Toward Seasonal Climate Forecasting and Climate Projections in Future

Akio KITOHMeteorological Research Institute, Tsukuba, Japan

Tokyo Climate Conference, 6 July 2009, Tokyo

ENSO influences worldwide climate even out of the tropical Pacific on seasonal to

inter-annual scales.

Sea Surface Temperature anomaly in November 1997

Accumulated Precipitation Anomalyduring Nov.1997-Apr.1998

from BAMS, 1999, 80, S1-48

from JMA webpage

ENSO is the most successfully predicted large-scale phenomenon on seasonal to

inter-annual scales ObservationDec1997 - Feb1998

Prediction from

31 July 1997by

JMA/MRI model

Precipitation

Surface Air Temperature

Sea SurfaceTemperature

JMA/MRI

4-month lead

Atmosphere - Land – Ocean Coupled Models

Atmosphere – Land Models

Atmosphere-ocean coupled models are necessary for the seasonal prediction of

ENSO and its influences

Short-term Prediction

Model

Seasonal Prediction

Model

Given Sea Surface Temperature Coupled Ocean

Local Relationship between Sea Surface Temperature (SST) and Rain Anomalies in Coupled models is more realistic than in Atmospheric

models

Rain -> SST1month lead

Rain = SST

Rain <- SST1month lag

Wang et al. (2005)ECHAM

Next JMA Seasonal Prediction System

developed by JMA/MRIJMA/MRI Coupled Model

• JMA/MRI Unified Atmospheric Model

• 180km Resolution (TL95L40)

• Ocean Model (MRI.COM)• 1.0°by 0.3-1.0° 50-layer

• 1-hour Coupling• Wind-stress, Heat-flux Adjustment

Ocean Initials and Data• MOVE/MRI.COM• Usui et al. (2006)• 3D-VAR (T,S)

• TAO/TRITON array• Altimeter Data• Argo Float

Improved NINO3.4 SST Prediction Skill

持続予報

NEW

Operational

持続予報

気候値予報NEW

Operational

(170-120W, 5S-5N)

NEW

OLD

Persistent

NEWPersistent

JMA/MRI

Jan 31 => Jun-Aug (1984-2005)

Precipitation Anomaly Prediction Skill

ROC: Relative Operating Characteristic

Atmospheric Model

Coupled model shows better skill than Atmosphere-only model

blue region : Upper tercile ROC skill is better than climatological one

JMA/MRI

Coupled Model

Precipitation Anomaly Prediction Skill

ROC: Relative Operating Characteristic

Jan 31 => Jun-Aug Jul 31 => Dec-Feb

Skill for boreal winter is higher than that for boreal summer

blue region : Upper tercile ROC skill is better than climatological one

JMA/MRI

Surface Air Temperature Anomaly Prediction Skill

Jan 31 => Jun-Aug Jul 31 => Dec-Feb

ROC: Relative Operating Characteristicblue region : Upper tercile ROC skill is better than climatological one

JMA/MRI

Prediction skill of temperature is higher than that of precipitation

South Asia Summer Monsoon Index (WYI)(4-month lead: JJA from JAN)

AGCMCGCM

WYI Definition: U850–U200 [0-20N,40-110E]

Blue: ForecastRed: Analysis

ACC: 0.59

Blue: ForecastRed: Analysis

ACC: 0.35

JMA/MRI

East Asia Summer Monsoon Index (DU2)(4-month lead: JJA from JAN)

DU2 Definition ：U850[5-15N,90-130E] - U850[22.5-32.5N,110-140E]

Blue: ForecastRed: Analysis

ACC: 0.58

Blue: ForecastRed: Analysis

ACC: -0.05

CGCM AGCM

JMA/MRI

Multi-Model Ensemble

WMO Lead Centre for LRF MME APEC Climate Center

From APCC HomepageFrom LRF MME Homepage

European DEMETER Project

From ECMWF Web PageRPSS: Rank Probability Skill Score (Wilks 1995)

Multi-model ensemble skill out-performs single model ensemble with the same member size

DEMETER

Forecast quality of DEMETER hindcasts

WCRP Position Paper on Seasonal Prediction.

Report from the First WCRP Seasonal Prediction Workshop (Barcelona, Spain, 4-7 June 2007). February 2008. WCRP Informal Report No.3/2008, ICPO Publication No.127.

Skill depends on regions, seasons and variables

Significant skills for precipitation in DJF_Amazon and JJA_Southeast Asia

JJA & DJF_East Asia and JJA_Australia for temperature

DEMETER

SINTEX-F showed the highest ENSO prediction skill among 10 coupled

GCMs

Nino3.4 index(1982-2001)

Adapted from Jin et al. 2008, APCC CliPAS

JAMSTEC

Nino3.4 SSTA prediction

Luo et al. (2008)

(120º-170ºW, 5ºS-5ºN)

ENSO can be predicted out to 1-year lead and even up to 2-years ahead in some cases

by SINTEX-F

JAMSTEC

SINTEX-F Coupled Model Components

AGCM (MPI, Germany): ECHAM4 (T106L19)OGCM (LODYC, France): OPA8 (2 x 0.52, L31)Coupler (CERFACS, France): OASIS2

*No flux correction, no sea ice model

Seasonal prediction for # tropical cyclones

ECMWF Newsletter No. 112 – Summer 2007

ECMWF

has already started and shows some skill …

Occurrence location of tropical cyclones are well predicted in the

Northwest Pacific

latitude

longitude

JMA/MRI

… as occurrence location is related with ENSO

Wand and Chan (2002) etc

more tropical storms form in the SE quadrant during the warm phase, and in the NW quadrant during the cold phase,

thus ENSO prediction is the key

Toward further improvement of seasonal prediction

NWP modelTyphoon prediction modelEl Niño prediction modelSeasonal prediction modelClimate modelEarth system model

Climate model development (IPCC AR4)

It is necessary to explore other predictability sources in the Earth system

Toward further improvement of seasonal prediction

NWP modelTyphoon prediction modelEl Niño prediction modelSeasonal prediction modelClimate modelEarth system model

Improving atmosphere-ocean coupled models will lead to constant improvement of seasonal predictions based on slow-coupled process like ENSO.

On the other hand, high predictability from ENSO seems to be limited within relatively low-latitudes.

Therefore, for more complete seasonal prediction, we need to explore other influential elements that show relatively long-range persistency or predictability in the Earth system that consists of upper and/or polar atmosphere, land, snow and ice, chemical processes besides the low-latitude troposphere.

It is necessary to explore other predictability sources in the Earth system

Xie et al. (1999)

JAPAN Winter Temperature is significantly correlated with Arctic Oscillation besides

ENSO　　

・　 Atlantic SST anomaly

・　 Snow over Eurasia

・　 Arctic Sea Ice Cover

・　 Stratosphere, Ozone

・　 Volcano Eruption

・　 Global Warming

AO

ENSO

Possible Causes

Stratospheric Harbingers of Anomalous Weather

(Troposphere-Stratosphere Interaction)

Baldwin and Dunkerton (2001)

AR4 to AR5: Need of climate change information for adaptation studies in

near future

• fill a gap between seasonal-to-interannual prediction and climate change projections• sufficiently high resolution projection is needed for resolve weather extremes• changes in weather extremes will become significant much earlier than mean climate change

Another emerging issue is a projection of future changes in weather extremes in order to contribute to decision-makings for the disaster prevention and other adaptation studies under the global warming environment.

Projected changes in extremes

Intensity of precipitation events is projected to increase. Even in areas where mean precipitation decreases, precipitation intensity is projected to increase but there would be longer periods between rainfall events.“It rains less frequently, but when it does rain, there is more precipitation for a given event.” (Tebaldi et al. 2006)Extremes will have more impact than changes in mean climate

IPCC AR4IPCC AR4 CMIP3 modelsCMIP3 models

Number of TC Generated in Each Latitude

Present-day(25yr)Future(25yr)

Observation

Latitude

TC fr

eqen

cy

２０％decrease

Annual global average Present =82 Future =66 (20% decrease)

(Observation:84)

Radial Profile Change around TC

・ Large changes occur near inner-core region, 40-60% for precipitation and 15-20% for surface wind.

・ A surface wind speed increase of more than 4% can be seen up to 500 km from storm center.

Surface Wind

Radial Distance in km from Storm Center

Precipitation

Future ExperimentPresent Experiment

Change rate

Cooperation activities of the MRI groupCooperation activities of the MRI group (by Earth Simulator Earth Simulator computed model outputscomputed model outputs for adaptation for adaptation studiesstudies)

Adaptation study in Coastal Zones of Caribbean countries: Barbados(one, 2005), Belize (one, 2005)

Adaptation studies in Colombian coastal areas, high mountain ecosystems: Colombia (two, 2005; 2009)

Adaptation to Climate Impacts in the Coastal Wetlands of the Gulf of Mexico: Mexico (two, 2006)

Adaptation to Rapid Glacier Retreat in the Tropical Andes: Peru (one, 2006), Ecuador (one, 2006; 2009), Bolivia (one, 2006; 2009)

Amazon Dieback: Brazil (two, 2008)

Cooperation under the Cooperation under the JICAJICA (Japan International Cooperation Agency)(Japan International Cooperation Agency) fundsfunds Adaptation studies in agriculture in Argentina: Argentina (three, 2008) Adaptation studies in monsoon Asia: Bangladesh, Indonesia, Philippines, Thailand, Vietnam (one each, 2008 & 2009) Adaptation studies in the Yucatan: Mexico (two, 2009)

Cooperation under the Cooperation under the World BankWorld Bank funds funds

Other collaborations with India, Korea, Thailand, USA, Switzerland, …

This collaboration started after COP10 (2004)

SUMMARY• ENSO is the major source of the predictability on

seasonal to inter-annual time-scales at the present. ENSO prediction was much improved for the past a few decades, and can be extended up to 1-year lead or longer.

• Probabilistic representation using initial ensembles is adapted for seasonal prediction of precipitation and surface air temperature because of small ratios of signal to noise. Multi-model ensemble technique contributes to improvement of seasonal prediction skills. Seasonal prediction skills are strongly dependent on regions, seasons and the elements to predict as well as ENSO situations.

• In addition to steady improvement of atmosphere-ocean coupled models, it is necessary to explore other predictability sources in the Earth system in future.

• High resolution model is now used to project future changes in weather extremes and tropical cyclones under the global warming environment. Such data is useful for various application studies, including adaptation to climate change.