computational performances of two models, we have actually shown that the icosahedral model can calculate faster than the spectral model for simulations in less than 10km grid interval. The next generation ocean modeling group has recently developed Arakawa Jacobian on the cubic grid, which could be applied all over the domain including the cubic grid singular points. This enabled us to make Arakawa-type momentum advection scheme on the cubic grid. It is expected that a high performance ocean simulation can be realized by using this method.
The next generation atmospheric modeling group is investigating
the performance of the icosahedral nonhydrostatic dynamical core.
For the development of cloud resolving Atmosphric General Circulation
Model (AGCM), the examination of dependency of model results on
the horizontal resolution and comparison between several turbulence
models are also being performed by using regional models including
physics. The next generation ocean modeling group is trying to parallelize
their ocean general circulation model code. The computational speed
of the model reaches 14 T Flops by using 3840 CPUs on the Earth
simulator (ES). This result is presented at the international workshop
"Use of High Performance Computing in Meteorology" held in U. K.
The data assimilation group starts a data assimilation experiment
on the ES. An experiment for climatologically seasonal variability
has been executed preliminarily (also presented at the aforementioned
workshop). That for interannual variability is the next target.
Preparation for the development of an earth system model is becoming
completed with two new full-time researchers who joined our Program;
Singo Watanabe from Kyushu University is an expert on middle atmosphere
modeling, and Michio Kawamiya from Kiel University, Germany, has
been conducting researches using ocean ecosystem models.
As the Earth Simulator has started its operation, each group in
our program has started model development for the large-scale simulation
on the Earth Simulator. In the next generation atmospheric modeling
group, test performance of the dynamics core of the non-hydrostatic
icosahedral atmospheric model on the Earth Simulator was started.
So far, we have got the sustained performance 43.5% of the peak
performance in case of horizontal grid of 30 km and 100 layers using
80 nodes. The elapsed time of one simulation day is 195 sec. The
next generation ocean modeling group is developing a general circulation
ocean model using a cubic grid system to realize the high speed
calculation on the Earth Simulator. The module of the shallow water
equations has been developed and some results are obtained.
In addition, the task team for global warming experiment using high
resolution coupled model was officially established, and started
preparation of modeling for global warming experiment on the Earth
Simulator, in co-operation with Center for Climate System Research,
University of Tokyo, and National Institute for Environmental Studies.
Meanwhile, the data assimilation group is taking two tasks on the
Earth Simulator. One is construction of a high-resolution ocean
data assimilation system using a 4-dimensional variational adjoint
method. The other is execution of a data assimilation experiment
for interannual variability in the ocean with a mid-resolution model.
These are unable to carry out on the conventional super computers.
For the prediction and research of the mid- to long-term climate
change, it is inevitable to improve the performance of the ocean
model. Especially, it is necessary to develop a high-resolution
ocean model in order to resolve meso-scale eddies, which are well
known to play significant role in the transport of heat and other
substances in the ocean.
The ocean model group in our program is developing a high-performance
parallel general ocean circulation code, UMI-1. Its computational
speed has reached 3.5 TFLOPS on the Earth Simulator using 1024 CPUs
(c.f. abstract of Parallel Computational Fluid Dynamic Conference
2002). In addition, we must also improve physical parameterizations
to cope with these high-resolution models.
One of the most essential issues about it is on the parameterization
of the bottom topographic effect, which directly controls the intensity
of the large-scale meridional circulation. Our group has improved
the "Bottom Boundary Layer Scheme" in the ocean model to reproduce
the realistic entrainment on the continental sloop and obtained
several new findings.
Furthermore, for the improvement of physical parameterization, a
data assimilation method is also effective. We are preparing to
estimate physical coefficients in the some parameterizations using
In the High Resolution Ocean Model Research Group, preparation of
the data necessary for the numerical experiment by super high resolution
MOM3 (0.1 degree horizontal, and 50 layers vertical), which will
be the world's first super gigantic numerical calculation, has been
almost completed in co-operation with Climate Variations Research
Program. The group also has started its test run, to be used also
for off-peak test of the Earth Simulator. Test run of the long integration
by high resolution OGCM for the global warming experiment has also
In the High Resolution Atmospheric General Circulation Model, the
cross-cutting project for the global warming prediction has been
set up into action. The Next Generation Ocean Model Group is trying
the high-speed parallelization of the ocean model, under development.
The group also has tried creating framework for uniform high resolution
model of the sphere to overcome the problems of the existing mesh
in a polar region. A test has been conducted for the high accuracy
advection scheme. The Data Assimilation group has developed a system
for ocean data assimilation using the four dimensional variational
method. Currently, the adjoint code is being applied to a sensitivity
analysis to investigate the structure of the oceans.
In the Next-Generation Model Research Group, develop-ment of a shallow-water
model using an icosahedral grid and a conformal cubic grid, and
a non-hydrostatic model meeting the conservation of mass and energy
has been almost completed. Following on from these two models, we
have shifted our development scheme to a global three-dimensional
non-hydrostatic model. Based on the program's non-hydrostatic model,we
are studying the physical process to be utilized for the next-generation
Drs. Sato and Tomita participated in the 2001 Workshop on Partial
Differential Equations over the Spherical Surface (PDE 2001) held
in Montreal, Canada in July, and presented their research results
on the numerical calculation method of the next generation model.
In September, Drs. Sato, Tomita, and Nasuno participated in the
Fourth International Workshop on Short Range Numerical Weather Prediction
(SRNWP 2001) held in Bad Orp, Germany and gave their presentation
on the current hydrostatic model.
In the Coupled-Model Development Group, sea ice has been integrated
into the group's ocean model. The group is also progressing with
the coupling of the atmosphere model and ocean model, in cooperation
with the Earth Simulator Research Center.
In the Data Assimilation Research Group, data assimilation using
ARGO data has been carried out in co-operation with Frontier Observation
System for Global Change (FORSGC). While most ocean observation
data are taken from the sea surface, this ARGO data is able to obtain
the sea temperature and salinity throughout the upper 2000m. It
is thus expected to help elucidate the detailed structure of the
global ocean circulation.
From July 23 to 25, the Global Oceanography Data Assimilation Experiment
(GODAE) workshop was held at the University of Hawaii. Group Leader
Dr. Awaji and researcher Dr. Sugiura attended the workshop and reported
the results of data assimilation research.
These results were the analysis of seasonal variation of ocean transport
processes. Four-dimensional variational method assimilation was
used to obtain initial conditions and surface fluxes for an ocean
general circulation model, which produces a dynamically consistent
seasonally varying climate data set. In model development, the Atmosphere
General Circulation Model Group is running a project for a high-resolution
model that is designed to have a 50-km horizontal resolution to
be the target of coupled atmosphere-ocean model, which is to be
run on the Earth Simulator.
The Ocean General Circulation Model Group has completed improvements
of the time integration method, which has resulted in dramatical
improvement of model efficiency. The Dynamical Core Group is proceeding
with construction of dynamic components for a global three-dimensional