Presentation on theme: "Regional Climate Change Scenario over East Asia 2003 June Trieste, ICTP RegCM Workshop Won-Tae Kwon Meteorological Research Institute, Korea."— Presentation transcript:
Regional Climate Change Scenario over East Asia 2003 June Trieste, ICTP RegCM Workshop Won-Tae Kwon Meteorological Research Institute, Korea
We are interested in the impacts of future climate change in Korea Climate Change Simulation with CGCM Regional Climate Change Scenario - dynamical downscaling - statistical adjustment
Need for impact assessment and adaptation on future climate change for various socio-economic and natural sectors for the sustainable development Korea is located at the eastern coast of the largest continent of the earth – large climate variability Meso-scale complex topography and high population density Most people want to hear about what will happen in their own back yards Need for high-resolution regional climate information for impact study Issues we need to consider….
Coupled Climate Model ECHO-G ECHAM4 T30/L19 dt = 30 minute Roeckner et al. 1996, MPI HOPE-G T42 + equ. ref. /L20 dt = 2 hours Wolff et al. 1997, DKRZ OASIS dt = 1 day Valcke et al. 2000, CERFACS 10 fluxes 4 surface conditions MPI M&D* coupled climate model - AGCM: ECHAM4 T30 (3.75 ) - OGCM: HOPE-G T42 (2.8 ) (0.5 at 10S~10N) sea ice model included - Coupler: OASIS Flux corrections - annual mean heat and fresh-water flux correction - no momentum flux correction *MPI M&D: Max-Planck Institute for Meteorology Models and Data Group
ECHO-G 1000-year Control Simulations Performed at MPI M&D, Germany Present day values (1990) for GHG concentrations Stable global mean surface temperature and thermohaline circulation ENSO - similar pattern to observed - 2-year period dominant (Legutke and Cubasch, 2001) Annual mean T 2m and precipitation rate (red line: 11-yr moving average)
Performed at METRI/KMA, Korea Greenhouse gases only - CO 2, CH 4, N 2 O, CFCs etc - 1860~1990: observed - 1990~2100: SRES scenarios SRES updated scenarios - A2: pessimistic scenario (CO 2 820 ppmv by 2100) - B2: optimistic scenario (CO 2 610 ppmv by 2100) A2 B2 ECHO-G SRES A2, B2 scenario simulations GHGs scenarios for 1860-2100
ECHO-G Scenario Simulation Results (numbers are 2090s mean) Global Precipitation (%) Temperature( ℃ ) 4.6 3.0 6.5 4.5 4.4 2.8 10.5 6.0 East Asia A2 B2 A2 B2
2050s Climate Change Patterns: A2 Global Precipitation (%) Temperature( ℃ ) East Asia
2050s Climate Change Patterns: B2 Global Precipitation (%) Temperature( ℃ ) East Asia
Seasonal Projection over East Asia DJF Precipitation (%) Temperature( ℃ ) MAM JJA SON A2-B2: Mitigation effect
Climate Change Projection over East Asia (Multi-Model Ensemble) A2 B2 Temperature ( ℃ ) Precipitation (%)
Northern limit of Bamboo habitation Check point 0 100km Distribution of Phyllostachys 19 th C 2002 경북 영천 자양면 충효리 경북 예천 풍양면 와룡리
TOPOGRAPHY (M) 27km resolution 400 km resolution
ECAHM4/HOPE-G (spectral data) After post process Initial Condition ( p-level grid data) Regional Climate Model (MM5) Horizontal, vertical interpolation detailed topography CD-Rom INTERPB I.C B.C
Raid storage sever NAS storage sever Myrinet hub 10/100 switch hub Monitoring system UPS 16 nodes ( dual CPU ) cluster ElectrometerMyrinetEthernet Computing Resources : HPC CLUSTER (ENVICOM) · CPU - AMD MP2000+ 16Node ( 32 CPUs ) · MEMORY - ECC Registered DDR Ram 2 GB · Myrinet - optical cable & switch, 2U high, 3-slot enclosure for switch, 16 ports · NAS - 1.8 TB, Network attached Storage, SCSI raid Storage 15 cpu hours for 1 year integration ! = 1 week for 10 year integration !
Grid size of global model Grid size of regional model ECHAM4USGS data 96 x 48 (~400 km)125 x 105 (~27 km) Orography of global model Z echam (bilinear interpolation, Z echam ) Orography of regional model Z mm5 (Z mm5 ) Orography Blending (Z bln )
18601950200020302100 A2-G Control Finished (2002) In Progress (2003) Dynamic Downscaling Progress
We provide regional climate information with dynamic downscaling. Does it good enough for assessment studies with confidence?
Transfer function using statistical method 70% of RMS error were reduced Transfer Function [ G2G Pilot transfer function ] RCMANAL EOFA REGA TC RCM = f(TC ANAL ) GRID DATACORRECTION Eigen Mode Significant Eigen Mode RCMC_RCM
TC1, TC2 in RCM (red), O_KMA (blue), after adjustment (green)
RMS error Season RCMRCM_C1RCM_C2 CORRECTED PERCENT MAM4.731.781.4868.7% JJA9.701.451.3186.5% SON6.642.201.8672.0% DJF4.032.162.1147.6% ANNUAL6.291.901.6973.1% RMS error of daily mean temperature
Summary We may be able to provide reasonable future regional climate information for impact assessment studies with combination of dynamic downscaling and statistical adjustment. Statistical adjustment is successful for temperature, however, we still need more efforts for precipitation because there is no outstanding eigen mode.
Reduction rate of GCM to RCM – a nested domain? Understanding the variability of future climate change – mean, range, extreme events, seasonal and local difference, etc. – how can we analyze these issues? Statistical downscaling of RCM data Understanding and communication with experts from various sectors – what kind of data they need for impact assessment Further Thoughts on Unsolved Obstacles
Future Plans….. EHCO run with A2 GHG+Aerosol scenario in 2003 and maybe more later on Using RegCM3 for the downscaling of EHCO model projections Sensitivity test and Optimization for East Asia domain Statistical downscaling (transfer function) for regional scenario
0 100km IDEAS for Future Works Integrated local climate change assessment Trend Flood/drought Water resource Agriculture Fishery Health Ecosystem Forest Road Tourism Recreation Energy Industry Transportation Construction Economy… Multi-disciplinary efforts Local climate change scenario (240 years)