Climate Change Projection in Latin America by Global 20-km and 60 ...

ambiente.gov.ar

Climate Change Projection in Latin America by Global 20-km and 60 ...

25 February 2009, Buenos AiresClimate Change Projection in LatinAmerica by Global 20-km and 60-kmMesh Atmospheric ModelAkio KITOH 1 , Shoji KUSUNOKI 1 , Hiroki KONDO 2,31Meteorological Research Institute (MRI), Tsukuba, Japan2Advanced Earth Science and Technology Organization, Tsukuba, Japan3Frontier Research Center for Global Change, Yokohama, Japan


CMIP3 multi-model A1B resultsIPCC AR4Most modelsproject a wetterclimate near theRio de la Plataand drierconditions alongmuch of thesouthern Andes.Over Amazon,precipitation isprojected toincrease in DJFand decrease inJJA.


Projected changesin extremesIntensity of precipitationevents is projected toincrease.Even in areas where meanprecipitation decreases,precipitation intensity isprojected to increase butthere would be longerperiods between rainfallevents.Extremes will have moreimpact than changes inmean climate.IPCC AR4High resolution model isneeded to handle extremes


Needs for High-Resolution ModelDaily precipitation amounton certain day (mm/day)Future Change in the Annual Maximum PrecipitationFuture – Present(mm/day)20-km high-resolution gridExtreme eventsarehighly localphenomenalow-resolution280-km gridIn the lowresolutionmodel,heavy rain issmoothed outCorresponding to simulationwith high-resolution modelRightupperRightlowerThe annual maximum rain calculated byoriginal 20-km mesh daily rainTo compare with the lower figure, the annualmaximum rain is regridded to low resolution(280-km) gridThe annual maximum rain calculated byregridded daily rain, which is interpolated to280-km gird from original 20-km gridThis can be considered as a proxy of lowresolution modelCorresponding to simulationwith low-resolution modelChange in precipitation extremes can notbe captured well by low-resolution model


QuickTimeý DzÉVÉlÉpÉbÉN êLí£ÉvÉçÉOÉâÉÄÇDZÇÃÉsÉNÉ`ÉÉǾå©ÇÈǞǽDžÇÕïKóvÇ-ÇÅB


Sample tropical cyclone track and max surface windsMRIPresentFuture


Tropical Cyclones


Innovative Program ofClimate ChangeProjection for the 21stcentury(KAKUSHIN Program)FY2007-FY2011Participating groups and their studies Long-term global environmental projectionwith an earth system model- Frontier Research Center for Global Change (FRCGC) et al. Near-term climate predictionwith a high-resolution coupled ocean-atmosphere GCM- Center for Climate System Research (CCSR) of the University of Tokyo et al. Projection of changes in extremes in the futurewith super-high-resolution atmospheric models- Meteorological Research Institute (MRI) et al.http://www.kakushin21.jp/eng/index.html


Cooperation activities of the MRI group(by Earth Simulator computed model outputs for adaptation studies)Cooperation under the World Bank funds Adaptation study in Coastal Zones of Caribbean countries:Barbados(one, 2005), Belize (one, 2005) Adaptation studies in Colombian coastal areas, high mountainecosystems: Colombia (two, 2005; 2009?) Adaptation to Climate Impacts in the Coastal Wetlands of the Gulf ofMexico: Mexico (two, 2006; 2009?) Adaptation to Rapid Glacier Retreat in the Tropical Andes: Peru (one,2006; 2009?), Ecuador (one, 2006), Bolivia (one, 2006) Amazon Dieback: Brazil (two, 2008)Cooperation under the JICA (Japan International Cooperation Agency) funds Adaptation studies in agriculture in Argentina: Argentina (three, 2008) Adaptation studies in monsoon Asia: Bangladesh, Indonesia,Philippines, Thailand, Vietnam (one each, 2008 & 2009)Other collaborations with India, Korea, Thailand, USA, …This collaborationstarted after COP10(Buenos Aires 2004)


The Earth Simulator50 meterNode (8 CPU)65meterNodes: 640Magnetic DisksCPUs: 5120Peak performance: 40 Tera flopsCrossbar switchhttp://www.es.jamstec.go.jp/esc/jp/ES/index.html


Using the Earth Simulator• Peak performance: 8 GFLOPS x 8 cpu x 640 nodes = 40TFLOPS: to be replaced by about twice performancemachine ES r by March 2009• 4.3 %/year (~240,000 node hour) is allocated to theMRI group in FY2008• Costs: the above costs about 400 Million Yen/year• MRI/JMA 20-km mesh AGCM– TL959(20km) with 60 layers– Uses 30 nodes of Earth Simulator– DT = 6 min• Turnaround time integration:– 1 year integration needs between 2 days and 2 weeks• Data transfer is a big issue


Time-slice experimentsAtmosphere-OceanCouple Model, A1BScenarioHigh resolutionAtmosphere modelexperimentWCRP CMIP320 km gridOceanSSTPredictedSSTPresent1979-2003AtmosphereSSTAtmosphereSSTFuture2075-2099YearNear-future(2015-2039)run will alsobe done


How to prescribe SSTPresent SSTAR4_A1B_Scenario Exp. SST2001-Observed SST1979~2003ΔSSTAR4_20thCenturyExp. SST -2001Future SSTA1979 2003 207525209925 yearsyearsalso applies for 2015-2039AR4_SSTChangeAR4 SSTTrend2075-2099Inter-annual variabilityof Observed SST1979-2003ΔSST++=Mean SST(2075-2099) – Mean SST(1979-2003)


MRI/JMA Atmospheric GCM• JMA : Operational global NWP model since Nov 2007• MRI : Next generation climate model• Resolution: TL959(20km) with 60 layers• Time integration: Semi-Lagrangian Scheme (Yoshimura, 2004)2 days/1 year integration with DT=6 min and30 nodes of Earth Simulator (ES has total 640 nodes)• Physics– SW radiation: Shibata & Uchiyama (1992)– LW radiation: Shibata & Aoki (1989)– Cumulus convection: Prognostic Arakawa-Schubert (Randall and Pan, 1993)– Land hydrology: MJ-SiB: SiB with 4 soil-layers and 3 snow-layers– Clouds: large-scale condensation, Cumulus, stratocumulus– PBL: Mellor & Yamada (1974,1982) level-2 closure model– Gravity wave drag: Iwasaki et al. (1989) + Rayleigh friction180km20kmThis model will beused in MRI-CGCM3and MRI-ESM afterintroducing additionalphysics and tuning.


Simulation status• 20-km– Present (1979-2003): 25 years done– Near-future (2015-2039): 15 years finished• Remaining 10 years to be calculated this boreal spring– Future (2075-2099): 25 years done• 60-km– 3 initial conditions– 4 different SSTs– 25 years for present, near-future and future– All done• New simulations for AR5 to be started aroundsummer 2009


Present-day climateresolution dependency


Precipitation: Present DJF climateCMAPGPCPCRUTRMM3A25180 km120 km60 km20 km


Precipitation: Present JJA climateCMAPGPCPCRUTRMM3A25180 km120 km60 km20 km


Precipitation: Present [70W-40W]CMAP GPCPCRU TRMM3A25180 km 120 km60 km20 kmJul Jan Jul


Performance of the Model for Reproduction of Precipitation Extremes20-km GCM presentGPCP-1DDTRMM3B42PavAnnualprecipitationCDDIndex fordrynessR5dIndex forheavy rainSimulated extremes indices are in goodagreement with that of the observationsModel: 20-year meanObs : 7-year meanfrom 1998 to 2003


Future Climate20-km AGCM vs 60-km AGCM


20-km model usesthis CMIP3ensemble meanSST anomaliesEnsemble runs with 60-km AGCM


Precipitation [F-P]DJF 60kmJJA 60kmDJF 20kmJJA 20km


Mean Tmax [F-P]


Precipitation [F-P] ANNrobust among differentSSTs used


Precipitation [70W-40W, 20S-0S] DJF• scatter among I.C. members is small• mean of SST ensemble is almost same to CMIP3 SST run


Precipitation [70W-40W, 20S-0S] JJA• slight decrease in area mean winter precipitation due tocancellation between north (increase) and south (decrease)


Precipitation [F-P] 60km_ENS vs 20kmDJFJJAMAMconsistent between20km and 60km modelsSON


MRI 60km/20km vs CMIP3 MMEDJFJJAGeneral agreement. Differencesover high mountains.CMIP3_DJFCMIP3_JJA


Total moisture flux & Divergence (DJF)PF-P• Increasing moisture inflow from the Atlantic• Northerly moisture into the Parana Basin


Total moisture flux & Divergence (JJA)PF-PMoisture divergence in JJA in western Amazon


index for heavy precipitationPINT: Simple Daily precipitation IntensityIndex (aka SDII)PINT=Annual total precipitationNumber of rain daywhere “rain day”: day of precipitation ≧ 1 mm/dayRx5D: Maximum 5-day total precipitation


PINT: Precipitation Intensity 20kmPF-PF


PINT: Precipitation Intensity F-P60km vs 20km60km20kmconsistent between20km and 60km models


RX5D: maximum 5-day precipitation 20kmPF-PFLarge RX5D overAmazon and SouthernBrazilRX5D increases almostall South Americaexcept for Chile


RX5D: maximum 5-day precipitation F-P60km vs 20km60km20km20km projects largerincrease in RX5D


Drought indexMaximum number of consecutivedry days (CDD)where “dry day”: day of precipitation < 1 mm/dayPrecipitationConsecutive dry daysDay


CDD: consecutive dry days 20kmPF-PFLarge CDD change overBrazilian highland andChile


CDD: consecutive dry days F-P60km vs 20km60km20kmlarger CDD increase inChile by 20km model


10-year data over southern Amazon (Ariquemes 63W 10S)Daily precipitation: present vs futurePresent-day simulationDaily precipitation maximum temperatureminimum temperature diurnal temperature rangeIntermittent rain in dry season stops in the future climate run


Extremes indices on temperatureUchiyama et al. (2006) SOLAETR(Intra-annual extreme temperature range)ETR increases over Brazilian highland due to increase in Tmax


PCMIP3 MMEAnnual RiverflowNohara et al. 2006(F-P)/P• decreasing river flow in western Amazon• increasing river flow in Parana Basin


Summary• Resolution of climate models becomes finer; nowwe can use 60-km or even 20-km mesh globalclimate models• Topography is better represented by highresolution model• Weather extremes are better represented byhigh resolution model• Spatial patterns of precipitation extremesindices are similar between 60-km and 20-kmmodels, but intensity differs in some indices


AcknowledgementsThis study was conducted as a part of researchtheme ”Projection of the change in the future weatherextremes using super-high-resolution atmosphericmodels” under the framework of the Innovative Programof Climate Change Projection for the 21st Century(KAKUSHIN), funded by the Ministry of Education,Culture , Sports, Science and Technology (MEXT). TheKAKUSHIN program is a 5-year research programstarted in April 2007. This is also a collaboration workbetween MRI and JICA and between MRI and WorldBank.

More magazines by this user
Similar magazines