Database for Policy Decision making
for Future climate change (d4PDF)
Hideo Shiogama (NIES, Japan)
& the MRI+MIROC joint team
The joint team of the MRI & MIROC groups has produced the huge
ensembles of the 60-km MRI AGCM:
1. 100 member historical runs during 1950-2010
2. 90 member +4K runs of 60-year length
• △SST patterns are taken from RCP8.5 runs of 6 AOGCMs
• We scaled △SST of each AOGCM to adjust the global mean
temperature changes to be +4K.
• We add the △SST to the detrended SST pattern of the
observations and use the 2090 forcing of RCP8.5.
• 15 initial condition ensembles for each △SST pattern
3. 100 member historical DETRENDED runs during 1950-2010
• We use a detrended SST and the preindustrial forcing.
We also downscaled the historical and +4K runs using the 20-km MRI
RCM around Japan.
Database for Policy Decision making
for Future climate change (d4PDF)
Daily precipitation (%) Daily precipitation (%) F req uency (%) Ra tio o f fr e q. (+ 4 K/ pr e se n t)
(a) Present day (b) Changes in the +4K world
Return Period = 1yr
10yr
100yr
Histogram of daily mean precipitation at Tokyo
(60km AGCM)
(R. Mizuta, H Shiogama) 6 patterns of △SST (CMIP5
GCMs) & the ensemble mean
• The high resolution MRI AGCM has a good skill on the simulations of daily precipitation. • The 100 member ensemble enables us to investigate very rare events (return prd > 100 yr). • In the +4K world, heavier precipitation events have larger increases of frequency.
The continental avraged annual mean surface air
temperature changes.
Obs.(CRUTEM4)
ALL NAT
Shading: min-max of 100 members Anomalies from 1951-1970 averages.
• The ALL runs well reproduce the observed warming of the continental averaged annual mean temperature.
Attribution of historical changes in frequencies
of record breaking temperature and
precipitation extreme events
Hideo Shiogama
1*, Yukiko Imada
2, Ryo Mizuta
2, Kohei Yoshida
2, Masato Mori
3,
Osamu Arakawa
4, Mikiko Ikeda
5, Chiharu Takahashi
3, Miki Arai
3, Masayoshi Ishii
2,
Nobuhito Mori
6,Izuru Takayabu
2, Eiichi Nakakita
6, Masahiro Watanabe
3&
Masahide Kimoto
31Center for Global Environmental Research, National Institute for Environmental Studies, Tsukuba, Japan 2Meteorological Research Institute, Tsukuba, Japan
3Atmosphere and Ocean Research Institute, the University of Tokyo, Kashiwa, Japan 4Faculty of Life and Environmental Sciences, University of Tsukuba, Tsukuba, Japan 5Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan
Area factions of occurrence of
record breaking daily extreme events
Annual Warmest Nights
Shading: 90% confidence intervals Dotted lines: min-max of 100 runs
Annual most intense rain
• Area fractions of record breaking events decline as the observation data accumulate (e.g. Meehl et al. 2009, GRL).
• Anthropogenic climate change induced significant increases of probability of record breaking annual warmest nights and annul most intense rain after 1990s.
PDFs of area fractions of record extreme events during
the 2001-2010 period.
Annual Warmest Nights
Annual most intense rain
H ad E X 2 ( m e an ) H ad E X 2 ( m e an )
24 times
2 times
• Anthropogenic climate change increase the chance of more area fractions of new record than the observed area fractions during the 2001-2010 period by 24 times for the annual warmest nights and 2 times for the annual most intense rain events.
Summary
• The Japanese modelling centers (MRI&MIROC) produced the large
ensembles of AGCM and RCM to mainly investigate the uncertainty
of internal variability in the D&A analyses and the future projections.
• The huge ensemble of the high resolution AGCM allows the robust
attribution of historical changes in extreme events.
• The output data will be available at http://www.diasjp.net/. We are
transferring the data, but it will take about 6 month due to the 2PB
data size!
• In the next financial year (April 2016-), we will perform +2K
ensemble which is closely related to the HAPPI-MIP.
Anthropogenic changes in probability (ALL minus NAT) of
record extreme events during the 2001-2010 period
1951-2010 trends of HadEX2 (Donat et al. 2013, JGR)
Annual Warmest Nights
Annual most intense rain
Histogram of daily mean precipitation at Tokyo of
OBS and the MRI AGCM
Daily mean precipitation
(mm/day)
Fr
eq.
(
%
)
OBSThe MRI AGCM has
amazing skills in simulations
of extreme events.
Daily mean precipitation
(mm/day)
Fr
eq.
(
%
)
OBS
Histogram of daily precipitation at Tokyo of OBS
Daily mean precipitation
(mm/day)
Fr
eq.
(
%
)
OBS
Histogram of daily precipitation at Tokyo of OBS
and the MRI AGCM
The large ensemble enable
us to investigate very
Global area factions of occurrence of
record breaking extreme events
Shading: 90% confidence range Dotted lines: min-max
PDFs of area fractions of new record extreme events
during the 2001-2010 period.
Changes in probability (ALL minus NAT) of new record
extreme events during the 2001-2010 period.
確 率 密 度 地球シミュレーター「特別推進課題」
地球温暖化対策に資するアンサンブル気候予測実験データベース
日本の温暖化施策決定のための統一シナリオ 「環境省『適応』データセット」を大幅拡充 創生P テーマ間連携最新
高解像度
大気モデル実験
大量アンサンブル
で
高精度の統計情報
気温将来変化 現在・温暖化時4℃上昇
した将来気候を予測
多様な
影響評価
に活用可能
ES最新機を利用内部変動 発生頻度の低い異常天候や 極端気象の変化の不確実性 を十分に評価できていない。 気候モデル 排出シナリオ
気候モデルを用いた地球温暖化予測における不確実性
Global, Large-scale: CMIP5実験
Extremes, Regional-scale: 60kmモデル実験 (創生プロC実験 + 環境省・気象庁 気候変動予 測データ)でカバー。
高解像度・
大量アンサンブル
で
統計情報が必要
(IPCC AR5)
モデルと実験設定1
• 過去実験: 1951–2010 の60年、100メンバー – SSTはCOBE-SST2(Hirahara et al. 2014)に時空間的に連続な100種類の摂動を加算 – 摂動は観測の不確実性（解析誤差）の情報から生成 • 非温暖化過去実験: 1951–2010 の60年、100メンバー – 過去実験において温暖化トレンドを除いたSSTを使用 温暖化トレンドを含む 過去60年の時間変動 (赤線;COBE-SST2) 観測不確実性を表す 摂動 (δT) 60km全球大気モデル: MRI-AGCM3.2 (Mizuta et al. 2012) 文部科学省の「21世紀気候変動予測革新プログラム」、 環境省の「地域気候変動予測データ」でも使用 日本域は 20km領域気候モデルへ ダウンスケーリング
モデルと実験設定2
• 将来実験: 産業革命前から4℃昇温した状態を60年、6×15=90メンバー – 海面水温の温暖化パターンとして、CMIP5の6種類のCGCMで地上気温が4℃上昇 したときの海面水温変化を算出し、温暖化トレンドを除いた過去60年の海面水温 に上乗せ – 過去実験と同様の摂動を15種類 – 温室効果ガス濃度はRCP8.5シナリオの2090年相当 6種の温暖化 パターン(CMIP5) (ΔT) 温暖化トレンドを除いた 過去60年の時間変動 (青線;COBE-SST2) 観測不確実性を 表す15摂動 (δT)日平均降水量(mm/day) 日平均降水量(mm/day) 頻度 (%) 頻度の比 (将来 /現在 ) (a) 現在の東京の日降水量頻度分布 (b) +4℃で頻度が何倍になるか 1年に1度 10年に1度 100年に1度
東京での日降水量頻度分布 (60km model)
(H Shiogama, R. Mizuta)熱帯低気圧の１年あたりの発生数 発生確率 [％ ] 過去実験 [100] N=84.6 (60年) 過去実験 [1] N=84.9 (60年) 観測 [1] N=84.3 (32年) 将来実験 [1] N=55.7 (60年) 将来実験 [90] N=54.7 (60年) ※[ ]はメンバー数 年々変動の標準偏差 HPB(1951-2010): 10.35 (±0.90) HPB(1979-2010): 9.74 (±1.24) 観測(1979-2010): 8.91 HFB(2051-2110): 8.76 (±0.88)
熱帯低気圧全球年発生数の確率分布
(by K. Yoshida)
10年に1度の日降水量
メンバー数を増やすことによって、よりはっきりした増減の分布が得られる。
1 member 10 members 90 members
Past +4K Change Rate ↑(将来の10年に1度)/(現在の10年に1度) (R. Mizuta)
We selected the periods 2031-2050 (year 2040), 2081-2100 (2090) and 2131-2150 (2140) for the SST warming patterns (forcing conditions) in the +2˚C, +4˚C and +6˚C warmer world.
CMIP5 GCM names scaling factors for +2˚C scaling factors for +4˚C scaling factors for +6˚C CCSM4 1.22758 1.10981 1.14547 GFDL-CM3 0.785844 0.75166 HadGEM2-AO 1.24267 0.902224 MIROC5 1.17132 1.06162 MPI-ESM-MR 1.28834 1.01852 MRI-CGCM3 1.35323 1.13509
XX年に1度の日降水量
過去・将来変化とも分布は大きく変わらないが、 Return periodが長いほど増加が大きい。
Once in 10 years Once in 30 years Once in 100 years
Past
+4K Change
Rate
