ImpactsImpacts ofof GlobalGlobal ClimateClimate andand EmissionEmission ChangesChanges onon U.S.U.S. AirAir QualityQuality Xin-Zhong Liang (PI), [email protected] Illinois State Water Survey (ISWS) Co-PIs: Ho-Chun Huang, Allen Williams, Michael Caughey, Kenneth Kunkel (ISWS) Donald J. Wuebbles Department of Atmospheric Sciences (DAS) Other Investigators: Jinhong Zhu, Zhining Tao, Min Xu (ISWS) Kenneth O. Patten, Jintai Lin (DAS) University of Illinois at Urbana-Champaign RD830963010 2003-2006 AcknowledgementAcknowledgement Federal grant EPA Science to Achieve Results (STAR) Supercomputing support NCSA/UIUC, NOAA/FSL, NCAR RD830963010 2003-2006 ObjectiveObjective To quantify, and understand the uncertainties of, the individual and combined impacts of global climate and emission changes on U.S. air quality, from the present to 2050 and 2100. 2003-2006 RD830963010 USAUSA AQAQ FactorsFactors U.S. air quality is determined by complex interactions over a range of spatial and temporal scales from (1) chemical processes and emissions on local to regional scales, (2) long-range transport of global pollutants and precursors, and (3) global and regional climate changes and variability. 2003-2006 RD830963010 Global Model System RegionRegionalal Model System Target Areas Resolution: 150-300 km Resolution: 30 km Resolution: 10 km Present Climate & Regional Focal Areas GCM Projections Climate Model Northeast Midwest West Coast Texas Intercontinental Regional Chemical Transport Air Quality Model Model Emissions Model O3 Global Emissions NOx VOCs CO & Projections U.S. NEI99 & Projections The nested global-regional modeling system: its components and their interactions UncertaintyUncertainty • Biases in the climate and chemistry arising from the driving global models and the incomplete physics of the regional models • Different climate sensitivities of the driving global models and the regional models • Inconsistencies in the coupled global- regional modeling system • Pollutant emissions error • Natural variability 2003-2006 RD830963010 GlobalGlobal ClimateClimate ModelsModels • LOW climate sensitivity: Parallel Climate Model (PCM) of the National Center for Atmospheric Research/U.S. Department of Energy • HIGH climate sensitivity: Hadley Centre third generation climate model (HAD) 2003-2006 RD830963010 FutureFuture ProjectionsProjections • PCM simulations for 2050 and 2100 – A1Fi : high emissions scenario – B1: low emissions scenario • HAD simulations for 2100 – A2: intermediate high emissions scenario – B2: intermediate low emissions scenario 2003-2006 RD830963010 D1 D2a D2c D2b D2d Regional computational domain design. The global models provide the regional models with lateral boundary conditions in the buffer zones (shaded outer edges). The climate model uses a grid spacing of 30-km over domain D1, while the air quality model uses 90-km in D1 and 30-km in subdomains D2a-d. RCM Resolves Orographic Effects Better Than GCM CMM5 Skill: Daily Fluctuations CMM5 Precipitation Diurnal Cycle Climate JJA PR (norm) Diurnal Cycle: Central Plains 1.8 Dai 1.6 KF CTL 1.4 MUL_4km 1.2 1 0.8 0.6 0.4 0.2 0 0 3 6 9 12151821 Hour Much More Than That… CumulusCumulus SensitivitySensitivity The following examples show • the sensitivity of the RCM downscaling to the cumulus parameterization for both the present-day climate simulation and future change projection • emphasis the need for ensemble cumulus prediction in climate and air quality 2003-2006 RD830963010 RCM versus GCM Simulated Heat Wave Days OBS PCM PGR HAD HGR GCM Simulated RCM Downscaled RCM Downscaling Sensitivity to Cumulus Schemes RCM Ensemble Cumulus Prediction MKF MGR OBS EOP AirAir QualityQuality ModelModel ResultsResults The following examples show • the difference between AQM and CTM in simulating the present ozone and projecting future changes • the effect of the RCM climate sensitivity on AQM simulations and projections • the AQM ozone projections for 2100 under different emissions scenarios and incorporating climate changes 2003-2006 RD830963010 CTMCTM andand AQMAQM simulatedsimulated O3O3 diurnaldiurnal cyclecycle Driven by R-2 Driven by PCM 100 80 60 40 Northeast 20 100 80 60 Midwest 40 20 100 80 60 40 California 20 100 80 60 Texas 40 20 048121620240 4 8 12162024 2003-2006 OBS AQM MOZART AQM+LBC RD830963010 CTMCTM andand AQMAQM simulatedsimulated O3O3 episodesepisodes Episodes Comp Episodes Comp 35 100 30 (ppb) 3 25 80 20 60 15 10 40 5 20 0 -5 Daily max 8-hour O max 8-hour Daily 123456 0 123456 mean (ppb) 5-yrs from Departure OBS AQM MOZART 2003-2006 RD830963010 Improvements of model chemistry and physical parameterizations Comparisons with the observed ozone in 1999 June – August Control Isop. chem. (SMOKE improved emis.) Isop. chem. O3 dry dep. and O3 dry improved dep. improved Isop. chem., Isop. chem., O3 dry dep., O3 dry dep., and Rc and Rc improved improved (MOZART emis.) ¾The improvement of the chemistry of isoprene nitrate reduces the model biases of ozone by 5 – 10 ppb over the eastern U.S. ¾The improvement of ozone dry deposition parameterization reduces the model biases of ozone by 5 – 10 ppb over the eastern U.S. ¾The modification of the critical Richardson number has little effects on ozone. Long-Range Transport Impact on Local-Regional AQ The chemical LBCs predicted by MOZART cause AQM to produce overall more O3 than assuming clean air boundaries. The effect is especially large in northwest U.S., where 10-18 ppb more O3 may result from the long-range transport. -8 0 8 16 24 PPB Long-range transport can exacerbate local and regional air quality problems by elevating the background and loading during episodes. OverallOverall RegionalRegional UncertaintyUncertainty The following examples show • the spread among models of Midwest temperature and ozone biases and future projections under various emissions scenarios • the sensitivity of ozone temperature- dependence to regional emissions changes 2003-2006 RD830963010 -2 0 2 4 6 8 Regional Temperature BiasandFutureProjection PCM PGR Present PKF Bias HAD HGR HKF B1 PCM PGR PKF B2 HAD HGR HKF 2050 HAD Midwest A2 HGR HKF A1Fi PCM PGR Future projection PKF PCM B1 PGR PKF HAD B2 HGR HKF 2100 HAD A2 HGR HKF A1Fi PCM PGR PKF Midwest Bias Future projection 30 Present 2050 2100 20 10 0 -10 -20 AQPKF AQPKF AQPKF AQPKF AQPKF AQPGR AQPGR AQPGR AQPGR AQPGR AQHGR AQHGR AQHGR MZPCM B1 B2 A2 A1Fi B1 B2 A2 A1Fi U.S. Midwest regional ozone biases and future projections by GCMs (PCM, HAD) and their downscaling RCMs (GR, KF) Daily Mean [O3] 60 Northeast Midwest 40 20 0 -20 -40 2000 2050 2100 2000 2050 2100 60 California Texas 40 20 0 -20 -40 Relative change (%) Relative 2000 2050 2100 2000 2050 2100 60 Southeast 40 RCM/PCM A1Fi RKF/PCM A1Fi 20 RCM/PCM B1 RKF/PCM B1 0 RCM/HAD A2 RCM/HAD B2 -20 MOZART A1Fi -40 MOZART B1 2003-2006 2000 2050 2100 RD830963010 Link between Temperature and Ozone Changes 30 25 Northeast Midwest 20 California Texas 15 Southeast 10 5 0 -5 Daily mean ozone trend [ppb] Daily mean ozone trend -10 {PGR,PKF}[A1Fi,B1](2050,2100) {HGR}[A2,B2](2100) -15 01234567 Temperature trend [ºC] U.S. background O3 and contributions from nearby/remote sources 1999 June – August Mean in the Surface Layer MOZART SMOKE emissions emissions Canada + Europe Mexico + Asia ¾Current U.S. ozone background ranges from 20 – 40 ppb in summer. ¾Anthropogenic emissions from Canada/Mexico contribute up to 13 ppb of ozone near the political boundaries; there is little effect in the central and southeast U.S. ¾Anthropogenic emissions from Europe/Asia contribute up to 3 ppb of ozone over the northwest; there is little effect in the central U.S. Effect of Including Canadian and Mexican Emissions on Simulated 8-hour Maximum Ozone Simulated by regional climate-air quality modeling system (RCM+SMOKE+AQM) Biogenic Emissions Dependent on Climate Present 2050 2100 2.0 2.0 GR scheme GR scheme 1.5 1.5 1.0 1.0 0.5 0.5 PCM_A1Fi HadCM_A2 HadCM_B2 PCM_B1 PCM_A1Fi HadCM_A2 HadCM_B2 PCM_B1 2.0 2.0 KF scheme KF scheme 1.5 1.5 Fraction ofFraction present emissions 1.0 1.0 0.5 0.5 PCM_A1Fi HadCM_A2 HadCM_B2 PCM_B1 PCM_A1Fi HadCM_A2 HadCM_B2 PCM_B1 1.5 1.5 1.0 1.0 Fraction of GR emissions PCM_A1Fi HadCM_A2 HadCM_B2 PCM_B1 PCM_A1Fi HadCM_A2 HadCM_B2 PCM_B1 Isoprene Monoterpene Climate/emission change effects on U.S. regional ozone air quality June – August Mean in the Surface Layer ∆O3 B1 Current climate ∆O3 A1fi ∆O3 A1fi climate climate+ emissions ¾Current ozone levels range from 30 – 70 ppb over the U.S. ¾Climate change (only) increase ozone in many inland areas largely due to the increased air temperature; the decrease near the coasts are a result of changes in marine air influence. ¾Effects of emission changes, if included, dominate the climate change effects. Relative Contributions of Emissions & Climate Changes (a) A1Fi (b) B1 Change (ppb) 3 O (c) (d) MET vs EMS Summer average daily maximum 8-hour surface ozone concentration changes (ppb) between 2050 and 1998: (a) A1Fi scenario and (b) B1 scenario; and relative contributions (%) of the projected emissions (EMS) and climate (MET) changes to total surface O3 concentration trends between 2050 and 1998: (c) A1Fi scenario and (d) B1 scenario. The color code 1 (green) denotes for the dominance of the EMS effect (contribution > 70%), 2 (yellow) for that of the MET effect (contribution > 70%), and 3 (red) for both effects to be comparable. ConclusionsConclusions [1]Precipitation processes modulate surface temperature and are closely associated with moist convection, thunderstorm activity, cloud formation, and boundary layer development. Collectively they control regional-local variation of air quality processes. We have demonstrated that the RCM downscaling substantially improves precipitation prediction (Liang et al. 2004a-b, Hayhoe et al. 2006, Zhu and Liang 2005, 2006) by capturing the dominant regional and local forcing processes. [2]The coarse grid spacing and incomplete physics representation cause substantial GCM biases and inter-model differences at regional-local scales (Kunkel and Liang 2005, Kunkel et al.