A New Global Climate Model of the Meteorological Research Institute: MRI-CGCM3 ^|^mdash;Model Description and Basic Performance^|^mdash;

Abstract

A new global climate model, MRI-CGCM3, has been developed at the Meteorological Research Institute (MRI). This model is an overall upgrade of MRI’s former climate model MRI-CGCM2 series. MRI-CGCM3 is composed of atmosphere-land, aerosol, and ocean-ice models, and is a subset of the MRI’s earth system model MRI-ESM1. Atmospheric component MRI-AGCM3 is interactively coupled with aerosol model to represent direct and indirect effects of aerosols with a new cloud microphysics scheme. Basic experiments for pre-industrial control, historical and climate sensitivity are performed with MRI-CGCM3. In the pre-industrial control experiment, the model exhibits very stable behavior without climatic drifts, at least in the radiation budget, the temperature near the surface and the major indices of ocean circulations. The sea surface temperature (SST) drift is sufficiently small, while there is a 1 W m-2 heating imbalance at the surface. The model’s climate sensitivity is estimated to be 2.11 K with Gregory’s method. The transient climate response (TCR) to 1 % yr-1 increase of carbon dioxide (CO2) concentration is 1.6 K with doubling of CO2 concentration and 4.1 K with quadrupling of CO2 concentration. The simulated present-day mean climate in the historical experiment is evaluated by comparison with observations, including reanalysis. The model reproduces the overall mean climate, including seasonal variation in various aspects in the atmosphere and the oceans. Variability in the simulated climate is also evaluated and is found to be realistic, including El Niño and Southern Oscillation and the Arctic and Antarctic oscillations. However, some important issues are identified. The simulated SST indicates generally cold bias in the Northern Hemisphere (NH) and warm bias in the Southern Hemisphere (SH), and the simulated sea ice expands excessively in the North Atlantic in winter. A double ITCZ also appears in the tropical Pacific, particularly in the austral summer.

Full text

Journal of the Meteorological Society of Japan, Vol. 90A, pp. 23--64, 2012. 23
DOI:10.2151/jmsj.2012-A02
A New Global Climate Model of the Meteorological Research Institute: MRI-CGCM3
—Model Description and Basic Performance—
Seiji YUKIMOTO, Yukimasa ADACHI, Masahiro HOSAKA, Tomonori SAKAMI,
Hiromasa YOSHIMURA, Mikitoshi HIRABARA, Taichu Y. TANAKA, Eiki SHINDO,
Hiroyuki TSUJINO, Makoto DEUSHI, Ryo MIZUTA, Shoukichi YABU, Atsushi OBATA,
Hideyuki NAKANO, Tsuyoshi KOSHIRO, Tomoaki OSE, and Akio KITOH
Meteorological Research Institute, Tsukuba, Japan
(Manuscript received 16 March 2011, in final form 4 August 2011)
Abstract
A new global climate model, MRI-CGCM3, has been developed at the Meteorological Research Institute
(MRI). This model is an overall upgrade of MRI’s former climate model MRI-CGCM2 series. MRI-CGCM3 is
composed of atmosphere-land, aerosol, and ocean-ice models, and is a subset of the MRI’s earth system model
MRI-ESM1. Atmospheric component MRI-AGCM3 is interactively coupled with aerosol model to represent di-
rect and indirect e¤ects of aerosols with a new cloud microphysics scheme. Basic experiments for pre-industrial
control, historical and climate sensitivity are performed with MRI-CGCM3. In the pre-industrial control experi-
ment, the model exhibits very stable behavior without climatic drifts, at least in the radiation budget, the temper-
ature near the surface and the major indices of ocean circulations. The sea surface temperature (SST) drift is suf-
ficiently small, while there is a 1 W m2heating imbalance at the surface. The model’s climate sensitivity is
estimated to be 2.11 K with Gregory’s method. The transient climate response (TCR) to 1 % yr1increase of car-
bon dioxide (CO2) concentration is 1.6 K with doubling of CO2concentration and 4.1 K with quadrupling of
CO2concentration. The simulated present-day mean climate in the historical experiment is evaluated by compar-
ison with observations, including reanalysis. The model reproduces the overall mean climate, including seasonal
variation in various aspects in the atmosphere and the oceans. Variability in the simulated climate is also eval-
uated and is found to be realistic, including El Nin˜o and Southern Oscillation and the Arctic and Antarctic oscil-
lations. However, some important issues are identified. The simulated SST indicates generally cold bias in the
Northern Hemisphere (NH) and warm bias in the Southern Hemisphere (SH ), and the simulated sea ice expands
excessively in the North Atlantic in winter. A double ITCZ also appears in the tropical Pacific, particularly in the
austral summer.
1. Introduction
Climate models have been advanced to simulate
many aspects of the observed climate. In the
Fourth Assessment Report of the Intergovernmen-
tal Panel on Climate Change (IPCC) (hereafter,
IPCC-AR4), the projections of future climate
change were based on numerous experiments with
more than 20 state-of-the-art climate models, and
yielded results with quantitative confidence levels.
The range of uncertainties in the projections,
however, remained as large as in the 3rd Assess-
ment Report (IPCC 2001). Bony and Dufresne
(2005) suggested that a major source of uncertainty
in climate sensitivity is feedback in tropical low
clouds. The uncertainty related to radiative forcing
was also a major factor. Many questions remain re-
garding the modeling of the indirect e¤ects of aero-
sols (e.g., Forster et al. 2007), which must take into
account sophisticated cloud microphysics (involv-
ing high computational costs). No models thus far
Corresponding author: Seiji Yukimoto, Climate Re-
search Department, Meteorological Research Institute,
1-1 Nagamine, Tsukuba, 305-0052, Japan.
E-mail: yukimoto@mri-jma.go.jp
62012, Meteorological Society of Japan

have involved a fully interactive atmospheric chem-
istry of aerosols with cloud microphysics for cli-
mate change experiments, although major uncer-
tainties are associated with the aerosol e¤ects.
The Meteorological Research Institute (MRI),
with the former climate model MRI-CGCM2.3.2
(Yukimoto et al. 2006), contributed to the IPCC-
AR4 by providing results from numerous experi-
ments for the third phase of the Coupled Model
Intercomparison Project (CMIP3). For the experi-
ments, MRI-CGCM2.3.2 used global flux adjust-
ments (Manabe and Stou¤er 1988) for heat and
freshwater, and partially for momentum. The main
reason for using the flux adjustments is that the
model output is used as boundary conditions of
the regional climate model (RCM) for downscaling
the future climate change in the vicinity of Japan
(Kurihara et al. 2005), which requires high preci-
sion and reproducibility of climate. Except for the
RCM’s boundary condition, the model exhibited
su‰cient performance without flux adjustments in
simulating the mean climate and variability on
global and sub-continental scales (Kitoh 2004).
We have developed a new climate model, MRI-
CGCM3. This model is an overall upgrade of the
MRI-CGCM2 series. This climate model is a core
subset of MRI’s earth system model MRI-ESM1
(Yukimoto et al. 2011). MRI-CGCM3 consists of
the atmosphere-land model (MRI-AGCM3), the
ocean and sea ice model (MRI.COM3), and the
aerosol model (MASINGAR mk-2). These compo-
nent models are coupled with a simple and flexible
coupler ‘‘Scup’’, which enables us to make a variety
of combinations of the component models with ar-
bitrary resolutions and grid coordinates.
Phase five of the CMIP (CMIP5) experiments is
planned (Taylor et al. 2011; available at http://
cmip-pcmdi.llnl.gov/cmip5/experiment_design
.html). These experiments are divided into two
major categories, long-term experiments and near-
term (decadal prediction) experiments. The long-
term experiments are further divided into an exper-
iment group driven by concentrations of the GHGs
and other forcing agents (C-driven), and an experi-
ment group driven by emissions (E-driven). MRI
will perform all these experiments with this unified
model. The E-driven experiments will be performed
with the fully configured MRI-ESM1, including
the atmospheric chemistry climate model (MRI-
CCM2), and submodels of the carbon cycle pro-
cess, representing terrestrial and marine ecology.
The other experiments will be performed with
MRI-CGCM3, which has the same configuration
as the MRI-ESM1 but without MRI-CCM2 and
the carbon cycle. We will perform all the experi-
ments for CMIP5 without any flux adjustments.
Atmospheric aerosols influence the climate by
perturbing the Earth’s radiation budget in several
ways. A direct radiative e¤ect is caused by the di-
rect scattering and absorption of atmospheric radi-
ation by aerosols. An indirect radiative e¤ect is
aerosols acting as cloud condensation nuclei and af-
fects cloud albedo (Twomey 1974; Twomey 1991),
precipitation formation, and cloud lifetime charac-
teristics (Albrecht 1989). Absorptive aerosols, such
as black carbon (BC) or mineral dust, warm the at-
mosphere and reduce solar radiation at the surface,
thus increasing atmospheric stability. Absorptive
aerosols can locally inhibit cloud formation or re-
duce cloud cover by heating cloud droplets. This is
called the semi-direct aerosol e¤ect (Hansen et al.
1997). Moreover, absorptive aerosols deposited on
a snow surface reduce the albedo of the snow sur-
face and enhance the melting of the snow (Hansen
and Nazarenko 2004). The aerosol model is interac-
tively incorporated into MRI-CGCM3, which ex-
plicitly presents the interaction between such e¤ects
of aerosols and climate perturbation.
We conducted a pre-industrial control experi-
ment (piControl in the CMIP5 syntax) and a his-
torical experiment. With the piControl experiment,
we evaluate the stability of the model’s climate.
The experiment is also used as a reference for vari-
ous experiments. By comparing the instrument ob-
servations with the historical experiment of 1850
through 2005, we can examine the capability of the
model to reproduce the historical climate change
as well as the present-day climate. In particular, it
is important to assess the reproducibility of recent
decades since 1979 from a comprehensive perspec-
tive, since observations are enhanced with advances
in various satellites for this period. Based on the
results of some of the climate sensitivity experi-
ments, we also examine the climate sensitivity of
the model.
This paper is structured as follows. Section 2
describes the model’s configuration. Section 3 de-
scribes the experiment design and the spin-up of
the model. Section 4 demonstrates the stability and
climate drift of the model in the piControl experi-
ment. Section 5 examines the climate sensitivity of
the model and the reproducibility of historical cli-
mate change, and Section 6 evaluates how the cli-
mate in the recent past is reproduced, including the
24 Journal of the Meteorological Society of Japan Vol. 90A

mean field and the variability. Section 7 presents a
summary and discussion.
2. Model description
The model we describe in this paper is MRI-
CGCM3, which is a subset of MRI-ESM1 (Yuki-
moto et al. 2011). It is a coupled atmosphere-ocean
global climate model that is composed of MRI-
AGCM3 and MRI.COM3, whose atmospheric
part (MRI-AGCM3) is interactively coupled with
aerosol model MASINGAR mk-2. See Yukimoto
et al. (2011) for a detailed description of the model.
2.1 Atmospheric model (MRI-AGCM3)
Newly developed atmospheric model MRI-
AGCM3 is based on JMA’s operational forecast
model. Its dynamics frame is a global spectral
model with hydrostatic primitive equations as prog-
nostic equations. A two-time-level semi-implicit
semi-Lagrangian scheme is used for time integra-
tion. This scheme permits a longer time-step than
the formerly used semi-implicit Eulerian scheme
(JMA 2002) and realizes high e‰ciency. The semi-
Lagrangian advection scheme is vertically conser-
vative (Yoshimura and Matsumura 2003, 2005)
as well as globally conservative. The model we
use here has a horizontal resolution of TL159
(A120 km), with 48 layers in the vertical eta co-
ordinate. The top of the atmosphere (TOA) is
0.01 hPa, so the stratosphere is fully covered.
This subsection describes each physical process
scheme of MRI-AGCM3 used in MRI-CGCM3 to
participate in the CMIP5. Table A1 summarizes the
configuration of the atmospheric model. For refer-
ence, a di¤erent model configuration is also pre-
sented. This configuration uses the same MRI-
AGCM3 series to participate in the CMIP5 as a
di¤erent modeling group with a set of very-high-
resolution atmospheric model experiments (MRI-
AGCM3.2S) (Mizuta et al. 2011).
a. Cumulus convection
A new mass-flux cumulus scheme has been devel-
oped (Yoshimura et al. in preparation; hereafter,
the Yoshimura cumulus scheme; see also Yukimoto
et al. 2011, for a detailed description). In this
scheme, convective updrafts are calculated as de-
tailed entraining and detraining plumes, as in
Tiedtke (1989). It also represents multiple convec-
tive updrafts with di¤erent heights, as in an AS-
type scheme (Arakawa and Schubert 1974), by
considering continuous convective updrafts be-
tween the minimum and maximum turbulent
entrainment/detrainment rates (lmin and lmax ). At
the cloud bottom, lmin is set to 0:5104m1and
lmax is set to 3:0104m1. Magnitudes of the
convective updrafts are determined by using a clo-
sure assumption (Nordeng 1994), which is based
on convective available potential energy (CAPE).
Convective updrafts are assumed to have virtual
temperature, water vapor mixing ratios, and other
variables linearly interpolated between the two up-
drafts with lmin and lmax.
In addition to turbulent entrainment, two kinds
of organized entrainment are considered: entrain-
ment from a layer with high moist static energy
from which updrafts originate and entrainment
nearly proportional to the grid-scale mass conver-
gence and the convective updraft mass flux. When
the convective updraft becomes negatively buoyant,
an organized detrainment occurs at that level.
The scheme represents a convective downdraft
when a detrained air mixed with the environment
air becomes negatively buoyant. The turbulent
entrainment/detrainment rate for the convective
downdraft is 2:0104m1. The mass flux of the
convective downdraft is limited so as not to exceed
0.3 times the sum of the mass fluxes of the con-
vective updrafts. When the convective downdraft
becomes positively buoyant at some level, the en-
tire downdraft mass flux is detrained at that level
as organized detrainment. If the downdraft does
not become positively buoyant, organized detrain-
ment from the downdraft occurs at the sub-cloud
layer.
The vertical transport of horizontal momentum
by convection is calculated. The e¤ect of the sub-
grid-scale horizontal pressure gradient force is in-
troduced based on Gregory et al. (1997). This e¤ect
acts to adjust the direction of the horizontal wind in
convection toward that of the wind in the environ-
ment. Without this e¤ect, the momentum transport
is overestimated. The pressure gradient force is set
proportional to the vertical wind shear in the envi-
ronment.
b. Radiation
The radiation scheme is basically the same as
that of JMA’s operational global model (see details
in JMA 2007), but with some di¤erences (e.g., in
the interaction with aerosols in MRI-AGCM3). In-
frared (i.e., longwave (LW )) radiation of up to
3000 cm1and solar (shortwave (SW)) radiation
are treated separately. The radiative flux is calcu-
lated in 9 LW bands and 22 SW bands. In the LW
February 2012 S. YUKIMOTO et al. 25

and SW schemes, major absorptions due to water
vapor (line and continuum absorption), carbon di-
oxide (CO2) (e.g., in the 15 mm band and near-
infrared region), and ozone (O3) (in the 9.6 mm
band and the visible and ultraviolet regions) are
considered. In addition, absorptions due to
methane (CH4), dinitrogen monoxide (N2O), and
chlorofluorocarbons (CFCs; CFC-11, CFC-12,
and HCFC-22) are taken into account in the LW
scheme because of their greenhouse gas (GHG) ef-
fect. Absorption by oxygen (O2) and Rayleigh scat-
tering by molecules of atmospheric gas are also cal-
culated in the SW scheme.
To represent the direct e¤ects of aerosols, optical
parameters are configured for five aerosol species,
corresponding to those in aerosol model MASIN-
GAR mk-2. The extinction and absorption coe‰-
cients and asymmetry factors of these species are
computed under the assumption of Mie scattering
by spherical particles, by using the complex refrac-
tion index data of OPAC software (Hess et al.
1998). The dependence of hygroscopic species on
ambient relative humidity is also considered (Chin
et al. 2002).
The e¤ect of aerosols on the optical properties
of clouds (i.e., the first indirect e¤ect) is considered
in the configuration of the e¤ective radius of cloud
particles. The e¤ective radius of a liquid water
cloud particle is computed as a function of cloud
droplet number density, based on Liu et al. (2006)
and Peng and Lohmann (2003). Since simple cloud
microphysics is introduced for the cloud scheme
(see the next subsection), the aerosol indirect e¤ect
on ice-cloud particles is also considered (Lohmann
2002).
In terms of the optical properties of cloud par-
ticle, LW emissivity is parameterized depending on
cloud water content (Kiehl and Zender 1995), and
the corresponding absorption coe‰cient is parame-
terized as a function of e¤ective radius (Hu and
Stamnes 1993, for liquid water clouds, and Ebert
and Curry 1992, for ice clouds). Optical depth,
single-scattering albedo, and the asymmetry factor
in SW are similarly parameterized by cloud water
content and e¤ective radius (Slingo 1989, for water
clouds and Ebert and Curry 1992, for ice clouds).
The vertical overlap of clouds greatly influences
estimates of radiative fluxes in a cloudy column.
In the LW scheme, maximum-random overlap (Ge-
leyn and Hollingsworth 1979) is assumed. In the
SW scheme, total cloudiness in a column is first
computed according to the maximum-random
overlap assumption, and then random overlap is
assumed to solve radiative fluxes in a cloudy sub-
column.
Because of relatively high computational costs,
full radiation computations are calculated for every
two grids in the zonal direction, for every hour in
the SW region, and for every three hours in the
LW region.
c. Cloud model
Cloud droplet and ice crystal concentrations are
important variables for representing the detailed in-
direct e¤ects of aerosols. A new two-moment bulk
cloud scheme (the MRI-TMBC scheme) has been
developed by expanding the Tiedtke cloud scheme
(Tiedtke 1993; ECMWF 2004; Jakob 2000). This
scheme predicts the cloud liquid water (CLW ) mix-
ing ratio, cloud-ice mixing ratio, and cloud droplet
and ice crystal concentrations. It represents forma-
tion of cloud droplets and ice crystals by convective
processes (i.e., cumulus detrainment) and stratiform
processes, and phase changes by immersion freez-
ing, contact freezing, and homogeneous freezing.
The aerosol model is coupled to the MRI-TMBC
scheme for activation of aerosol species into cloud
droplets and ice crystals. The aerosol species em-
ployed in the cloud scheme are SOx þdimethyl sul-
fide (DMS), BC, OC, sea-salt (two bins of particle
size), and mineral dust (six bins of particle size).
The activation of some aerosols into cloud droplets
is based on the parameterizations of Abdul-Razzak
and Ghan (2000, 2002) and Takemura et al. (2005).
The activation of some aerosols into ice crystals is
based on the parameterizations of Bigg (1953) for
immersion freezing, Lohman and Diehl (2006) for
contact freezing, and Ka
¨rcher et al. (2006) for cir-
rus clouds. Cloud droplet and ice crystal concentra-
tions evaluated by the scheme reflect radiation pro-
cesses through their e¤ective radii.
The deposition terms are based on Murakami
(1990), the depositional growth terms are based
on Rutledge and Hobbs (1983), and the condensa-
tion and condensation growth terms are based on
Tiedtke (1993). Melting occurs when the atmo-
spheric temperature is above 273.15 K, and homo-
geneous freezing occurs at temperatures below
235.0 K. A semi-Lagrangian scheme is used for the
advection process. The precipitation process is basi-
cally the same as that of Tiedtke (1993), except the
parameterization of Rotstayn (2000) is adopted for
the rainfall term. The Bergeron-Findeisen process is
also incorporated into the MRI-TMBC scheme.
26 Journal of the Meteorological Society of Japan Vol. 90A

This process occurs when the cloud-ice mixing ratio
exceeds 0.5 mg kg1(Lohman et al. 2007). Further-
more, a saturation adjustment proposed by Tao
(1989) is introduced.
d. Planetary boundary layer
The planetary boundary layer (PBL) scheme in
JMA’s operational model is the turbulence model
advocated by Mellor and Yamada (1974, 1982)
(the MY scheme), which is a second-order closure
model based on the Reynolds averaging method.
Nakanishi (2001) and Nakanishi and Niino (2004,
2006, 2009) proposed an improved version of the
MY scheme (called the MYNN scheme) for deter-
mining closure constants, mixing length, and the
stability of time integration. We introduced the first
two improvements into MRI-AGCM3 as follows.
First, based on data obtained by large eddy simula-
tion (LES), the MYNN scheme re-evaluates the
closure constants and introduces terms for the e¤ect
of buoyancy and wind shear, which are neglected in
the MY model. Second, a new diagnostic equation
for the mixing length is proposed; it evaluates the
stability of the surface layer by using the Monin-
Obukhov length and the e¤ect of buoyancy (Hara
2007a, 2007b).
e. Land surface model
A new land surface model named the Hydrology,
Atmosphere and Land (HAL) model (Hosaka et al.
in preparation) has been developed for MRI-
AGCM3. HAL consists of three submodels: SiByl
(vegetation), SNOWA (snow), and SOILA (soil).
Vegetation submodel SiByl has surface vegeta-
tion processes similar to those of JMA/SiB (Hirai
et al. 2007). SiByl has two vegetation layers (can-
opy and grass) and calculates heat, moisture, and
momentum fluxes between the land surface and the
atmosphere. Precipitation interception and evapo-
transpiration are included as moisture processes.
Implemented heat processes include SW radiation
(direct/di¤use, visible/infrared), LW radiation, sen-
sible heat, latent heat, and ground heat fluxes. A
bulk formulation scheme (Louis 1979) is used for
estimating surface flux. Air surface information
(e.g., air temperature at 2 m height) is diagnosed.
Snow submodel SNOWA can have any number
of snow layers; the maximum value is set to eight
for the CMIP5 experiments. The number of layers
depends on the snow water equivalent (SWE) and
the snow accumulation history. The predicted vari-
ables for snow are temperature, SWE, density,
grain size, and the aerosol deposition contents of
each layer. Water phase change (snow melting) oc-
curs when the temperature of each layer exceeds
1C. The bottom snow melts when the tempera-
ture of the uppermost underlying soil layer is above
1C. Snow properties, including grain size, are pre-
dicted due to snow metamorphism (Niwano et al.
in preparation), and the snow albedo is diagnosed
from the aerosol mixing ratio, the snow properties,
and the temperature (Aoki et al. 2011).
Soil submodel SOILA can have any number of
soil layers. In the CMIP5 experiments, it is com-
posed of 14 soil layers with depths of 2, 3, 5, 10,
10, 20, 30, 30, 40, 100, 100, 150, 200, and 300 cm,
for a total depth of 10 m. The temperature of each
layer is predicted by solving heat-conduction equa-
tions. The water phase change occurs between
1C and 1C. The number of layers in which soil
moisture is predicted depends on the vegetation
type (350 cm in forest grids to 50 cm in desert
grids). The infiltration flux of liquid water is esti-
mated by solving the Darcy equation, in which
hydraulic conductivity depends on soil moisture.
Drainage (gravitational runo¤ ) occurs from the
bottom layer, and surface runo¤ occurs from the
top layer.
f. Ocean-surface process
An important function of the ocean-surface
scheme is calculating turbulent heat, moisture, and
momentum fluxes. A simple skin sea surface tem-
perature (SST) scheme is used in MRI-AGCM3
as the lower boundary over the sea surface. This
scheme is designed to represent short-term tempera-
ture variation (e.g., diurnal variation) at the air–sea
interface, caused by short-term variations in wind
and solar radiation. The scheme has one sub-skin
layer with a linear temperature profile. Bulk SST is
the temperature at the bottom of the sub-skin layer,
which is given by the first layer temperature of the
OGCM in MRI-CGCM3, and is fixed during each
time step in the calculation of the fluxes over the
sea surface.
The interface temperature is the temperature at
the top of the sub-skin layer, which is estimated
from heat fluxes (including radiation) to the atmo-
sphere and from the bottom of the sub-skin layer.
The interface temperature and the sensible and la-
tent heat fluxes to the atmosphere are determined
at the same time by an implicit method from verti-
cal di¤usivity, which is calculated in the PBL
scheme. The bulk coe‰cients for the heat flux esti-
mation follow Louis (1979) and Louis et al. (1982),
February 2012 S. YUKIMOTO et al. 27

except those for turbulent fluxes in an unstable state
follow Miller et al. (1992).
Since the sub-skin layer is mixed well with the
underlying layer under windy conditions (and vice
versa), the empirical coe‰cient for the heat flux
from the bottom of the layer depends on wind
speed; it is larger (smaller) when the wind is
stronger (weaker) (Kawai and Wada 2007; Gente-
mann et al. 2003; Castro et al. 2003). See Yukimoto
et al. (2011) for the detailed formulation. The
OGCM is thermally driven by the heat flux at the
bottom of the sub-skin layer.
g. Sea ice surface process
The sea ice surface temperature is a prognostic
variable in MRI-AGCM3, and its variation is cal-
culated at the same time as the sensible and latent
heat fluxes by the implicit method. The snow sur-
face processes are calculated in the same way as
those for snow on land when snow (either partial
or full coverage) is on the sea ice. The roughness
length is set to a constant (0.001 m). The other pa-
rameters are fundamentally the same as those for
the sea surface. The snow amount (water equiva-
lent) on sea ice is also a prognostic variable in
MRI-AGCM3. When there are multiple categories
for sea ice thickness in a single grid box, these
values are calculated for each category.
Albedos of ice and snow are separately esti-
mated, and the former is similar to that by Hunke
and Lipscomb (2006). The parameterization of
Aoki and Tanaka (2008) is applied for estimating
snow albedo, which accounts for the decline of al-
bedo due to pollution by aerosol deposition. More-
over, the e¤ect of solar penetration into the snow
on albedo is taken into account for both visible
wavelengths for ice and near-infrared wavelengths
for snow.
h. River-lake model and ice discharge from
ice-sheet
The river-flow component in MRI-AGCM3 is
the Global River model using TRIP (GRiveT).
The river-channel dataset we use for MRI-
CGCM3 is the 11version of the Total River
Integrated Pathway (TRIP) (Oki and Sud 1998).
River runo¤ is calculated by the land surface model
in MRI-AGCM3 and is transported by GRiveT to
the river mouths via TRIP. GRiveT also has a lake
in each TRIP grid, which is connected to the lake
surface component in MRI-AGCM3.
The river and lake water masses and their inter-
nal energies are predicted. The water runo¤ esti-
mated by the land surface component and pre-
cipitation minus evaporation by the lake-surface
component are input to GRiveT. In each grid, river
water flows at a constant velocity of 0.4 m s1. Half
of the river water flows into the lake, and the lake
water returns to the river, depending on the water
level, with a 10-day e-folding timescale. The inter-
nal energies are a¤ected only in the lake by the sur-
face heat fluxes estimated at the lake-surface com-
ponent, and are reflected on the ocean temperature
when they are discharged at the river mouth. Freez-
ing of the lake is not considered.
As MRI-CGCM3 does not include a dynamic
ice-sheet model, a simple model for snow on ice
sheet (SMIST) is used as the ice-sheet component
of MRI-CGCM3 in order to balance the global
ocean water mass in the unforced climate. When
the SWE is greater than 10 m over the land surface,
the excess snow is taken away from the land surface
model and passed to SMIST. The ice mass and
its energy are transported to the ocean coast by
SMIST and discharged into the ocean as an ice-
berg. The algorithm is almost the same as the
GRiveT algorithm, but without any lake.
2.2 Ocean-ice model (MRI.COM3)
The ocean-ice component of MRI-CGCM3
is MRI Community Ocean Model Version 3
(MRI.COM3). Users may refer to its reference
manual (Tsujino et al. 2010) for details. The free-
surface, depth-coordinate ocean component solves
primitive equations using Boussinesq and hydro-
static approximation. A split-explicit algorithm is
used for the barotropic and baroclinic parts of
the equations (Killworth et al. 1991). MRI.COM3
can be used to simulate ocean and sea ice with
various specific configurations. Here, we describe
the MRI.COM3 specifications used in the MRI-
CGCM3 (or MRI-ESM1) component. Since the
specification is very close to Tsujino et al. (2011),
see their description also.
a. Resolution and topography
Horizontal resolutions are 1longitude and 0.5
latitude. A generalized orthogonal coordinate sys-
tem is used in the Arctic region (latitudes higher
than 64N) with polar singularities at Siberia
(64N, 80E), Canada (64N, 100W), and the
South Pole (tripolar grid).
The ocean model consists of 50 vertical levels
plus a bottom boundary layer (BBL) (Nakano and
Suginohara 2002). The surface layer is 4 m thick,
and the upper layers above 1000 m are resolved
28 Journal of the Meteorological Society of Japan Vol. 90A

by 30 layers. Vertical levels shallower than 32 m
follow the surface topography, as in s-coordinate
models (Hasumi 2006). BBLs with a thickness of
50 m are added in dense-water formation regions
of the present climate (i.e., the northern North At-
lantic (50Nto70
N and 60Wto0
E) and the
Southern Ocean around the Antarctica (south of
60S)), assuming that the dense-water formation
regions do not change significantly in a warmer
climate.
Model topography is constructed from the
Global Gridded 2-minute Database (ETOPO2v2;
National Geophysical Data Center). The topogra-
phy of the model is modified to represent important
oceanic current systems, including those around
complex archipelagos such as the Philippine Is-
lands.
b. Transport algorithm
The generalized Arakawa scheme as described
by Ishizaki and Motoi (1999) is used to calculate
the momentum advection terms. This momentum
advection scheme conserves total momentum and
energy for three-dimensionally non-divergent flows
over arbitrary topographies, and total quasi-
enstrophy (ðqv=qxÞ2or ðqu=qyÞ2) for horizontally
non-divergent flows. A numerical advection scheme
based on conservation of second-order moments
(SOMs) (Prather 1986) is employed for advection
of all tracers (temperature, salinity, and biogeo-
chemical tracers). The SOM scheme is computa-
tionally stable and almost free of numerical di¤u-
sion; therefore, it can reproduce relatively realistic
tracer distributions in OGCMs (e.g., Hofmann and
Maqueda 2006).
c. Sub-grid scale mixing
A flow-dependent anisotropic horizontal viscos-
ity scheme (Smith and McWilliams 2003) is
adopted to reduce viscosity in the direction normal
to the flow (nN). Viscosity in the flow direction (nF)
is set as Smagorinsky harmonic viscosity (Smagor-
insky 1963), nF¼ð4D=pÞ2jDj, where Dis the grid
size and jDjis the strain rate. Anisotropic viscosity,
nN¼0:2nF, allows the equatorial undercurrent
to be narrow and swift, as observed. At the lateral
boundary, nNis set to be half of nFin order to
produce a Munk boundary layer. Isopycnal tracer
di¤usion (Redi 1982; Cox 1987) is used with a coef-
ficient 1000 m2s1. Eddy-induced transport is par-
ameterized as isopycnal layer thickness di¤usion
(Gent and McWilliams 1990) with a coe‰cient
varying in space and time (Visbeck et al. 1997).
Vertical di¤usivity and viscosity are set by the tur-
bulence closure model (Noh and Kim 1999; Noh
et al. 2005). Background vertical di¤usivity consists
of a horizontally uniform vertical profile, as pro-
posed by Tsujino et al. (2000), and parameteriza-
tion for the tidally driven mixing (St. Laurent et al.
2002) near the Kuril Islands and the Sea of
Okhotsk. The latter treatment is to realize the inter-
mediate layer ventilation in the northern North Pa-
cific (e.g., Nakamura et al. 2004) while avoiding
unexpected side e¤ects that might arise when this
parameterization is applied globally. Seawater den-
sities are calculated by an accurate equation of state
(Tsujino et al. 2010). The vertical gravitational in-
stabilities calculated by the model are completely
eliminated at each time step by a convective adjust-
ment scheme.
d. Sea ice model
The sea ice model solves the evolution of frac-
tional area, heat content, and thickness; the trans-
port of ice categorized according to its thickness;
and the dynamics of the grid-cell-averaged ice pack.
Using the coupler, the sea ice model sends surface
temperature, interior temperature, snow and ice
thicknesses, and fractional area to MRI-AGCM3,
and receives surface fluxes calculated by MRI-
AGCM3. The ice model is part of MRI.COM3,
and ice–ocean exchange processes are internal.
The thermodynamic part is based on Mellor
and Kantha (1989). For processes that are neither
explicitly discussed nor included by Mellor and
Kantha (1989) (e.g., categorization by thickness,
ridging, and rheology), we adopt those of the Los
Alamos sea ice model (CICE) (Hunke and Lips-
comb 2006). Fractional area, snow volume, ice vol-
ume, ice energy, and ice-surface temperature of
each thickness category are transported using the
multidimensional positive definite advection trans-
port algorithm (MPDATA) (Smolarkiewicz 1984).
See the MRI.COM3 reference manual (Tsujino
et al. 2010) for details.
2.3 Aerosol model (MASINGAR mk-2)
In MRI-CGCM3, atmospheric aerosols are cal-
culated with the Model of Aerosol Species in the
Global Atmosphere (MASINGAR) mk-2, which is
coupled with the Scup coupler library (Yoshimura
and Yukimoto 2008). This improved version of the
MASINGAR aerosol model (Tanaka et al. 2003)
treats five aerosol species: non-sea-salt sulfate, BC,
OC, sea-salt, and mineral dust. The grid resolution
of the model is variable, and the horizontal resolu-
February 2012 S. YUKIMOTO et al. 29

tion is set as TL95 (A180 km) for MRI-CGCM3,
which di¤ers from that of the coupled AGCM
(TL159, A120 km). The vertical layers are set the
same as those of AGCM.
a. Coupling with MRI-AGCM3
The aerosol model receives the meteorological
fields and surface conditions from MRI-AGCM3
through the coupler. The meteorological fields in-
clude horizontal wind components, air temperature,
specific humidity, convective mass flux, precipita-
tion and evaporation with convective and large-
scale clouds, the vertical eddy di¤usion coe‰cient,
and surface pressure. The surface conditions in-
clude near-surface wind speed, surface air temper-
ature (SAT), ground temperature, SST, sea ice
coverage, snow amount, land-use type, vegetation
amount, and leaf area index. MASINGAR mk-2
sends the concentrations of the five aerosol species
and the deposition fluxes of absorptive aerosols
(BC and mineral dust) to MRI-AGCM3 to be
used in calculating the direct and indirect radiative
e¤ects of the aerosols and of the snow albedo. Sea-
salt aerosol is calculated in six size bins in the
aerosol model, but is sent in two bins (smaller and
larger than 1 mm).
b. Atmospheric transport
Atmospheric transport is calculated using a semi-
Lagrangian advection scheme and schemes for sub-
grid turbulent vertical di¤usion and convective
transport. The vertical eddy di¤usion coe‰cient
is taken from that of water vapor, calculated in
MRI-AGCM3. Convective transport is calculated
using the updraft mass flux from the cumulus con-
vection scheme in MRI-AGCM3.
Aerosol particles are subject to gravitational set-
tling relative to air motion and are assumed to fall
with terminal velocity Vs. The terminal velocity Vs
is calculated under the assumption that the particles
are spherical and is proportional to the square of
the radius of the particle and the Cunningham slip-
flow correction.
c. Dry and wet deposition processes
Dry deposition is parameterized by the
resistance-in-series model (Seinfeld and Pandis
1997), which includes turbulent impaction and
gravitational settling. Wet deposition is distin-
guished between in-cloud and below-cloud scaveng-
ing and is categorized by cloud type (convective or
large-scale) and species (aerosol or gas; accommo-
dation with water droplets). For in-cloud scaveng-
ing by large-scale precipitation, we use the parame-
terization developed by Giorgi and Chameides
(1986). Both the dry and wet deposition schemes
for sea-salt and mineral dust aerosols are particle-
size-dependent.
For water-soluble gases such as SO2, wet
scavenging is calculated simultaneously with the
aqueous-phase chemistry. The fraction of a water-
soluble gas that is in liquid water is assumed
to follow Henry’s law of equilibrium, which is
temperature- and pH-dependent. Evaporation of
rainwater is considered when calculating the wet
deposition rate. When evaporation occurs, a frac-
tion of the trace elements is released back to the
air.
d. Emission processes and chemical reactions
Prescribed emission inventories are used for the
emission processes of anthropogenic sulfur com-
pounds and carbonaceous aerosols. The emissions
of oceanic DMS, sea-salt, and mineral dust are cal-
culated from the meteorological and surface condi-
tions. See Yukimoto et al. (2011) for the detailed
formulations.
2.4 Coupler (Scup)
Scup (Yoshimura and Yukimoto 2008) was de-
veloped at MRI as a simple general-purpose cou-
pler for coupling component models for integration
into an ESM. Each component model in MRI-
CGCM3 (or MRI-ESM1) (atmospheric, ocean-ice,
aerosol, and atmospheric chemistry models) uses
Scup to exchange data with the other component
models. Scup makes it easy to develop an inte-
grated model composed of an arbitrary combina-
tion of these component models. Distributing the
communications reduces the amount of transferred
data and leads to high computational e‰ciency.
Using the settings in the Scup configuration file,
the models can be executed in parallel or sequen-
tially; accordingly, we can distribute the execu-
tion of the component models in the most e‰cient
way on the computer being used. For the CMIP5
experiments, the coupling interval is 1 hour for
atmosphere-ocean coupling (including the sea ice
model) and 30 minutes for atmosphere-chemistry
coupling (in the aerosol model).
3. Spin-up and basic experiments
In this section, we first describe how the initial
state of the climate system in the model is made
(i.e., spin-up procedure) for the base-line ( pi-
Control) experiment. We then describe the design
30 Journal of the Meteorological Society of Japan Vol. 90A

of the basic experiments introduced in the present
paper.
3.1 Spin-up procedure
Before coupling the atmosphere and the ocean,
initial atmospheric and land surface data is taken
from JMA’s operational analysis for 00UTC, July
9, 2002, and data on the initial oceanic state are
taken from the 2360-year integration of the
OGCM forced by the interannually varying forcing
boundary conditions of Version 2 of the Coordi-
nated Ocean-ice Reference Experiments (CORE-2)
datasets (Large and Yeager 2009). The initial distri-
butions of aerosol species and their precursor gases
are set to zero, except carbonyl sulfide (OCS) below
100 hPa, which is set to 500 pptv.
A spin-up run with the coupled model MRI-
CGCM3 was then started and continued for 305
years under the present-day forcing agents (e.g.,
GHG concentration and solar irradiance). After
the spin-up of the present-day condition, GHG
concentration is gradually decreased (during 53
years) to that of the year 1850. An additional 62-
year spin-up is then conducted under the fixed year
1850 condition in order to derive the initial state
of the piControl experiment. During this spin-up
phase, we tune some parameters slightly. We also
try to improve the sea ice distribution bias by re-
storing sea surface salinity (SSS) to the observed cli-
matology for only five years at the beginning of the
last 62-year spin-up, but with no apparent e¤ect on
the final spin-up state. Since MRI-CGCM3 uses a
vegetation map from the USGS-GLCC land-use
data and SAGE Global Potential Vegetation Data-
set (Ramankutty and Foley 1999) as the standard
(preset-day) condition, the spin-up starts with the
standard condition but is replaced with the 1850
condition at the beginning of the last 62-year spin-
up run.
3.2 Experiment design
Numerous coordinated experiments in CMIP5
(Taylor et al. 20112) have been conducted. Of
these, some of the core experiments for evaluat-
ing the model’s basic performance are described
here.
a. piControl experiment
The piControl experiment serves as the baseline
for all other experiments in CMIP5. According to
the CMIP5 protocol, all external forcing agents
have the same values at year 1850 as for the his-
torical experiment (see below). Concentrations of
GHGs and anthropogenic aerosols or their precur-
sors are fixed at 1850 values of the Representative
Concentration Pathways (RCP) database.3For
example, the concentrations of CO2,CH
4, and
N2O are 284.725 ppmv, 790.97924 ppbv, and
275.42506 ppbv. Emissions from eruptive volca-
noes are not included in the piControl experiment.
The initial condition of the piControl experiment
is the spin-up state described above. The simulation
period for the piControl experiment is planned to
be 500 years or more. In the present study, we pres-
ent results for the first 500 years.
b. historical experiment
The historical experiment is designed to evaluate
how realistically the model can simulate the present
climate and the recent past climate changes. Also, it
provides initial conditions for future RCP scenario
experiments and decadal prediction experiments of
the CMIP5. The length of the experiment is 156
years, starting from 1850 (1850 through 2005). The
reproducibility of the era since the mid-nineteenth
century is particularly important, since instrument
observations are available for that era.
Historical records of concentrations of GHGs
and anthropogenic aerosols or their precursors in
the RCP database are used. Emission fluxes of sea-
salt, mineral dust, and oceanic DMS are calculated
in the aerosol model, as described in the previous
section. Emission of terrestrial biogenic DMS is
adopted from the inventories compiled by Spiro
et al. (1992). Historical SO2emission from sporadi-
cally erupting volcanoes is compiled from the
Global Emissions Inventory Activity (GEIA) data-
base (Andres and Kasgnoc 1998), total ozone map-
ping spectrometer (TOMS) satellite measurement
(Bluth et al. 1997), and ground-based observations
of aerosol characteristics from pre-satellite era spec-
tral extinction measurements (Stothers 1996, 2001).
The SO2emission from non-eruptive volcanoes is
set as invariant throughout the piControl and his-
torical experiments. Historical ozone concentration
is taken from the Atmospheric Chemistry and Cli-
mate (AC&C) and Stratospheric Processes And
their Role in Climate (SPARC) database.4The his-
2 available from http://cmip-pcmdi.llnl.gov/cmip5/
experiment_design.html
3 available from http://www.iiasa.ac.at/web-apps/tnt/
RcpDb
4 available from http://www.pa.op.dlr.de/CCMVal/
AC&CSPARC_O3Database_CMIP5.html
February 2012 S. YUKIMOTO et al. 31

torical solar forcing data is taken from the database
reconstructed for 1850 through 2005 based on Lean
et al. (2005).
Change of land-use is evaluated by using the
land-use type (LUT) datasets (gcrop and gpast)
provided by the CMIP5 land-use group for 1700
through 2100. A normalized and monotonically in-
creasing ‘‘gcrop þgpast’’ function is calculated at
each model grid as the index of vegetation type
change from forest to grass. By using the index,
land-use changes are reflected in the standard vege-
tation type (assumed as for year 1990).
The historical experiment consists of three mem-
ber ensemble runs. Each initial state is taken from
years 100, 130, and 160 of the piControl experi-
ment. Conditions other than those described above
are the same as in the piControl experiment.
c. abrupt4xCO2 experiment
An experiment with abrupt quadrupling of CO2
concentration (abrupt4xCO2) is executed to evalu-
ate the equilibrium climate sensitivity of the model
following the Gregory regression approach (Greg-
ory et al. 2004). The experiment consists of 5-year,
12-member ensemble runs initiated from each
month of year 40 in the piControl run. The evalua-
tion becomes more stable when ensemble runs are
used, since each run has a short response that tends
to be disturbed by interannual variations. Only
the first run (started in January) is extended to 150
years long to examine the long response. Condi-
tions other than those described above are the
same as in the piControl experiment.
d. 1pctCO2 experiment
A run with the idealized 1 % yr1increase of
CO2concentration is executed to measure the tran-
sient climate response (TCR). This run allows an
idealized climate response without such complica-
tions as aerosols and land-use changes. The results
of this experiment can be compared with previous
CMIP model results (e.g., CMIP3). The initial state
of the experiment is taken from year 40 of the pi-
Control experiment. Except for the increased CO2
concentration, all the conditions are the same as in
the piControl experiment.
4. Model stability and drift
The climate system requires millennia to achieve
an equilibrium state, even if imbalance of the global
radiation budget at the TOA is very small. The sta-
bility and drift should be carefully checked for the
piControl run, to determine whether the spin-up is
su‰cient or not.
The temporal variation of the globally averaged
annual mean SAT (2-m screen level) for the pi-
Control experiment is presented in Fig. 1a. It is
su‰ciently stable for a long term (at least 500
years), and there is almost no climate drift. The
500-year average temperature is 13.6C, and the
linear trend of þ0.016Cyr
1over 500 years is
not statistically significant at the 99% confidence
level.
Figure 2a presents the global radiation budget
at TOA, which is downward solar irradiance minus
reflected solar radiation and outgoing LW radia-
Fig. 1. Time series of globally averaged
annual-mean surface air temperature
(SAT) for the (a) piControl experiment
(thin solid line) and (b) historical experi-
ment (solid line, ensemble mean; ‘x’, each
member), and for the observation (grey
lines). Eleven-year running mean (thick
line in panel a) and linear trend (dashed
lines) for the piControl experiment are
also plotted.
32 Journal of the Meteorological Society of Japan Vol. 90A

tion (OLR). The 0.5 W m2downward net imbal-
ance seems very stable except for interannual varia-
tion. In addition, there is an unknown energy
source of 0.5 W m2in the atmosphere, due to the
lack of strict energy conservation. We do not be-
lieve that this unknown energy source is a critical
issue in evaluating climate change, since the value
does not change throughout the period or in other
experiments (e.g., the 1pctCO2 experiment, not
shown). Consequently, there is a net energy input
(or downward flux) of about 1 W m2at the sur-
face. The land surface model is well-conserved,
and the annual mean energy budget is almost zero,
which leads to net energy absorption by the ocean
due to the surface energy imbalance.
We next examine the drift of the ocean. Figure
2b depicts temporal variations of globally averaged
annual mean SST, and vertically averaged ocean
temperature for the upper 635 m (VAT635) and
for full depth (VATFULL). The global SST is
18.1C on average and exhibits no significant trend.
However, a warming trend is apparent in the time
series of VAT635, which represents the ocean sub-
surface layer. A relatively large warming trend of
0.1Cyr
1occurs 200 years from the beginning,
though after that the trend becomes much smaller.
The variation of VATFULL reflects an energy im-
balance at the surface, since the ocean-ice model
conserves heat (and mass) with high precision. A
small warming trend (0.017C/100yr) is found in
the first 200 years, followed by a reduced warming
trend. The temporal change of VATFULL toward
equilibrium is likely to be a logarithm function,
and the trend is expected to become gradually
weaker. We estimate the logarithm function for the
500-year VATFULL to be 0:0112 lnðnÞþ4:1837,
where nis years from the start. Thus, VATFULL
is estimated to be 4.279C (4.287C), and in-
creases 0.095C (0.103C) by 5000 (10000) years
later. It is reasonable to assume that the SST in-
creases similarly to VATFULL in the equilibrium
state. The temperature change (<0.1C in 10000
years) is su‰ciently small compared to the climate
change in the 20th and 21st Centuries that we are
targeting.
The climate system cannot be stable if the gen-
eral circulation of the oceans is not stable, even
though the drift in the globally averaged ocean
temperature is su‰ciently small. It is necessary to
examine how realistic and how stable the model
simulates meridional overturning circulation
(MOC) and other important volume transports in
the oceans, particularly for evaluating long-term
climate change. Figure 3 illustrates temporal varia-
tions of volume transports for the Atlantic MOC at
45N, the Antarctic MOC at 70S–65S, and the
Antarctic Circumpolar Current (ACC). The MOCs
are closely related to formation of the North Atlan-
tic deep water (NADW) and the Antarctic bottom
water (AABW), and are associated with a major
ocean heat up-take. The Atlantic MOC fluctuates
from 12 through 18 Sv (106m3s1) and is rather
stable at an average of 15 Sv throughout the simu-
lation. The observation by Talley et al. (2003) is
18 Sv with a 3 to 5 Sv error. The simulated value
is within the observation error and is hence realis-
tic. The Antarctic MOC is only around 4 Sv at the
beginning of the experiment; however, it increases
within the next 50 years and then almost stabilizes
Fig. 2. Time series of globally averaged an-
nual mean (a) radiation imbalances (posi-
tive downward) at the top (solid line) and
the bottom (dashed line) of the atmo-
sphere, (b) sea surface temperature (SST)
(solid line, left axis), upper 635 m (dashed
line, right inner axis), and total (grey line,
right outer axis) averaged ocean tempera-
ture of the piControl experiment. Eleven-
year running means are also plotted with
thick solid lines.
February 2012 S. YUKIMOTO et al. 33

at 8 Sv. Estimated values (e.g., Talley et al. 2003;
Sloyan and Rintoul 2001) range from 20 Sv to
50 Sv, though they are thought to have much uncer-
tainty. The simulated value, however, is somewhat
smaller than the observed value. The simulated
ACC demonstrates a small but very long-term vari-
ation between 110 Sv and 120 Sv. Transport of the
ACC is estimated at 134 Sv, based on observation
by Cunningham et al. (2003). The simulated trans-
port of the ACC is also probably underestimated.
We next examine the global water budget.
Global sea level change would occur with even a
small water budget imbalance, since the OGCM ex-
changes real freshwater flux in MRI-CGCM3 (not
virtual salt flux, as in MRI-CGCM2.3.2). It has
been confirmed that the model conserves the global
water with high precision. The trend of the sea level
attributable to water budget imbalance in the
piControl experiment is less than 2 mm/100yr,
which we believe is su‰ciently small to evaluate
sea level change due to thermal expansion and a
hydrological cycle change by anthropogenic GHG
increase. Combining the e¤ects of thermal imbal-
ance in the piControl experiment results in an
increasing trend of approximately 5 mm/100yr in
global sea level.
5. Climate sensitivity and historical climate change
Various methods of estimating the climate sensi-
tivity of models have been proposed since the 1980s
(e.g., Hansen et al. 1984). Each method has both
advantages and disadvantages. The method with
the slab mixed layer ocean coupled to AGCM was
generally used to obtain an equilibrium response,
but has some drawbacks, especially in including
the e¤ects of oceanic circulations. Recently, Greg-
ory et al. (2004) proposed a new, simple method
by regressing the radiative flux at TOA against the
global average SAT, which is recommended by
CMIP5.
Figure 4 depicts the Gregory plot (Gregory et al.
2004) for the abrupt4xCO2 ensemble simulations
with MRI-CGCM3. The regression line crosses at
the change in global SAT DT¼4:22 K with the
axis of net radiative flux change at TOA DR¼0,
which means that the radiative flux balances
Fig. 3. Time series of meridional overturn-
ing volume transport (unit: Sv ¼
106m3s1) in (a) the Atlantic (45N ) and
(b) the Antarctic (70Sto65
S), and (c)
ACC in the piControl experiment.
Fig. 4. Scatter plot and its linear regression
of globally averaged annual mean values.
Downward radiation flux change at TOA
versus SAT change from 12-member en-
semble (5 years each) abrupt4xCO2 experi-
ment, for net radiation (RT), clear-sky
longwave (LN), clear-sky shortwave (SN),
longwave cloud forcing (LC), and short-
wave cloud forcing (SC).
34 Journal of the Meteorological Society of Japan Vol. 90A

(DR¼0) if the SAT change achieves equilibrium
(with DT¼4:22 K) for the quadrupling of CO2
concentration. The regression line crosses at DR¼
7:66 W m2with the axis of DT¼0, which means
the radiative forcing is 7.66 W m2for quadru-
pling of CO2concentration. The radiative forcing
for doubling CO2concentration is estimated as
3.83 W m2, since radiative forcing is known to be
approximately linear relative to CO2concentration.
The radiative forcing is comparable to 3.7 W m2
of MRI-CGCM2.3.2 using a di¤erent radiation
scheme (Shibata and Aoki 1989; Shibata and
Uchiyama 1992). Consequently, the climate sensi-
tivity of MRI-CGCM3 is 2.11 K (¼4.22/2) for
doubling of CO2concentration. This value is mar-
ginally in the lower range of multi-model results
(2.1 K to 4.4 K) estimated from slab mixed layer
ocean equilibrium in IPCC-AR4 and is only 73%
of the climate sensitivity 2.9 K (equilibrium re-
sponse with slab ocean) of MRI-CGCM2.3.2 (Yu-
kimoto et al. 2006). It is di‰cult to estimate real
climate sensitivity from the observations, due to
complicated variations induced by various forcings
and internal climate variability. Therefore, the re-
sults do not necessarily mean that the model under-
estimates climate sensitivity.
Another orthodox measure of climate response
to a forcing is evaluating TCR with an ideal 1 %
yr1CO2increase. The result (not shown) demon-
strates global SAT increases 1.6 K for doubling
and 4.1 K for quadrupling CO2concentration.
These increases are 84% and 89% relative to those
for MRI-CGCM2.3.2.
Yukimoto et al. (2006) concluded that improving
the cloud basic state results in positive cloud feed-
back in the tropics and leads to greater climate sen-
sitivity. In MRI-CGCM3, the cloud model is so-
phisticated, and the cloud variables are predicted
based on physical theories, keeping satisfactory re-
alistic distributions for the cloud forcings for the
present climate (presented later). Furthermore, the
cloud microphysics in the model is fully interactive
with variations of aerosols, taking into account
direct and indirect aerosol e¤ects. More detailed
analysis of the experiments of MRI-CGCM3 will
lead to a better understanding of the mechanism
associated with cloud and climate sensitivity. This
subject will be investigated in future studies.
The globally averaged SAT variations in the
three-member ensemble historical experiment are
plotted in Fig. 1b. In contrast with the piControl
experiment, the simulated SAT in the historical
run demonstrates a moderate increase from the be-
ginning through the middle of the 20th Century
after recovery from consistently downward spikes
in 1883 and 1902 associated with the eruptions of
Kurakatau and Santa Marı
´a. A slight decrease
of SAT is observed during the 1940s through the
1960s, followed by a rapid increase since the 1970s.
These multi-decadal tendencies are consistently si-
mulated in all the members and are similar to the
observation (HADCRUT3v) (Brohan et al. 2006).
The ensemble mean of the 2001 to 2005 average is
14.17C, and the increase relative to the 1850 to
1899 average is 0.59C. This value is 0.17C smaller
than that in IPCC-AR4 (0:76 G0:19C), though it
is marginally in the lower limit of the estimation
error range.
6. Mean climate and variability
Reproducing the present-day climate is an essen-
tial requirement for models used for future climate
change projection. A recent period of a few decades
is favorable for comparing the simulated fields with
the observation, since abundant instrument obser-
vations, including satellite observations, are avail-
able for this period. Available multiple reanalysis
data for this period enable us to verify the model
rather conveniently. We use JRA-25 (Onogi et al.
2007) for the present verification.
6.1 Mean climate
We compare the simulated mean climate of the
last 27 years of the historical experiment (1979
through 2005) with observations of the same pe-
riod. To evaluate the mean climate, we use the en-
semble mean of three members of the historical
experiment.
Representative globally averaged mean values
are compared among the observations, MRI-
CGCM3, and MRI-CGCM2.3.2 (Table 1). The
downward solar radiation at TOA of 341.7 W m2
in MRI-CGCM3 is reasonably accurate, since it
depends only on the solar input (forcing) and the
Earth’s orbital parameters for the present day. At
TOA, the upward SW radiation is 103.3 W m2,
and the LW radiation is 237.6 W m2, in agree-
ment with the satellite observations of the ISCCP
FD dataset (Zhang et al. 2004) and CERES obser-
vation adjusted by Loeb et al. (2009) within the
error range of satellite observations. The upward
SW and LW radiation at TOA are also close to
those of the present-day control experiment of
MRI-CGCM2.3.2 (di¤erences within 2 W m2).
February 2012 S. YUKIMOTO et al. 35

The net radiation at TOA is downward
0.85 W m2, indicating slight heating of the system.
Generally, SW radiation at TOA depends
strongly on clouds, particularly tropical and sub-
tropical low-level clouds. LW radiation at TOA,
however, depends mainly on the atmospheric tem-
perature and high clouds associated with tropical
convective activity. Agreement of both SW and
LW radiation with the observation supports the
realistic simulation of cloud height distribution.
Cloud forcing is simulated to be 46:3Wm
2for
SW radiation and þ23.7 W m2for LW radiation,
in reasonable agreement with the observations.
The energy budget at the surface exhibits down-
ward 1.30 W m2that includes an unknown energy
source of 0.45 W m2in the atmosphere, which is
as large as in the piControl experiment. The ocean
is heated by an additional 0.31 W m2in 1979
through 2005, compared to the average in the pi-
Control experiment (0.99 W m2for the corre-
sponding period). Therefore, the change in surface
energy budget corresponds to the ocean heat up-
take due to the simulated climate change. The value
is one third of a recent observation of 0.9 W m2
(Trenberth et al. 2009). However, other recent ob-
servations of the upper ocean heat content change
are 0:582 G0:151 1022 Jyr
1for 1993 through
2005 (Ishii and Kimoto 2009) and 0:73 1022 J
yr1for 1993 through 2007 (Levitus et al. 2009).
These values correspond to 0.36 W m2and
0.45 W m2for ocean heat up-take.
Each energy flux at the surface (downward SW
and LW radiation) reflects SW and upward LW ra-
diation, and sensible and latent heat fluxes. These
results are satisfactorily close to the new estimation
by Trenberth et al. (2009) based on the new satellite
observations and three reanalyses. The latent flux,
however, is somewhat overestimated, which is con-
sistent with the larger global mean precipitation
(2.90 mm day1) compared to the CMAP observa-
tion (Xie and Arkin 1997) of 2.67 mm day1for
the same period. This overestimation of global pre-
cipitation is common in many models and reanaly-
ses (e.g., Onogi et al. 2007). More precipitation
means a larger global water cycle that probably
leads to overestimated poleward water transport,
which in turn suggests freshening the upper oceans
at high latitudes.
Meridional distribution of the radiation budget
at TOA regulates meridional energy transport by
Table 1. Globally averaged radiation and cloud radiative forcing (CRF ) at TOA and radiative and heat fluxes at the
surface (SFC), and SAT and precipitation for the observation, the MRI-CGCM3 experiments ‘piControl’ and en-
semble mean of historical, and the present-day control experiment with MRI-CGCM2.3.2. Units for radiation
(heat) fluxes are W m2.
Experiment Observations
MRI-CGCM3
piControl
MRI-CGCM3
historical
MRI-CGCM2.3.2
present-day
Incoming solar radiation 341.3 341.61 341.72 342
Reflected solar at TOA 101.9 101.04 103.26 102
Outgoing longwave radiation 238.5 240.05 237.61 236
Net downward radiation at TOA 0.9 0.51 0.85 4
SW CRF at TOA 51.0a/46.6b44.93 46.26 49
LW CRF at TOA 26.5a/29.5b23.29 23.65 22
Net CRF at TOA 24.5a/17.1b21.65 22.61 27
Downward SW radiation at SFC 184 199.89 197.58 197
Absorbed SW radiation at SFC 161 168.56 166.02 165
Downward LW radiation at SFC 333 329.86 333.00 337
Net LW radiation at SFC 63 64.41 63.16 59
Sensible heat flux at SFC 17 17.97 17.34 31
Latent heat flux at SFC 80 85.19 84.23 75
Net downward energy at SFC 0.9 0.99 1.30 0
SAT (C) 14 13.62 13.99 13.65
Precipitation (mm day1) 2.67 2.94 2.90 2.6
a: ISCCP FD (Zhang et al. 2004)
b: CERES adjusted (Loeb et al. 2009)
other values: (Trenberth et al. 2009)
36 Journal of the Meteorological Society of Japan Vol. 90A

the climate system. Figure 5 depicts the simulated
northward energy transports by the system, the at-
mosphere, and the oceans, along with those esti-
mated based on the observation (Trenberth and
Caron 2001; Fasullo and Trenberth 2008). Oceanic
energy transport is implied from the integrated
ocean surface heat flux. Atmospheric transport is
calculated by subtracting the implied oceanic trans-
port from the total energy transport produced by
the system. The unknown energy source in the at-
mosphere is also subtracted, on the assumption
that it is globally uniform. We confirmed that result
from another assumption (where the unknown en-
ergy source is proportional to the SAT ) is very sim-
ilar to this one.
The simulated energy transport by the system
agrees well with the observational estimations in
the Northern Hemisphere (NH) but is under-
estimated approximately 1 PW in the tropical and
mid-latitude Southern Hemisphere (SH). This bias
arises from overestimated SW radiation input in
the Southern Ocean. The atmospheric transport,
however, is realistically simulated at all latitudes.
Results indicate a 4.9 PW northward peak at 40N
and a 5.0 PW southward peak at 40S. Due to the
radiation bias in the SH, the oceanic transport indi-
cates a large (1 PW) deficit in SH low latitudes.
The model simulates only a 0.4 PW peak of the
southward transport, while 1 PW is observed at
10S. The oceanic transport in the NH is generally
well-simulated, despite a slightly weaker northward
transport than observed in the subtropics.
Distribution of simulated annual mean SST is
compared with the observation (COBE) (Ishii et al.
2005) (Fig. 6). The model generally reproduces the
basin-scale SST distribution fairly well. The warm
pool in the Indo-Pacific region and the relatively
high SST in the western tropical Atlantic are simu-
lated. Northward bulges of the contours around the
Kuroshio and the Gulf Stream are slightly exagger-
ated compared to those observed. The cold tongue
along the equator in the Pacific is simulated, al-
though it is too strong.
A hemispheric bias is apparent in the di¤erence
map (Fig. 6c); it is generally colder in the NH and
warmer in the SH than actually observed. The
Kuroshio Extension region and the Labrador Sea
through the Greenland-Iceland-Norwegian (GIN )
Seas exhibit strong cold biases exceeding 4C. The
latter cold bias is accompanied by an excessive
extent of sea ice in winter (discussed later). The
SST in most of the tropical oceans exhibits small
biases of less than 1C, except for a slightly cold
equatorial bias associated with the exaggerated
cold tongue in the Pacific. Warm biases are com-
mon in the subtropical eastern part of the southern
basins, where marine low-level clouds predominate.
Warm bias is also dominant across the Southern
Ocean.
Some of the SST errors are related to radiation
biases. Figure 7 depicts the simulated annual mean
SW and LW upward radiation at TOA and their
di¤erence from the ERBE observation (Barkstrom
et al. 1989). The simulated overall distribution of
SW and LW radiation agrees with the observations.
Large reflected SW appears in the tropical regions
(e.g., over the Maritime Continent and the Amazon
basin). Consistent with the large SW, the regions
display smaller OLR, suggesting dominant active
convection. The reflected SW is low and the OLR
is high over the tropical and subtropical oceans
where subsidence is dominant.
The di¤erences from the observation for both
SW and LW are generally small, with magnitudes
of less than 10 W m2in most global areas. How-
ever, there are some noticeable biases, which are
thought to be consistent with errors in cloud distri-
bution. Reflected SW is overestimated and OLR is
Fig. 5. Northward heat transport by the sys-
tem (black), the atmosphere (red), and the
oceans (blue) for 1979 through 2005 aver-
age of the historical experiment (solid lines)
and observational estimations (dots, Tren-
berth and Caron 2001; crosses, Fasullo
and Trenberth 2008).
February 2012 S. YUKIMOTO et al. 37

Fig. 6. Annual mean SST (a) observed and (b) simulated in the historical experiment, and (c) its bias
(simulation–observation) for 1979 through 2005 average. Units are C.
38 Journal of the Meteorological Society of Japan Vol. 90A

Fig. 7. Annual mean radiation fluxes at the top of the atmosphere (TOA) for (a) reflected shortwave, (b) outgoing longwave in the histor-
ical experiment (1979 through 2005 average), and di¤erences from the ERBE observation (c, d). Units are W m2.
February 2012 S. YUKIMOTO et al. 39

underestimated over the Maritime Continent, the
tropical western Indian Ocean, and the tropical
Atlantic coast of South America, suggesting too
strong a cloud forcing. In the equatorial western
Pacific, a distinct overestimated OLR implies a
lack of convection, possibly attributable to the ex-
aggerated cold tongue. Across the Southern Ocean,
the cloud forcing is underestimated, as implied
from the negative bias in SW reflection and the
overestimated OLR. This underestimated cloud
forcing leads to an excessive energy input in the
SH mid-latitude and a consequently underesti-
mated required poleward energy transport in the
SH (Fig. 5).
Geographical SAT distributions reproduced by
the model for June through August (JJA) and De-
cember through February (DJF) seasonal averages
are presented in Fig. 8. To facilitate evaluating the
model, di¤erences from observations over the land
(the latest CRU TS3.1 dataset based on Mitchell
and Jones (2005)) are also depicted. SAT over land
is generally subject to hydrological conditions (e.g.,
precipitation, snow cover, and soil wetness), de-
pending on the season. The model realistically re-
produces the overall SAT distributions for both
winter and summer. In JJA, the bias is less than
2C in most of the continental regions. Cold biases
exceeding 2C in Scandinavia through western Rus-
sia, northeastern Canada, and the subtropical NH
in DJF may be associated with the dominant SST
cold bias in the NH (Fig. 6c) and overestimated
sea ice in the North Atlantic.
Distributions of seasonally averaged precipita-
tion for JJA and DJF are illustrated in Fig. 9. The
large-scale precipitation pattern associated with the
summer Asian monsoon in JJA is fairly well simu-
lated. Active convection regions over the Indochi-
nese Peninsula and the South China Sea through
the western tropical Pacific east of the Philippines
are qualitatively realistic. The precipitation band
associated with the Baiu front is satisfactorily real-
istic. The northern intertropical convergence zone
(ITCZ) along the 10Nto15
N latitude band
is also qualitatively well simulated, with a con-
centrated precipitation band extending from the
middle-eastern Pacific through northern South
America, the tropical Atlantic, and western Africa.
For the austral summer (DJF), the heavy precip-
itation associated with South American and the
South African monsoons is fairly well simulated.
The distinct precipitation in the North Pacific and
the North Atlantic associated with winter storm
tracks is also realistic.
Precipitation biases exceeding 1 mm day1are
limited but prominent in the tropics, where strong
precipitation is primarily observed. The Indian
Monsoon precipitation (JJA) is significantly under-
estimated over the Indian subcontinent, the Bay of
Bengal, and the western coast of India. The warm
bias in the SAT in JJA (Fig. 8c) is possibly associ-
ated with dry bias due to the shortage of Indian
Monsoon precipitation. The cold SST bias in the
northern Arabian Sea (Fig. 6c) is a possible reason
for this shortage. In contrast, the western Pacific
ITCZ has too much precipitation. Separate test
runs with modified cumulus convection parameters
(not shown) exhibit a negative correlation between
Indian and western Pacific precipitation. There is
unrealistic precipitation along the south o¤ the
equatorial Pacific, extending from east of New
Guinea. The overestimated precipitation o¤ the
equator in the SH becomes more distinct in aus-
tral summer (DJF), and underestimated equatorial
western Pacific precipitation results in the double
ITCZ. This problem is discussed in the next section.
Snow coverage, one of the most important fac-
tors a¤ecting the SAT over land, is evaluated for
geographical distributions for the boreal fall (Sep-
tember through November (SON )), winter (DJF),
and spring (March through May (MAM)) (Fig.
10). In comparison with the observations (Arm-
strong and Brodzik 2005; dataset obtained from
the National Snow and Ice Data Center (NSIDC)),
the model reasonably reproduces the observed dis-
tribution with its seasonal march. However, the
snow cover is slightly excessive in spring and fall in
the western part of the Eurasian continent. Also,
the simulated snow cover is excessive over the Tibe-
tan Plateau during all seasons.
The horizontal distribution of column-integrated
CLW content is realistically simulated compared
with Special Sensor Microwave/Imager (SSM/I )
observations (Wentz 1997) (Fig. 11). The simu-
lated values are 0.10 through 0.12 kg m2in the
regions along the storm tracks, and 0.06 through
0.08 kg m2in the regions of subtropical subsi-
dence in the North Pacific and the North Atlantic.
The values in these regions are quantitatively con-
sistent with those of the observations. The overall
geographical pattern is also similar to that of the
observations, though it is available only over the
ocean. However, the model overestimates the CLW
amount in the tropical western North Pacific and
the o¤-equator area of the Southern Pacific. These
40 Journal of the Meteorological Society of Japan Vol. 90A

Fig. 8. SAT distribution for (a) JJA and (b) DJF simulated in the historical experiment (1979 through 2005 average), and (c, d) di¤erences
from the observation (CRU TS3.1) over land. Units are C.
February 2012 S. YUKIMOTO et al. 41

Fig. 9. Precipitation distribution for (a) JJA and (b) DJF simulated in the historical experiment (1979 through 2005 average), and (c, d)
di¤erences from the CMAP observation. Units are mm day1.
42 Journal of the Meteorological Society of Japan Vol. 90A

Fig. 10. Snow coverage climatological distributions (1979 through 2005 average; in percent; data obtained from NSIDC) observed in
the NH in (a) fall (SON), (b) winter (DJF ), (c) spring (MAM), and (d) summer (JJA). (e–h) same as (a–d) but for simulated dis-
tributions in the historical experiment.
February 2012 S. YUKIMOTO et al. 43

biases indicate distributions similar to those of
the biases in precipitation (Fig. 9), implying over-
estimated convections with significant cloud droplet
detrainment, although there is less CLW along the
ITCZ in the middle-eastern Pacific. In the Southern
Ocean, however, the model generally underesti-
mates the CLW, which is consistent with underesti-
mated reflected SW radiation and overestimated
OLR at TOA in the region (Figs. 7c, d).
Figure 12 compares the simulated aerosol optical
depth (AOD) at 550 nm with the satellite-retrieved
AOD with the MODerate resolution Imaging Spec-
trometer (MODIS) (e.g., Remer et al. 2008). The
spatial pattern of the simulated AOD is consistent
with that of the satellite retrievals. Large AOD val-
ues are found over a dust belt (Prospero et al. 2002)
that extends from the west coast of North Africa
over the Middle East, Central Asia, and South
Asia to East Asia. The simulated AOD is, however,
systemically less than the satellite retrieved values
over both the oceans and land. The globally aver-
aged simulated AOD (0.6) is within the range but
at the lower end of the values among the aerosol
models in AeroCom (Kinne et al. 2006). Compari-
sons with satellite retrievals suggest that the simula-
tion underestimates the biomass burning aerosol in
the tropics. The negative bias is also possibly due
to the treatment of the dependence of AOD on
hygroscopicity, especially sea-salt aerosol, or may
be partly the result of overestimating the satellite
retrievals of sea-salt, due to possible cloud contam-
ination (Kaufman et al. 2005).
Figure 13 displays mean sea-level pressure (SLP)
distributions simulated for JJA and DJF, with dif-
Fig. 11. Annual mean column-integrated
cloud liquid water (CLW) (in kg m2) for
(a) SSM/I observations (the Defense Mete-
orological Satellite Program) and (b) simu-
lation in the historical experiment. Each
climatology is for 1988 through 2005. The
SSM/I data for January 1988 to November
1991 is from the F8 satellite, that for De-
cember 1991 to April 1997 is from the F11
satellite, and that for May 1997 to Decem-
ber 2005 is from the F14 satellite.
Fig. 12. Annually averaged global distribu-
tions of the aerosol optical depth (AOD)
from 2001 through 2005, obtained from
(a) simulation (historical experiment), and
(b) MODIS-Terra 5.1 observation.
44 Journal of the Meteorological Society of Japan Vol. 90A

ferences from JRA-25. The Pacific and Atlantic
subtropical highs in the boreal summer (JJA) are
appropriately simulated in their location and cen-
tral pressure. In the boreal winter (DJF), the major
climatological cyclones of the Aleutian Low and
the Icelandic Low are satisfactorily simulated,
though they are stronger than the reanalysis with a
negative bias southeast of the climatological center.
These low SLP biases imply a tendency of stronger
subpolar gyres in the North Pacific and the North
Atlantic. There are relatively weaker SLP biases in
the SH in both seasons, despite the systematic radi-
ation bias, except for a slightly stronger meander-
ing south of Australia in winter.
We now examine the simulated atmospheric
meridional-vertical structure. Figure 14 depicts the
simulated zonally averaged temperature and zonal
wind for the seasonal mean (JJA and DJF ). The
overall temperature structure agrees well with the
reanalysis for both seasons. In particular, biases in
the tropics are su‰ciently small, with di¤erences of
less than 1 K. The tropopause height is accurately
simulated, though the tropopause temperature is
lower in JJA and higher in DJF by 2 to 3 K com-
pared with the reanalysis. A cold bias with a baro-
tropic structure is outstanding at 40N in summer.
In contrast, an overall cold bias of 2 to 3 K is dom-
inant in winter mid-high latitudes in the NH, with
a strong cold bias (>5 K) near the surface at high
latitude.
The zonal wind structure is also well-simulated
for both seasons. Each subtropical jet is accurately
simulated in position and strength in JJA and DJF.
Separation between the subtropical jet and the stra-
tospheric polar night jet is also properly simulated,
implying a realistic vertical propagation of plane-
Fig. 13. Mean sea-level pressure for (a) JJA and (b) DJF simulated in the historical experiment (contour,
c.i. ¼4 hPa) and di¤erences from the JRA-25 reanalysis (shading). The di¤erences are masked in the
high-altitude region where the climatological surface pressure is less than 850 hPa. Each climatology is for
1979 through 2005 average.
February 2012 S. YUKIMOTO et al. 45

Fig. 14. Zonally averaged atmospheric temperature (unit: K) for (a) JJA and (b) DJF, and zonal wind (unit: m s1) for (c) JJA and
(d) DJF. Black and dark green contours denote simulated and JRA-25 reanalysis values, and shading denotes simulated biases rel-
ative to the JRA-25 reanalysis. Each climatology is for the 1979 through 2005 average.
46 Journal of the Meteorological Society of Japan Vol. 90A

tary waves and divergence. Associated with the
temperature bias in the NH there is a weak easterly
bias in the JJA equatorial side of the subtropical jet
axis.
SSS can be an important measure for evaluating
the water cycle simulated in the climate model,
since it is strongly a¤ected by precipitation minus
evaporation, river runo¤, and sea ice formation/
melting. Figure 15 illustrates the simulated annual-
mean SSS for the 1979 through 2005 averages,
along with a comparison with the observations for
the same period (Ishii et al. 2006). The basin-scale
patterns (e.g., the saline regions in the subtropical
oceans, the higher SSS in the Atlantic than in the
Pacific, and the low SSS in the Arctic and around
the Maritime Continent) are realistic. In most ocean
areas, the SSS is close to that observed, with biases
of less than 1 psu. A remarkable low SSS bias,
however, is found in the region around the North
Atlantic Current (NAC), particularly east of New-
foundland. This SSS bias can be attributed to a
weaker northward transport of saline water by the
NAC and implies the possibility of suppressed con-
vection, leading to weaker thermohaline circulation
or MOC (15 Sv) in the North Atlantic. Some sig-
nificant biases in limited regions in the Bay of
Bengal (positive bias) and in the tropical South Pa-
cific (negative bias) are consistent with precipitation
biases (Fig. 9).
Zonal mean ocean temperature and salinity are
presented in Fig. 16a, along with their biases from
the Polar science center Hydrographic Climatology
(PHC) version 3 (Steele et al. 2001). The overall
structure for both temperature and salinity is rea-
sonably simulated. A warm bias of more than 2C
north of 60N suggests that the GIN Seas (not
shown) are warmly biased beneath the subsurface,
in contrast with the cold surface bias. Another
warm bias of more than 1C in the abyss is at-
tributed to a warmly biased AABW. Similar but
weaker warm biases are also found in a long-term
integration of MRI.COM3 by using CORE-2 inter-
annual forcing (Tsujino et al. 2011). Intrusions of
saline water into the Weddell and Ross Seas are
not su‰cient in the present simulation, compared
to the CORE-2 simulation.
Temperature biases above 1000 m mainly reflect
North Pacific characteristics (not shown). The ther-
mocline sinks (rises) at low (mid) latitudes, proba-
bly because the subtropical gyre is biased toward
the equator. Large salinity biases seem to be con-
fined to upper layers. Fresh (saline) biases are
found at high (low) latitudes in the Atlantic, while
a fresh (saline) bias is found in the South (North)
Pacific. We assume that these biases are attribut-
able mainly to the biased ocean current for the At-
lantic (e.g., insu‰cient northward transport by the
NAC) and to the biases in precipitation pattern for
the Pacific, which are also consistent with the above
suggestion for the SSS biases.
Meridional overturning of each basin (i.e., the
North Atlantic, Pacific, and Indian Oceans) is pre-
sented in Fig. 16b. Estimates of MOCs based on
the inverse method are 16 G2Svat48
N for the
North Atlantic, 21 G6 Sv for the Southern Ocean,
and 18 G6 Sv for the deep Indo-Pacific Ocean
(Ganachaud 2003). These values are consistent
with the simulated MOCs. In the Southern Ocean,
however, part of the AABW formation process is
absent as noted above, resulting in a northward-
shifted meridional overturning cell around Antarc-
tica (Fig. 16b), consistent with a warmer AABW
than that in the CORE-2 simulation (Tsujino et al.
2011).
Sea ice interacts with the ocean currents due to
influences on freshwater transport as well as strong
influences on the local climate with well-known ice-
albedo feedback. Therefore, realistically reproduc-
ing the sea ice distribution is a crucial element in
climate modeling. Figure 17 depicts the simulated
sea ice coverage (concentration or compactness) in
September and March for the NH and the SH. The
observed sea ice extent (region where the mean sea
ice concentration exceeds 15%) from HadISST1.1
(Rayner et al. 2003) is also presented for com-
parison.
The geographical distribution in March indicates
good simulation in the North Pacific sector, but
an excessive extent in the Labrador Sea through
the GIN and Barents Seas. This unrealistic sea
ice expansion in the North Atlantic is associated
with the fresher sea surface (Fig. 17b) and the
possibly weaker and southward-shifted NAC, and
they possibly interact each other. The stronger
Icelandic Low (Fig. 13b) is possibly somewhat re-
lated to this sea ice bias; however, the influence
seems relatively small, based on a brief examination
of the relationship with the interannual variation of
the Icelandic Low (not shown). In the Antarctic,
the sea ice extent is generally well-simulated in
both seasons.
The seasonal cycle of the sea ice area (sum of
the ocean area multiplied by the sea ice concentra-
tion) for the NH and the SH is also plotted in
February 2012 S. YUKIMOTO et al. 47

Fig. 15. Annual mean sea surface salinity (SSS) (unit: psu) (a) observed and (b) simulated in the historical
experiment, and (c) its bias (simulation–observation), for the 1979 through 2005 average.
48 Journal of the Meteorological Society of Japan Vol. 90A

Fig. 16. (a) Zonally averaged annual mean global ocean (left) temperature [unit: C] and (right) salinity
[unit: psu] (contours) and biases (color shading) relative to the observation (PHC; Steele et al. 2001). (b)
Meridional overturning stream function [units: Sv] in (left) the Southern Ocean, (center) the Atlantic
Ocean, and (right) the Indian plus Pacific Oceans.
February 2012 S. YUKIMOTO et al. 49

Fig. 17. Simulated sea ice concentration distributions in September for (a) NH and (b) SH, and in March for
(c) NH and (d) SH. Red contours denote 15% lines of the observed concentration (HadISST ). Seasonal
cycle of simulated (black) and observed (red) sea ice area for (e) NH and (f ) SH. Each climatology is for
the 1979 through 2005 average.
50 Journal of the Meteorological Society of Japan Vol. 90A

Fig. 17. The simulated sea ice area in the NH ex-
hibits a 19 106km2maximum in March and a
5106km2minimum in August. The months of
maximum and minimum are close to those ob-
served; however, the model simulates too large a
sea ice area in winter through spring. In the SH,
despite the well-simulated distributions, the sea-
sonal variation of sea ice area has an overestimated
18 106km2maximum area in September, in
comparison with the observed maximum area of
15 106km2. This result suggests that the Antarc-
tic sea ice compactness is higher than indicated by
observation.
6.2 Variability
In projecting future climate change, projecting
how the variabilities, including El Nin˜ o and South-
ern Oscillation (ENSO) change, is becoming as im-
portant as, or more important than, projecting the
mean climate. Improving the reliability of the pro-
jected change in variabilities requires realistic simu-
lation of the variability in the present-day climate.
Here, we present SST variations as representative
ocean variabilities, including ENSO, the Pacific
decadal oscillation (PDO) (Mantua et al. 1997),
and the Atlantic multi-decadal oscillation (AMO)
(Kerr 2000). In addition, we present the most
dominant atmospheric variability, known as the
Arctic Oscillation (AO) or northern annular mode
(NAM), and the Antarctic oscillation (AAO) or
southern annular mode (SAM). To examine these
variabilities, we use the results from one member
in the ensemble historical experiment. Other mem-
bers have similar results.
Figure 18 presents the standard deviation of the
monthly SST anomalies observed and simulated
by the historical experiment. We evaluated the
SST anomalies for 116 recent years (1890 through
2005) with respect to the 116-year climatology
without detrending, in order to capture the inter-
annual, decadal, and multi-decadal variations. A
distinct variation in the equatorial Pacific suggests
SST variation associated with ENSO. Relatively
larger variation is found around the Kuroshio Ex-
tension, from which the variation seems to spread
around the subtropical gyre in a horseshoe-like pat-
tern. Decadal variation in this region is related to
PDO, which is an important variability in decadal
prediction for the near future (Mochizuki et al.
2010). In the North Atlantic, however, an unrealis-
tically strong variation meanders along the simu-
lated NAC. The simulated NAC shifts southward
in the east o¤ Newfoundland. This problem will be
discussed later.
We examine the simulated variations associated
with ENSO (Fig. 19) for 1890 through 2005. The
simulated time series of the NINO3 region (90W–
150W, 5S–5N) exhibits interannual variation
with a standard deviation of 0.65C, which is 14%
smaller than the observation (0.76C) for the same
period. The oscillation period seems similar to the
observed one, and there is no unrealistically con-
centrated dominant period (e.g., biennial oscillation
was rather dominant in MRI-CGCM2.3.2). The
observed NINO3 SST indicates an apparent posi-
tively skewed variation in recent several decades,
while the simulated one does not. Stronger ENSO
variability tends to induce a stronger El Nin˜ o, due
to its non-linear e¤ects (Yukimoto and Kitamura
2003).
The SST anomaly with regression on the NINO3
SST anomaly is presented in Fig. 19b. A warm
anomaly in the equatorial Pacific extends from o¤
Peru, and a cold anomaly is observed in the mid-
latitude North Pacific. This pattern is consistent
with the observation (Fig. 19e), though the simu-
lated signal is generally weaker than the observed
one.
Associated with El Nin˜ o, the simulated precipi-
tation increases in the equatorial Pacific and de-
creases around it, with a relatively strong signal in
the Maritime Continent. This overall characteristic
roughly agrees with the observation. The observed
precipitation increase with El Nin˜ o is located in
the central Pacific along the equator (Fig. 19f );
however, the simulated one shifts westward. The
model indicates a much weaker east-west contrast
of the SLP anomaly (i.e., Southern Oscillation) as-
sociated with El Nin˜o. The negative SLP anomaly
simulated in the northern North Pacific is consis-
tent with the observation.
The simulated and observed SLP and SAT
anomalies are regressed on the AO index that is
the first EOF mode calculated for the month-to-
month SLP in November through March in 20N
to 90N. The observed (JRA-25) SLP anomaly
pattern (Fig. 20a) associated with AO indicates a
low-pressure anomaly in the polar region, and sur-
rounding high-pressure anomalies in the mid-
latitude. The simulated SLP anomaly pattern (Fig.
20b), however, only roughly agrees with the ob-
served pattern, indicating a zonally symmetric low-
pressure anomaly in the Arctic. The minimum of
the low-pressure anomaly is simulated around the
February 2012 S. YUKIMOTO et al. 51

North Pole, while it is observed around Iceland.
The observed high-pressure anomalies in the Atlan-
tic and the Pacific are simulated faint and shifted
westward. The simulated AO accounts for more
variance (29.2%) than the observation (20.4%).
Miller et al. (2006) suggested that the AO in models
tend to overestimate the variance.
In accordance with the SLP anomaly, warm SAT
anomalies over the Eurasian continent from north-
ern Europe through northern Japan, and colder
SAT anomalies in northeastern Canada through
Greenland, the Labrador Sea, and southern Alaska
are observed. This overall feature associated with
AO in MRI-CGCM3 is roughly consistent with
the observations.
The Antarctic oscillation, the counterpart of
AO in the SH, is also examined. Its index is the
first mode of the month-to-month 700 hPa geo-
potential height at 90Sto20
S. The plots for
AAO, which are similar to those for AO, are
depicted in Fig. 21. This figure indicates a low-
pressure anomaly with a center of action around
the Pacific coast of Western Antarctica, and sur-
rounding high-pressure anomalies with a tri-polar
pattern in the mid-latitude. In addition, it indi-
cates cold anomalies in the Pacific and Indian
Ocean sectors, and a warm anomaly around the
Antarctic Peninsula. These characteristics are con-
sistent with the observations. The simulated AAO
also overestimates the fraction of the accounted
Fig. 18. Standard deviations of (a) observed and (b) simulated monthly SST anomalies for 1890 through
2005. The anomalies are deviations from the 1890 through 2005 climatology without detrending. Units
are C.
52 Journal of the Meteorological Society of Japan Vol. 90A

variance (31.8%) compared to the observed one
(24.7%).
7. Summary and discussion
A new climate model, MRI-CGCM3, has been
developed as a subset of the new earth system model
MRI-ESM1 at MRI. A set of CMIP5 experiments
is performed using MRI-CGCM3. As a base-line
for detailed analysis of the results of various experi-
ments, we describe formulations of the model and
evaluate the basic performance of MRI-CGCM3
from the results of a set of basic experiments.
Fig. 19. Simulated (left column) and observed (right column) variations associated with ENSO. (a) Time se-
ries of SST anomaly in the NINO3 region (90W to 150W, 5Sto5
N), and anomalies of (b) SST (C)
and (c) precipitation (mm day1; color shading) and SLP (hPa; contours) regressed on the NINO3 SST
anomaly. The regressions are calculated over 116 years (1890 through 2005) for SST and 27 years (1979
through 2005) for precipitation and SLP. (d–f) Same as (a–c) but for the observations (the COBE SST,
the JRA-25 SLP, and the CMAP precipitation).
February 2012 S. YUKIMOTO et al. 53

Fig. 20. Anomaly patterns of mean sea-level pressure (SLP) (hPa; contours) and SAT (K; color shading) as-
sociated with the normalized AO index (1979 through 2005) for (a) JRA-25 reanalysis and (b) simulation in
the historical experiment.
Fig. 21. Same as Fig. 20 but for AAO.
54 Journal of the Meteorological Society of Japan Vol. 90A

To develop MRI-CGCM3, the AGCM of the
former climate model MRI-CGCM2 series has
been substantially upgraded. The upgrade includes
the dynamics frame, and physical parameteriza-
tions of cumulus convection, radiation, clouds,
PBL, land surface, and ocean surface processes.
The ocean-ice model is also a new model MRI
.COM3, which is a fundamental replacement of
the former ocean component in the MRI-CGCM2
series. In MRI-CGCM3, the AGCM is coupled in-
teractively with a new version of aerosol model
MASINGAR mk-2, which enables the model to ex-
plicitly represent the direct and indirect e¤ects of
aerosols on the climate system.
In the piControl experiment, the model exhibits
very stable behavior without climatic drifts, at least
in the radiation budget and the temperature near
the surface. Some of major indices associated with
ocean circulation (i.e., MOCs and ACC) are also
found to be stable with fairly realistic values.
A small net radiation imbalance at TOA of
0.5 W m2exists in the piControl experiment, and
an unknown energy source of 0.5 W m2in the at-
mosphere results in an increasing trend of the aver-
aged temperature of the global ocean. Since the
trend in SAT is su‰ciently small (0.016 K/100yr),
it can be neglected in analyses of at least centennial
timescale climate change. The global water is con-
served with a high precision of less than 2 mm/
100yr of the sea level trend.
Climate sensitivity is estimated with the Gregory
method (Gregory et al. 2004) and found to be
2.11 K, which is rather low compared with MRI-
CGCM2.3.2. The global SAT increase for the
1pctCO2 experiment is also less than that in MRI-
CGCM2.3.2, but exhibits relatively closer values
to MRI-CGCM2.3.2 compared to the ratio of cli-
mate sensitivities, implying a possible di¤erence in
ocean heat up-take. In the historical experiment
for reproducing climate change of 1850 through
2005 with all the forcing agents including concen-
tration of GHGs and emission of aerosols, the
global SAT increase during this period is underesti-
mated by 0.17C compared with the observation.
The contribution of cloud radiative forcing to cli-
mate sensitivity is also evaluated with a similar
regression method (Fig. 4), and the results indicate
almost neutral SW cloud feedback and negative
LW cloud feedback (0:38 W m2K1) to the
GHG forcing. The LW cloud feedback has a sig-
nificantly greater negative value than those for the
models in Gregory and Webb (2008). Further de-
tailed analysis of this subject will be required in fu-
ture studies.
For comparison with the recent observations and
reanalysis, the mean climate simulated by MRI-
CGCM3 is evaluated for its ability to reproduce
the present-day climate (1979 through 2005 aver-
age) of the historical experiment.
The simulated global-mean radiation at TOA
agrees well with the new satellite observations for
SW and LW. The cloud forcing is also realistic in
the global average. However, biases in the meri-
dional distribution of the radiation budget lead to
underestimation of 1 PW in the implied oceanic
southward heat transport in the SH, although the
atmospheric meridional heat transport generally
agrees with the observational estimations.
The global mean and geographical distribution
of the simulated SST indicate overall proper repro-
duction of the global-scale climate, even without
flux adjustment. The SST indicates small biases in
the tropics but contrasting biases in the mid-high
latitudes with cold NH and warm SH.
The distribution of radiation, which is strongly
a¤ected by the distribution of clouds, indicates
small biases in the large-scale pattern, except for re-
gions with errors in tropical convection and the
Southern Ocean. The basic mean fields that regu-
late the atmospheric general circulation (including
precipitation, SLP, and meridional-vertical struc-
ture) are evaluated by comparison with the obser-
vation and JRA-25 reanalysis. Consequently, the
model demonstrates generally reasonable reproduc-
tion of these mean fields, except for some important
issues in precipitation (i.e., a lack of Indian Mon-
soon precipitation and the double ITCZ in the Pa-
cific). These basic performances and features are
also consistent with the simulated CLW content.
The simulated ocean temperature and salinity
structures and important circulations are found to
be reasonable by considering the atmospheric and
surface states. Sea ice distribution generally agrees
with the observations, except for the excessive ex-
tension in the winter Atlantic and overestimated
Antarctic sea ice compactness.
For variability, the model simulates comparable
SST variation, including ENSO, with the observa-
tion for 1890 through 2005. The model suggests re-
alistic SST variability, including ENSO and PDO.
However, an unrealistic SST variation in the North
Atlantic seems to be associated with biased ocean
current and sea ice distribution. The dominant at-
mospheric variability for AO and AAO is found
February 2012 S. YUKIMOTO et al. 55

to be realistic for AAO in the model. The simu-
lated SLP pattern with AO is not so realistic partic-
ularly over the mid-latitude oceans (Fig. 20), which
means discussing the decadal variability of AO,
that would involve interactions with the oceans,
may not be suitable for this model.
In many respects, MRI-CGCM3 is capable of
reproducing the basic mean state and variability in
the climate system for investigating global and sub-
continental scale variability and climate change.
However, special attention should be paid to the
following issues, which need improvement.
There are systematic biases in the simulated SST.
The cold bias in the North Pacific concentrates
along the Kuroshio Extension and its return cur-
rent in the subtropical gyre (Fig. 6c). The region
downstream of the strong western boundary cur-
rent of the Kuroshio is dominated by eddies. One
possible reason for this cold bias is insu‰cient
northward eddy heat transport, since many climate
models with a low-resolution (non-eddy-resolving)
ocean model tend to indicate a similar bias (e.g.,
Delworth et al. 2006; Watanabe et al. 2011). More
sophisticated treatment, including parameterization
of the oceanic eddies, may be required.
Another possible factor is overestimation of the
subpolar gyre, which is attributable to the simu-
lated Aleutian Low (Fig. 13b) being stronger than
that observed. The stronger subpolar gyre in the
North Pacific possibly leads to a stronger south-
ward cold advection east of Japan, which is one
of the factors of the cold bias along the Kuroshio
Extension. As a result of cooling in the Kuroshio
Extension, the region of return current in the sub-
tropical gyre becomes colder.
The cold SST bias in the northern part of the
subtropical gyre results in overestimated low clouds
in summer and subsequent underestimation of solar
radiation at the sea surface (Fig. 7c) and decreased
SST, yielding a positive feedback for a colder SST.
Cold SST biases also dominate in the North At-
lantic (Fig. 6c), with a particularly large bias in the
Labrador and GIN Seas through the Barents Sea.
This region is along the NAC, which transports
warm and saline water. Possibly, the lack of trans-
port tends to result in a larger sea ice extent in this
region in winter (Fig. 17c). The overestimated sea
ice extent again results in a decreased SST with
ice-albedo feedback. Furthermore, once the sea ice
expands, a very stable layer with low SSS (Fig. 15)
is formed near the surface, due to melt water of the
sea ice in spring through summer. This very stable
sea surface prevents the deep convective mixing
that is active in the region, including the Labrador
Sea (e.g., Pickart et al. 2002), which again creates a
favorable situation for sea ice formation. The
model seems to simulate unrealistically large fluctu-
ations along the NAC (Fig. 18b). This unstable
fluctuation of the current is possibly a trigger for
the expansion of sea ice and entering the feedback
loop mentioned above.
The present MRI-CGCM3 ocean model has low
resolution and cannot resolve oceanic baroclinic
eddies in the mid-high latitudes. Improvements in
the parameterization of the ocean model will be re-
quired to simulate a realistic transport of heat and
salt by the eddy e¤ects in these important regions.
Also, parameters, including oceanic eddy viscosity,
must be set carefully so as to avoid unrealistic ed-
dies or meandering. This is an important problem
to be solved.
Warm SST biases dominate in the eastern basins
in the subtropical SH, where low-level clouds must
be realistically represented. The warm SST bias ex-
tends broadly in the Southern Ocean, partly due to
overestimated solar absorption (Fig. 7c). Results
from a separate test run with a cloud satellite simu-
lator (not shown) suggest underestimation of low-
level clouds, in addition to underestimation of col-
umn CLW content (Fig. 11b) in the mid-latitude
SH. The atmospheric circulation in the SH is fairly
realistic in the mean fields (Figs. 13 and 14), thus
supporting realistic heat transport (Fig. 5) by the
atmosphere, including by synoptic disturbances.
Reflected solar radiation is also underestimated in
atmosphere-only experiments. Tuning parameters
of the PBL scheme to increase low-level clouds did
not resolve the problem since the SW bias worsened
in the NH when parameters were adjusted for bet-
ter estimates in the SH. Once the SST increases,
the low-level cloud decreases in the region; as a re-
sult, the SST increases further due to excessive solar
heating, forming a positive feedback loop.
The double ITCZ is another serious problem in
the simulation by MRI-CGCM3. Less precipitation
is simulated on the equator in the western Pacific,
and excessive precipitation is simulated at the
ITCZ in the winter hemisphere. Many models have
su¤ered from the double ITCZ problem (e.g., Lin
2007), although some models indicate improvement
(e.g., Watanabe et al. 2011). We also have made
some progress in addressing this problem by intro-
ducing new cumulus parameterization with some
tuning. Experiments with the atmosphere-only
56 Journal of the Meteorological Society of Japan Vol. 90A

model (not shown) yield much more precipita-
tion on the equator in the western Pacific and a
diminished double ITCZ feature. However, tuning
further decreases the Indian summer monsoon pre-
cipitation, which may be related to the excess of
precipitation in the western Pacific ITCZ.
For the ocean-coupled MRI-CGCM3, we ad-
justed cloud forcing to agree with the observation
by tuning cloud water detrainment from the cumu-
lus convection. The model simulates fairly well the
precipitation pattern in the boreal summer, al-
though it generally overestimates precipitation in
the tropics. In the austral summer, however, the
precipitation pattern still forms a double ITCZ,
since the precipitation is simulated along the south-
ern edge of the cold tongue, not in the South Pacific
Convergence Zone (SPCZ) as observed. A possible
cause of this precipitation error is that the simu-
lated cold tongue is stronger than the observed one
(Fig. 6b). The double ITCZ tends to produce a
stronger easterly wind along the equator, which fur-
ther enhances the cold tongue, again forming a pos-
itive feedback loop (Lin 2007).
The exaggerated cold tongue may a¤ect the sim-
ulation of ENSO. The lack of precipitation in the
equatorial central Pacific could cause the weak El
Nin˜ o by reducing anomalous Bjerknes feedback in
the El Nin˜ o, as implied from the weaker zonal pres-
sure anomaly in the simulated El Nin˜o (Fig. 19c).
MRI-CGCM3’s present ocean resolution
(10:5) is probably insu‰cient to resolve equa-
torial instability waves, which may contribute to
the stronger cold tongue. Some special treatment
will be required to resolve this problem, which is
also an important subject for future study.
Acknowledgments
The atmospheric model development is based
upon JMA’s operational weather prediction model,
which is the result of enormous e¤orts by numer-
ous personnel at JMA, especially in the Numerical
Prediction Division. Other component models are
based on the results of research e¤orts by the per-
sonnel at MRI, especially in the Climate Research
Department, Oceanographic Research Department,
and Atmospheric Environment and Applied Mete-
orology Research Department. This work was con-
ducted under the framework of the ‘‘Comprehen-
sive Projection of Climate Change around Japan
due to Global Warming’’ supported by JMA, and
partly supported by the KAKUSHIN Program of
the Ministry of Education, Culture, Sports, Science,
and Technology (MEXT). The calculations were
performed on the HITACHI SR16000 located at
MRI.
References
Abdul-Razzak, H., and S. J. Ghan, 2000: A parameter-
ization of aerosol activation: 2. Multiple aerosol
type. J. Geophys. Res.,105, 6837–6844.
Abdul-Razzak, H., and S. J. Ghan, 2002: A parameter-
ization of aerosol activation 3. Sectional repre-
sentation. J. Geophys. Res.,107 (D3), 4026,
doi:10.1029/2001JD000483.
Albrecht, B. A., 1989: Aerosols, cloud microphysics, and
fractional cloudiness. Science,245, 1227–1230.
Andres, R. J., and A. D. Kasgnoc, 1998: A time-
averaged inventory of subaerial volcanic sulfur
emissions. J. Geophys. Res.,103, 25251–25261.
Aoki, Te., and Y. T. Tanaka, 2008: Influence of atmo-
spheric aerosol deposition on snow albedo. Tenki,
55, 538–547 (in Japanese).
Table A1. Comparison of configuration for the MRI’s climate models.
Model
MRI-CGCM3 (this work,
Yukimoto et al. 2011)
MRI-CGCM2.3.2
(Yukimoto et al. 2006)
MRI-AGCM3.2S
(Mizuta et al. 2011)
advection scheme Semi-Lagrangian Euler Semi-Lagrangian
Horizontal resolution TL159 (A120 km) T42 (A280 km) TL959 (A20 km)
Vertical levels (top) 48 (top: 0.01 hPa) 30 (top: 0.4 hPa) 64 (top: 0.01 hPa)
Radiation Modified JMA (2007) Shibata and Uchiyama (1992),
Shibata and Aoki (1989)
JMA (2007)
Cumulus convection Yoshimura Prognostic Arakawa –Schubert Yoshimura
Cloud MRI-TMBC Functions of relative humidity Tiedtke (1993)
PBL MYNN (level 2) Mellor Yamada (level 2) Mellor Yamada (level 2)
Land surface HAL Hirai et al. (2007) Hirai et al. (2007)
Gravity wave drag Iwasaki et al. (1989) Iwasaki et al. (1989) Iwasaki et al. (1989)
River and lakes GRiveT simple river routing
(Yukimoto et al. 2000)
N/A
February 2012 S. YUKIMOTO et al. 57

Aoki, Te., K. Kikuchi, M. Niwano, Y. Kodama, M.
Hosaka, and T. Y. Takana, 2011: Physically based
snow albedo model for calculating broadband al-
bedos and the solar heating profile in snowpack
for GCMs., Geophys. Res., submitted
Arakawa, A., and W. H. Schubert, 1974: Interaction of a
cumulus cloud ensemble with the large-scale envi-
ronment. Part I. J. Atmos. Sci.,31, 674–701.
Armstrong, R. L., and M. J. Brodzik, 2005: Northern
Hemisphere EASE-Grid Weekly Snow Cover and
Sea Ice Extent Version 3. Boulder, Colorado
USA: National Snow and Ice Data Center. Digital
media.
Barkstrom, B., E. Harrison, G. Smith, R. Green, J. Ki-
bler, R. Cess, and the ERBE Science Team, 1989:
Earth Radiation Budget Experiment (ERBE) ar-
chival and April 1985 results. Bull. Amer. Meteor.
Soc.,70, 1254–1262.
Bigg, E. K., 1953: The supercooling of water. Proc. Phys.
Soc. London,B66, 688–694.
Bluth, G. J. S., W. I. Rose, I. E. Sprod, and A. J.
Krueger, 1997: Stratospheric loading of sulfur
from explosive volcanic eruptions. J. Geology,
105, 671–684.
Bony, S., and J. L. Dufresne, 2005: Marine boundary
layer clouds at the heart of tropical cloud feedback
uncertainties in climate models. Geophys. Res.
Lett.,32, L20806, doi:10.1029/2005GL023851.
Brohan, P., J. J. Kennedy, I. Harris, S. F. B. Tett, and
P. D. Jones, 2006: Uncertainty estimates in re-
gional and global observed temperature changes:
a new dataset from 1850. J. Geophys. Res.,111,
D12106, doi:10.1029/2005JD006548.
Castro, S. L., G. A. Wick, and W. J. Emery, 2003: Fur-
ther refinements to models for the bulk-skin sea
surface temperature di¤erence. J. Geophys. Res.,
108, 3377, doi:10.1029/2002JC001641.
Chin, M., P. Ginoux, S. Kinne, O. Torres, B. N. Holben,
B. N. Duncan, R. V. Martin, J. A. Logan, A. Hi-
gurashi, and T. Nakajima, 2002: Tropospheric
aerosol optical thickness from the GOCART
model and comparisons with satellite and sun pho-
tometer measurements. J. Atmos. Sci.,59, 461–
483.
Cooke, W. F., C. Liousse, H. Cachier, and J. Feichter,
1999: Construction of a 11fossil fuel emission
data set for carbonaceous aerosol and implementa-
tion and radiative impact in the ECHAM4 model.
J. Geophys. Res.,104, 22137–22162.
Cox, M. D., 1987: Isopycnal di¤usion in a z-coordinate
ocean model. Ocean Modelling,74, 1–5.
Cunningham, S. A., S. G. Alderson, B. A. King, and
M. A. Brandon, 2003: Transport and variability of
the Antarctic Circumpolar Current in Drake Pas-
sage. J. Geophys. Res.,108, 8084, doi:10.1029/
2001JC001147.
Delworth, T. L., A. J. Broccoli, A. Rosati, R. J. Stou¤er,
V. Balaji, J. A. Beesley, W. F. Cooke, K. W.
Dixon, J. Dunne, K. A. Dunne, J. W. Durachta,
K. L. Findell, P. Ginoux, A. Gnanadesikan, C. T.
Gordon, S. M. Gri‰es, R. Gudgel, M. J. Harrison,
I. M. Held, R. S. Hemler, L. W. Horowitz, S. A.
Klein, T. R. Knutson, P. J. Kushner, A. R. Lan-
genhorst, H.-C. Lee, S.-J. Lin, J. Lu, S. L. Maly-
shev, P. C. D. Milly, V. Ramaswamy, J. Russell,
M. D. Schwarzkopf, E. Shevliakova, J. J. Sirutis,
M. J. Spelman, W. F. Stern, M. Winton, A. T.
Wittenberg, B. Wyman, F. Zeng, and R. Zhang,
2006: GFDL’s CM2 global coupled climate mod-
els. Part I: Formulation and simulation character-
istics. J. Climate,19, 643–674.
Ebert, E. E., and J. A. Curry, 1992: A parameterization
of ice cloud optical properties for climate models,
J. Geophys. Res.,97, 3831–3836.
European Centre for Medium-Range Weather Forecasts,
2004: IFS Documentation CY28r1, http://www
.ecmwf.int/research/ifsdocs/CY28r1. Part IV, Chap-
ter 6.
Fe
´can, F., B. Marticorena, and G. Bergametti, 1999:
Parameterization of the increase of the aeolian
erosion threshold wind friction velocity due to soil
moisture for arid and semi-arid areas. Annales
Geophysicae,17, 149–157.
Fasullo, J. T., and K. E. Trenberth, 2008: The annual
cycle of the energy budget. Part I: Global mean
and land–ocean exchanges. J. Climate,21, 2297–
2313.
Forster, P., V. Ramaswamy, P. Artaxo, T. Berntsen, R.
Betts, D. W. Fahey, J. Haywood, J. Lean, D. C.
Lowe, G. Myhre, J. Nganga, R. Prinn, G. Raga,
M. Schulz, and R. Van Dorland, 2007: Changes in
Atmospheric Constituents and in Radiative Forc-
ing. In: Climate Change 2007: The Physical Science
Basis. Contribution of Working Group I to the
Fourth Assessment Report of the Intergovernmental
Panel on Climate Change, S. Solomon, D. Qin,
M. Manning, Z. Chen, M. Marquis, K. B. Averyt,
M. Tignor and H. L. Miller, Eds., Cambridge Uni-
versity Press, Cambridge, United Kingdom and
New York, NY, USA.
Ganachaud, A., 2003: Large-scale mass transports, water
mass formation, and di¤usivities estimated from
World Ocean Circulation Experiment (WOCE)
hydrographic data. J. Geophys. Res.,108,
doi:10.1029/2002JC001565.
Geleyn, J.-F., and A. Hollingsworth, 1979: An economi-
cal analytical method for the computation of the
interaction between scattering and line absorption
of radiation. Beitr. Phys. Atmos.,52, 1–16.
Gent, P. R., and J. C. McWilliams, 1990: Isopycnal mix-
ing in ocean circulation models. J. Phys. Ocean-
ogr.,20, 150–155.
58 Journal of the Meteorological Society of Japan Vol. 90A

Gentemann, C., C. J. Donlon, A. Stuart-Menteth, and
F. J. Wentz, 2003: Diurnal signals in satellite sea
surface temperature measurements. Geophys. Res.
Lett.,30(3), 1140, doi:10.1029/2002GL016291.
Giorgi, F., and W. L. Chameides, 1986: Rainout lifetimes
of highly soluble aerosols and gases as inferred
from simulations with a general circulation model.
J. Geophys. Res.,91, 14367–14376.
Gong, S. L., 2003: A parameterization of sea-salt aerosol
source function for sub- and super-micron
particles. Global Biogeochem. Cy.,17, 1097,
doi:10.1029/2003GB002079.
Gong, S. L., L. A. Barrie, and J.-P. Blanchet, 1997: Mod-
eling sea-salt aerosols in the atmosphere 1. model
development. J. Geophys. Res.,102, 3805–3818.
Gregory, D., W. J. Ingram, M. A. Palmer, G. S. Jones,
P. A. Stott, R. B. Thorpe, J. A. Lowe, T. C. Johns,
and K. D. Williams, 2004: A new method for diag-
nosing radiative forcing and climate sensitivity.
Geophys. Res. Lett.,31, L03205, doi:10.1029/
2003GL018747.
Gregory, D., and M. Webb, 2008: Tropospheric ad-
justment induces a cloud component in CO2
forcing. J. Climate,21, 58–71, doi:10.1175/
2007JCLI1834.1.
Gregory, D., R. Kershaw, and P. M. Inness, 1997: Par-
ametrization of momentum transport by convec-
tion. II: Tests in single-column and general circula-
tion models. Quart. J. Roy. Meteor. Soc.,123,
1153–1183.
Guenther, A., C. N. Hewitt, D. Erickson, R. Fall, C.
Geron, T. Graedel, P. Harley, L. Klinger, M. Ler-
dau, W. A. Mckay, T. Pierce, B. Scholes, R. Stein-
brecher, R. Tallamraju, J. Taylor, P. Zimmerman,
1995: A global model of natural volatile organic
compound emissions. J. Geophys. Res.,100(D5),
8873–8892, doi:10.1029/94JD02950.
Hansen, J., and L. Nazarenko, 2004: Soot climate forcing
via snow and ice albedos. Proc. Natl. Acad. Sci.
USA,101, 423–428, doi:10.1073/pnas.2237157100.
Hansen, J., A. Lacis, D. Rind, G. Russell, P. Stone, I.
Fung, R. Ruedy, and J. Lerner, 1984: Climate sen-
sitivity: Analysis of feedback e¤ects. Climate Pro-
cesses and Climate Sensitivity, Geophys. Monogr.,
No. 29, Amer. Geophys. Union, 130–163.
Hansen, J., M. Sato, and R. Ruedy, 1997: Radiative forc-
ing and climate response. J. Geophys. Res.,102
(D6), 6831–6864.
Hara, T., 2007a: Update of the operational JMA meso-
scale model and implementation of improved
Mellor-Yamada level 3 scheme. Extended Ab-
stracts, 22nd Conf. on Weather Analysis and
Forecasting/18th Conf. on Numerical Weather Pre-
diction, Utah, USA, J3.5.
Hara, T., 2007b: Implementation of improved Mellor-
Yamada Level 3 scheme and partial condensation
scheme to JMANHM and their performance.
CAS/JSC WGNE Res. Act. in Atmos. and Ocean
Modelling,37, 0407–0408.
Hasumi, H., 2006: CCSR Ocean Component Model
(COCO) Version 4.0. Center for Climate System
Research, CCSR Report,25, 103 pp.
Hess, M., P. Koepke, and I. Schult, 1998: Optical proper-
ties of aerosols and clouds: The software package
OPAC. Amer. Meteor. Soc.,79, 831–843.
Hirai, M., T. Sakashita, H. Kitagawa, T. Tsuyuki, M.
Hosaka, and M. Oh’izumi, 2007: Development
and validation of a new land surface model for
JMA’s operational global model using the CEOP
observation dataset. J. Meteor. Soc. Japan,85A,
1–24.
Hofmann, M., and M. A. M. Maqueda, 2006: Perfor-
mance of a second-order moments advection
scheme in an Ocean General Circulation Model.
J. Geophys. Res.,111, C05006, doi:10.1029/
2005JC003279.
Hu, Y. X., and K. Stamnes, 1993: An accurate parame-
terization of the radiative properties of water
clouds suitable for use in climate models. J. Cli-
mate,6, 728–742.
Hunke, E. C., and W. H. Lipscomb, 2006: CICE: the Los
Alamos Sea Ice Model Documentation and Soft-
ware User’s Manual, 59 pp, (Available at http://
climate.lanl.gov/source/projects/climate/Models/
CICE/index.shtml).
IPCC, 2001: Climate Change 2001: The Scientific Basis.
Intergovernmental Panel on Climate Change, Con-
tribution of Working Group I to the Third Assess-
ment Report of the Intergovernmental Panel on
Climate Change, Houghton, J. T., Y. Ding, D. J.
Griggs, M. Noguer, P. J. van der Linden, X. Dai,
K. Maskell, and C. A. Johnson, Eds., Cambridge
University Press, Cambridge, United Kingdom
and New York, NY, USA, 881 pp.
Ishii, M., and M. Kimoto, 2009: Reevaluation of histori-
cal ocean heat content variations with time-varying
XBT and MBT depth bias corrections. J. Ocean-
ogr.,65, 287–299, doi: 10.1007/s10872-009-0027-7.
Ishii, M., M. Kimoto, K. Sakamoto, and S.-I. Iwasaki,
2006: Steric sea level changes estimated from his-
torical ocean subsurface temperature and salinity
analyses. J. Oceanogr., 62, 155–170.
Ishii, M., A. Shouji, S. Sugimoto, and T. Matsumoto,
2005: Objective analyses of sea-surface temperature
and marine meteorological variables for the 20th
century using ICOADS and the Kobe Collection.
Int. J. Climatol.,25, 865–879, doi: 10.1002/joc
.1169.
Ishizaki, H., and T. Motoi, 1999: Reevaluation of the
Takano-Onishi scheme for momentum advection
on bottom relief in ocean models. J. Atmos. Oce-
anic. Technol.,16, 1994–2010.
February 2012 S. YUKIMOTO et al. 59

Iwasaki, T., S. Yamada, and K. Tada, 1989: A parame-
terization scheme of orographic gravity wave drag
with the di¤erent vertical partitioning, part 1: Im-
pact on medium range forecasts. J. Meteor. Soc.
Japan,67, 11–41.
Jakob, C., 2001: The representation of cloud cover in At-
mospheric General Circulation Models, PhD the-
sis, Ludwig-Maximilians-Universitaet Muenchen,
193 pp.
Japan Meteorological Agency, 2002: Outline of the oper-
ational numerical weather prediction at the Japan
Meteorological Agency. (Appendix to WMO nu-
merical weather prediction progress report). Japan
Meteorological Agency, 157 pp.
Japan Meteorological Agency, 2007: Outline of the oper-
ational numerical weather prediction at the Japan
Meteorological Agency (Appendix to WMO nu-
merical weather prediction progress report). Japan
Meteorological Agency, 194 pp. (Available online
at http://www.jma.go.jp/jma/jma-eng/jma-center/
nwp/outline-nwp/index.htm).
Ka
¨rcher, B., J. Hendricks, U. Lohmann, 2006: Physically
based parameterization of cirrus cloud formation
for use in global atmospheric models. J. Geophys.
Res.,111, D01205, doi: 10.1029/2005JD006219.
Kaufman, Y. J., O. Boucher, D. Tanre
´, M. Chin, L. A.
Remer, and T. Takemura, 2005: Aerosol anthro-
pogenic component estimated from satellite data.
Geophys. Res. Lett.,32, L17804, doi:10.1029/
2005GL023125.
Kawai, Y., and A. Wada, 2007: Diurnal sea surface
temperature variation and its impact on the atmo-
sphere and ocean: A review. J. Oceanogr.,63, 721–
744.
Kerr, R. A., 2000: A North Atlantic climate pacemaker
for the centuries. Science,288, 1984–1986.
Kettle, A. J., M. O. Andreae, D. Amouroux, T. W. An-
dreae, T. S. Bates, H. Berresheim, H. Bingemer,
R. Boniforti, M. A. J. Curran, G. R. DiTullio, G.
Helas, G. B. Jones, M. D. Keller, R. P. Kiene,
C. Leck, M. Levasseur, G. Malin, M. Maspero, P.
Matrai, A. R. McTaggart, N. Mihalopoulos, B. C.
Nguyen, A. Novo, J. P. Putaud, S. Rapsomanikis,
G. Roberts, G. Schebeske, S. Sharma, R. Sim, R.
Staubes, S. Turner, and G. Uher, 1999: A global
database of sea surface dimethylsulfide (DMS)
measurements and a procedure to predict sea sur-
face DMS as a function of latitude, longitude and
month. Global Biogeochem. Cy.,13, 399–444.
Kinne, S., M. Schulz, C. Textor, S. Guibert, Y. Balkan-
ski, S. E. Bauer, T. Berntsen, T. F. Berglen, O.
Boucher, M. Chin, W. Collins, F. Dentener, T.
Diehl, R. Easter, J. Feichter, D. Fillmore, S.
Ghan, P. Ginoux, S. Gong, A. Grini, J. Hendricks,
M. Herzog, L. Horowitz, I. Isaksen, T. Iversen, A.
Kirkeva
˚g, S. Kloster, D. Koch, J. E. Kristjansson,
M. Krol, A. Lauer, J. F. Lamarque, G. Lesins, X.
Liu, U. Lohmann, V. Montanaro, G. Myhre, J.
Penner, G. Pitari, S. Reddy, O. Seland, P. Stier, T.
Takemura, and X. Tie, 2006: An AeroCom initial
assessment—optical properties in aerosol compo-
nent modules of global models. Atmos. Chem.
Phys.,6, 1815–1834, doi:10.5194/acp-6-1815-2006.
Kiehl, J. T., and C. S. Zender, 1995: A prognostic ice
water scheme for anvil clouds. World Climate Pro-
gramme Research, WCRP-90, WMO/TD-No. 713,
167–188.
Killworth, P. D., D. Stainforth, D. J. Webb, and S. M.
Paterson, 1991: The development of a free-surface
Bryan-Cox-Semtner ocean model. J. Phys. Ocean-
ogr.,21, 1333–1348.
Kitoh, A., 2004: E¤ects of mountain uplift on east asian
summer climate investigated by a coupled atmo-
sphere–ocean GCM. J. Climate,17, 783–802.
Kurihara, K., K. Ishihara, H. Sasaki, Y. Fukuyama, H.
Saitou, I. Takayabu, K. Murazaki, Y. Sato, S.
Yukimoto, and A. Noda, 2005: Projection of cli-
matic change over Japan due to global warming
by high-resolution regional climate model in MRI.
SOLA,1, 97–100.
Large, W. G., and S. Yeager, 2009: The global climatol-
ogy of an interannually varying air-sea flux data
set. Clim. Dyn.,33, 341–364, doi:10.1007/s00382-
008-0441-3.
Lean, J., J. Beer, and R. Bradley, 1995: Reconstruction
of solar irradiance since 1610: Implications for cli-
mate change. Geophys. Res. Lett.,22, 3195–3198.
Le Treut, H., R. Somerville, U. Cubasch, Y. Ding, C.
Mauritzen, A. Mokssit, T. Peterson, and M.
Prather, 2007: Historical Overview of Climate
Change. In: Climate Change 2007: The Physical
Science Basis. Contribution of Working Group I to
the Fourth Assessment Report of the Intergovern-
mental Panel on Climate Change, Solomon, S., D.
Qin, M. Manning, Z. Chen, M. Marquis, K. B.
Averyt, M. Tignor and H. L. Miller, Eds., Cam-
bridge University Press, Cambridge, United King-
dom and New York, NY, USA.
Levitus, S., J. I. Antonov, T. P. Boyer, R. A. Locarnini,
H. E. Garcia, and A. V. Mishonov, 2009: Global
ocean heat content 1955–2008 in light of recently
revealed instrumentation problems. Geophys. Res.
Lett.,36, L07608, doi:10.1029/2008GL037155.
Liss, P. S., and L. Merlivat, 1986: Air-sea exchange rates:
Introduction and synthesis. P. Buat-Me
´nard, Ed.,
The Role of Air-Sea Exchange in Geochemical Cy-
cling, Vol. C 185 of NATO ASI Series, D. Reidel
Publishing Company, 113–127.
Liu, Y., P. H. Daum, and S. S. Yum, 2006: Analytical
expression for the relative dispersion of the cloud
droplet size distribution. Geophys. Res. Lett.,33,
L02810, doi:10.1029/2005GL024052.
60 Journal of the Meteorological Society of Japan Vol. 90A

Large, W. G., and S. G. Yeager, 2009: The global clima-
tology of an interannually varying air-sea flux data
set. Climate Dynamics,33, 341–364, doi:10.1007/
s00382-008-0441-3.
Lin, J.-L., 2007: The double-ITCZ problem in IPCC
AR4 coupled GCMs: Ocean–Atmosphere feed-
back analysis. J. Climate,20, 4497–4525.
Loeb, N. G., B. A. Wielicki, D. R. Doelling, G. L. Smith,
D. F. Keyes, S. Kato, N. Manalo-Smith, and T.
Wong, 2009: Toward optimal closure of the Earth’s
top-of-atmosphere radiation budget. J. Climate,
22, 748–766. doi: 10.1175/2008JCLI2637.1
Lohmann, U., 2002: Possible aerosol e¤ects on ice clouds
via contact nucleation. J. Atmos. Sci.,59, 647–656.
Lohmann, U., and K. Diehl, 2006: Sensitivity studies of
the importance of dust ice nuclei for the indirect
aerosol e¤ect on stratiform mixed-phase clouds. J.
Atmos. Sci.,63, 968–982.
Lohmann, U., P. Stier, C. Hoose, S. Ferrachat, S. Klos-
ter, E. Roeckner, and J. Zhang, 2007: Cloud mi-
crophysics and aerosol indirect e¤ects in the global
climate model ECHAM5-HAM. Atmos. Chem.
Phys.,7, 3425–3446.
Louis, J.-F., 1979: A parametric model of vertical eddy
fluxes in the atmosphere. Boundary Layer Meteor.,
17, 187–202.
Louis, J. F., M. Tiedtke, and J. F. Geleyn, 1982: A short
history of the operational PBL-parameterization at
ECMWF. Workshop on Planetary boundary layer
parameterization, ECMWF, Reading, U.K., 25–27
November 1981, 59–80.
Manabe, S., and R. J. Stou¤er, 1988: Two stable equilib-
ria of a coupled ocean-atmosphere model. J. Cli-
mate,1, 841–866.
Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and
R. C. Francis, 1997: A Pacific interdecadal climate
oscillation with impacts on salmon production.
Bull. Amer. Meteor. Soc.,78, 1069–1079.
Mellor, L. G., and L. Kantha, 1989: An ice-ocean
coupled model. J. Geophys. Res.,94, 10937–10954.
Mellor, G. L., and T. Yamada, 1974: A hierarchy of
turbulence closure models for planetary boundary
layers. J. Atmos. Sci.,31, 1791–1806.
Mellor, G. L., and T. Yamada, 1982: Development of
a turbulence closure model for geophysical fluid
problems. Rev. Geophys. Space Phys.,20, 851–875.
Miller, M., A. Beljaars, and T. Palmer, 1992: The sensi-
tivity of the ECMWF model to the parameteriza-
tion of evaporation from the tropical oceans. J.
Climate,5, 418–434.
Miller, R. L., G. A. Schmidt, and D. T. Shindell, 2006:
Forced annular variations in the 20th century
IPCC AR4 simulations. J. Geophys. Res.,111,
D18101, doi:10.1029/2005JD006323.
Mitchell, T. D., and P. D. Jones, 2005: An improved
method of constructing a database of monthly cli-
mate observations and associated high-resolution
grids. Int. J. Climatol.,25, 693–712, doi: 10.1002/
joc.1181.
Mizuta, R., H. Yoshimura, H. Murakami, M. Matsueda,
H. Endo, T. Ose, K. Kamiguchi, M. Hosaka, M.
Sugi, S. Yukimoto, S. Kusunoki, and A. Kitoh,
2011: Climate simulations using the improved
MRI-AGCM with 20-km grid. J. Meteror. Soc.
Japan, in press.
Mochizuki, T., M. Ishii, and M. Kimoto, Y. Chikamoto,
M. Watanabe, T. Nozawa, T. Sakamoto, H. Shio-
gama, T. Awaji, N. Sugiura, T. Toyoda, S. Yasu-
naka, H. Tatebe, and M. Mori, 2010: Pacific
decadal oscillation hindcasts relevant to near-term
climate prediction. Proceedings of the National
Academy of Sciences of the United States of
America,107, 1833–1837, doi: 10.1073/pnas
.0906531107.
Monahan, E. C., D. E. Spiel, and K. L. Davidson, 1986:
A model of marine aerosol generation via white-
caps and wave disruption. E. C. Monahan and
G. M. Niocaill, Eds., Oceanic Whitecaps, D. Rei-
del, 167–174.
Murakami, M., 1990: Numerical modeling of dynamical
and microphysical evolution of an isolated convec-
tive cloud—the 19 July 1981 CCOPE cloud. J.
Meteor. Soc. Japan,68, 107–128.
Nakamura, T., T. Toyoda, T. Awaji, and Y. Ishikawa,
2004: Tidal mixing in the Kuril Straits and its im-
pact on ventilation in the North Pacific Ocean. J.
Oceanogr.,60, 411–423.
Nakanishi, M., 2001: Improvement of the Mellor-
Yamada turbulence closure model based on large-
eddy simulation data. Boundary-Layer Meteor.,
25, 63–88.
Nakanishi, M., and H. Niino, 2004: An improved
Mellor-Yamada level-3 model with condensation
physics: Its design and verification. Boundary-
Layer Meteor.,112, 1–31.
Nakanishi, M., and H. Niino, 2006: An improved
Mellor-Yamada level-3 model: Its numerical stabil-
ity and application to a regional prediction of ad-
vection fog. Boundary-Layer Meteor.,119, 397–
407.
Nakanishi, M., and H. Nino, 2009: Development of an
improved turbulence closure model for the atmo-
spheric boundary layer. J. Meteor. Soc. Japan,87,
895–912.
Nakano, H., and N. Suginohara, 2002: E¤ects of bottom
boundary layer parameterization on reproducing
deep and bottom waters in a world ocean model.
J. Phys. Oceanogr.,32, 1209–1227.
Noh, Y., and H.-J. Kim, 1999: Simulations of tempera-
ture and turbulence structure of the oceanic bound-
ary layer with the improved near-surface process.
J. Geophys. Res.,104, 15621–15634.
February 2012 S. YUKIMOTO et al. 61

Noh, Y., Y.-J. Kang, T. Matsuura, and S. Iizuka, 2005:
E¤ect of the Prandtl number in the parameteriza-
tion of vertical mixing in an OGCM of the trop-
ical Pacific. Geophys. Res. Lett.,32, L23609,
doi:10.1029/2005GL024540.
Nordeng, T. E., 1994: Extended versions of the convec-
tive parametrization scheme at ECMWF and their
impact upon the mean and transient activity of the
model in the tropics. Research Department Techni-
cal Memorandum No. 206, ECMWF, Shinfield
Park, Reading, Berks, UK.
Oki, T., and Y. C. Sud, 1998: Design of total runo¤ inte-
grating pathways (TRIP)—a global river channel
network. Earth Interactions,2, 1–37.
Onogi, K., J. Tsutsui, H. Koide, M. Sakamoto, S. Ko-
bayashi, H. Hatsushika, T. Matsumoto, N. Yama-
zaki, H. Kamahori, K. Takahashi, S. Kadokura,
K. Wada, K. Kato, R. Oyama, T. Ose, N. Man-
noji, and R. Taira, 2007: The JRA-25 reanalysis.
J. Meteor. Soc. Japan,85, 369–432, doi:10.2151/
jmsj.85.369.
Pickart, R. S., D. J. Torres, R. A. Clarke, 2002: Hydrog-
raphy of the Labrador Sea during Active Convec-
tion. J. Phys. Oceanogr.,32, 428–457.
Prather, M. J., 1986: Numerical advection by conserva-
tion of second order moments. J. Geophys. Res.,
91, 6671–6681.
Peng, Y., and U. Lohmann, 2003: Sensitivity study of the
spectral dispersion of the cloud droplet size distri-
bution on the indirect aerosol e¤ect. Geophys. Res.
Lett.,30, 1507, doi:10.1029/2003GL017192.
Prospero, J. M., P. Ginoux, O. Torres, S. E. Nicholson,
and T. E. Gill, 2002: Environmental characteriza-
tion of global sources of atmospheric soil dust
identified with the Nimbus 7 total ozone map-
ping spectrometer (TOMS) absorbing aerosol
product. Rev. Geophys.,40, 1002, doi:10.1029/
2000RG000095.
Ramankutty, N., and J. A. Foley, 1999: Estimating his-
torical changes in global land cover: Croplands
from 1700 to 1992. Global Biogeochem. Cy.,13,
997–1027.
Ra
¨isa
¨nen, P., 1998: E¤ective longwave cloud fraction and
maximum-random overlap of clouds: A problem
and a solution. Mon. Wea. Rev.,126, 3336–3340.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Fol-
land, L. V. Alexander, D. P. Row-ell, E. C. Kent,
and A. Kaplan, 2003: Global analyses of sea sur-
face tempera-ture, sea ice, and night marine air
temperature since the late nineteenth century. J.
Geophys. Res.,108(D14), 4407, doi:10.1029/
2002JD002670.20.
Redi, M. H., 1982: Oceanic isopycnal mixing by coordi-
nate rotation. J. Phys. Oceanogr.,12, 1154–1158.
Remer, L. A., R. G. Kleidman, R. C. Levy, Y. J. Kauf-
man, D. Tanre
´, S. Mattoo, J. V. Martins, C.
Ichoku, I. Koren, H. Yu, and B. N. Holben, 2008:
Global aerosol climatology from the MODIS
satellite sensors. J. Geophys. Res.,113, D14S07,
doi:10.1029/2007JD009661.
Rotstayn, L. D., 2000: On the ‘‘tuning’’ of autoconver-
sion parameterization. J. Geophys. Res.,105
(D12), 15495–15507.
Rutledge, S. A., and P. V. Hobbs, 1983: The mesoscale
and microscale structure and organization of
clouds and precipitation in midlatitude cyclones.
VIII: A model for the ‘‘seeder-feeder’’ process in
warmfrontal rainbands. J. Atmos. Sci.,40, 1185–
1206.
Seinfeld, J. H., and S. N. Pandis, 1997: Atmospheric
chemistry and physics: From air pollution to cli-
mate change. A Wiley-Interscience publication,
New York, NY, 1326 pp.
Shao, Y., and H. Lu, 2000: A simple expression for wind
erosion threshold friction velocity. J. Geophys.
Res.,105, 22437–22443.
Shao, Y., M. R. Raupach, and J. F. Leys, 1996: A model
for predicting aeolian sand drift and dust entrain-
ment on scales from paddock to region. Aust. J.
Soil. Res.,34, 309–342.
Shibata, K., and T. Aoki, 1989: An infrared radiative
scheme for the numerical models of weather and
climate. J. Geophys. Res.,94, 14923–14943.
Shibata, K., and A. Uchiyama, 1992: Accuracy of the
delta-four-stream approximation in inhomogene-
ous scattering atmospheres. J. Meteor. Soc. Japan,
70, 1097–1109.
Slingo, A., 1989: A GCM parameterization for the short-
wave radiative properties of water clouds. J. At-
mos. Sci.,46, 1419–1427.
Sloyan, B. M., and S. R. Rintoul, 2001: The Southern
Ocean limb of the global deep overturning circula-
tion. J. Phys. Oceanogr.,31, 143–173.
Smagorinsky, J., 1963: General circulation experiments
with the primitive equations. I. The basic experi-
ment. Mon. Wea. Rev.,91, 99–164.
Smith, R. D., and J. C. McWilliams, 2003: Anisotropic
horizontal viscosity for ocean models. Ocean Mod-
elling,5, 129–156.
Smolarkiewicz, P. K., 1984: A fully multidimensional
positive definite advection transport algorism with
small implicit di¤usion. J. Comput. Phys.,54,
325–362.
Spiro, P. A., D. J. Jacob, and J. A. Logan, 1992: Global
inventory of sulfur emissions with 11resolu-
tion. J. Geophys. Res.,97, 6023–6036.
Steele, M., R. Morley, and W. Ermold, 2001: PHC: A
global ocean hydrography with a high-quality
Arctic Ocean. J. Climate,14, 2079–2087.
St. Laurent, L., H. Simmons, and S. Jayne, 2002: Esti-
mating tidally driven energy in the deep ocean.
Geophys. Res. Lett.,29, 2106–2110.
62 Journal of the Meteorological Society of Japan Vol. 90A

Stothers, R. B., 1996: Major optical depth perturbations
to the stratosphere from volcanic eruptions: Pyrhe-
liometric period, 1881–1960. J. Geophys. Res.,
101(D2), 3901–3920.
Stothers, R. B., 2001: Major optical depth perturbations
to the stratosphere from volcanic eruptions: Stellar
extinction period, 1961–1978. J. Geophys. Res.,
106(D3), 2993–3003.
Takemura, T., T. Nozawa, S. Emori, T. Y. Nakajima,
and T. Nakajima, 2005: Simulation of climate re-
sponse to aerosol direct and indirect e¤ects with
aerosol transport-radiation model. J. Geophys.
Res.,110, D02202, doi:10.1029/2004JD005029.
Talley, L. D., J. L. Reid, and P. E. Robbins, 2003: Data-
based meridional overturning streamfunctions for
the global ocean. J. Climate,16, 3213–3226.
Tanaka, T. Y., and M. Chiba, 2005: Global simulation
of dust aerosol with a chemical transport model,
MASINGAR. J. Meteor. Soc. Japan,83A, 255–
278.
Tanaka, T. Y., and M. Chiba, 2006: A numerical study
of the contributions of dust source regions to
the global dust budget. Global and Planetary
Changes,52, 88–104, doi:10.1016/j.gloplacha.2006
.02.002.
Tanaka, T. Y., K. Orito, T. T. Sekiyama, K. Shibata, M.
Chiba, and H. Tanaka, 2003: MASINGAR, a
global tropospheric aerosol chemical transport
model coupled with MRI/JMA98 GCM: Model
description. Pap. Meteor. Geophys.,53(4), 119–
138.
Tao, W.-K., 1989: An ice-water saturation adjustment.
Mon. Wea. Rev.,117, 231–235.
Tegen, I., and I. Fung, 1995: Contribution to the atmo-
spheric mineral aerosol load from land surface
modification. J. Geophys. Res.,100 (D9), 18707–
18726, doi:10.1029/95JD02051.
Tiedtke, M., 1989: A comprehensive mass flux scheme
for cumulus parameterization in large scale mod-
els. Mon. Wea. Rev.,117, 1779–1800.
Tiedtke, M., 1993: Representation of clouds in large scale
models. Mon. Wea. Rev.,121, 3040–3061.
Trenberth, K., and J. M. Caron, 2001: Estimates of meri-
dional atmosphere and ocean heat transports. J.
Climate,14, 3433–3443.
Trenberth, K., T. Fasullo, and J. Kiehl, 2009: Earth’s
Global Energy Budget. Bull. Amer. Meteor. Soc.,
doi:10.1175/2008BAMS2634.
Tsujino, H., H. Hasumi, and N. Suginohara, 2000: Deep
pacific circulation controlled by vertical di¤usivity
at the lower thermocline depth. J. Phys. Oceanogr.,
30, 2853–2865.
Tsujino, H., T. Motoi, I. Ishikawa, M. Hirabara, H. Na-
kano, G. Yamanaka, T. Yasuda, and H. Ishizaki,
2010: Reference manual for the Meteorological
Research Institute Community Ocean Model
(MRI.COM) Version 3. Tech. Rep. of MRI,59,
241 pp.
Tsujino, H., M. Hirabara, H. Nakano, T. Yasuda, T.
Motoi, and G. Yamanaka, 2011: Simulating pres-
ent climate of the global ocean-ice system using
the Meteorological Research Institute Community
Ocean Model (MRI.COM): Simulation character-
istics and variability in the Pacific sector. J. Ocean-
ogr., doi:10.1007/s10872-011-0050-3.
Twomey, S., 1974: Pollution and the planetary albedo.
Atmos. Environ.,8, 1251–1256.
Twomey, S., 1991: Aerosols, clouds and radiation. At-
mos. Environ.,25A (11), 2435–2442.
Visbeck, M., J. Marshall, and T. Haine, 1997: Specifi-
cation of eddy transfer coe‰cients in coarse-
resolution ocean circulation models. J. Phys. Oce-
anogr.,27, 381–402.
Watanabe, M., T. Suzuki, R. O’ishi, Y. Komuro, S.
Watanabe, S. Emori, T. Takemura, M. Chikira,
T. Ogura, M. Sekiguchi, K. Takata, D. Yamazaki,
T. Yokohata, T. Nozawa, H. Hasumi, H. Tatebe,
and M. Kimoto, 2011: Improved climate simula-
tion by MIROC5: Mean states, variability, and cli-
mate sensitivity. J. Climate,23, 6312–6335.
Webb, R. W., C. E. Rosenzweig, and E. R. Levine, 2000:
Global soil texture and derived water-holding ca-
pacities (Webb et al.). data set. (available at http://
www.daac.ornl.gov) Oak Ridge National Labora-
tory Distributed Active Archive Center, Oak
Ridge, Tennessee, U.S.A.
Wentz, F. J., 1997: A well-calibrated ocean algorithm
for SSM/I. J. Geophys. Res.,102 (C4), 8703–
8718.
Yoshimura, H., and T. Matsumura, 2003: A semi-
Lagrangian scheme conservative in the vertical
direction. CAS/JSC WGNE Research Activities
in Atmospheric and Ocean Modeling, 33,3.19–
3.20.
Yoshimura, H., and T. Matsumura, 2005: A two-time-
level vertically-conservative semi-Lagrangian semi-
implicit double Fourier series AGCM. CAS/JSC
WGNE Research Activities in Atmospheric and
Ocean Modeling, 35, 3.25–3.26.
Yoshimura, H., and S. Yukimoto, 2008: Development of
a Simple Coupler (Scup) for Earth System Model-
ing. Pap. Meteor. Geophys.,59, 19–29.
Yukimoto, S., and Y. Kitamura, 2003: An investigation
of the irregularity of El Nin˜o in a coupled GCM.
J. Meteor. Soc. Japan,81, 599–622.
Yukimoto, S., A. Noda, A. Kitoh, M. Sugi, Y. Kita-
mura, M. Hosaka, K. Shibata, S. Maeda, T.
Uchiyama, 2000: The New Meteorological Re-
search Institute Coupled GCM (MRI-CGCM2)—
Model Climate and Variability—. Pap. Meteor.
Geophys.,51, 47–88. (Available from http://www
.jstage.jst.go.jp/article/mripapers/51/2/47/_pdf ).
February 2012 S. YUKIMOTO et al. 63

Yukimoto, S., A. Noda, A. Kitoh, M. Hosaka, H. Yosh-
imura, T. Uchiyama, K. Shibata, O. Arakawa,
and S. Kusunoki, 2006: Present-day climate and
climate sensitivity in the Meteorological Re-
search Institute Coupled GCM version 2.3 (MRI-
CGCM2.3). J. Meteor. Soc. Japan,84, 333–363.
Yukimoto, S., H. Yoshimura, M. Hosaka, T. Sakami, H.
Tsujino, M. Hirabara, T. Y. Tanaka, M. Deushi,
A. Obata, H. Nakano, Y. Adachi, E. Shindo, S.
Yabu, T. Ose, and A. Kitoh, 2011: Meteorological
Research Institute Earth System Model Version 1
(MRI-ESM1)—Model Description—. Tech. Rep.
of MRI,64, 83 pp. (Available from http://www
.mri-jma.go.jp/Publish/Technical/index_en.html).
Zhang, Y., W. B. Rossow, A. A. Lacis, V. Oinas, and
M. I. Mishchenko, 2004: Calculation of radiative
fluxes from the surface to top of atmosphere based
on ISCCP and other global data sets: Refinements
of the radiative transfer model and the input
data. J. Geophys. Res.,109, D19105, doi:10.1029/
2003JD004457.
Zobler, L., 1986: A world soil file for global climate
modeling. NASA Tech. Memo., 87802, NASA,
33 pp.
64 Journal of the Meteorological Society of Japan Vol. 90A