1. Impact of 2000-2050 climate change on PM 2.5 air quality inferred from a multi-model analysis of meteorological modes Loretta J. Mickley Co-Is: Amos P.K.A. Tai and Daniel J. Jacob School of Engineering and Applied Sciences Harvard University AQ Management Contacts: Susan Anenberg and Carey Jang, EPA/OAQPS 1 June 13-15, 2012

2. Climate change will likely affect PM2.5 concentrations. Models disagree on the sign and the magnitude of the impacts.  g m -3 Racherla and Adams, 2006 Pye et al., 2009 Response of sulfate PM 2.5 at the surface to 2000-2050 climate change. These model results are computationally expensive. How well do models capture variability in present-day PM 2.5 ? A2 A1 We need a simple tool that will allow AQ managers to readily calculate the climate consequences for PM2.5 air quality across a range of models and scenarios. 2

3. CMIP3 archive of daily meteorology: 15 IPCC models AQ response to climate change Apply observed relationships between PM 2.5 and met fields 3 AQ management tool Climate change over US PM 2.5 dependence on met variables Temperature ? ? ? Stagnation Relative humidity Precipitation Mixing depth The dependence of PM 2.5 on meteorological variables is complex. Different components have different sensitivities. Model projections have uncertainties.

4. Multiple linear regression coefficients for total PM 2.5 on meteorological variables. Units: μg m -3 D -1 (p-value < 0.05) Stagnation is strongly correlated with high PM 2.5. Mean PM 2.5 is 2.6 μg m -3 greater on a stagnant day Tai et al. 2010 Observed correlations of PM 2.5 with temperature and precipitation. 1998-2008 meteorology + EPA-AQS observations Increases in total PM 2.5 on a stagnant day vs. a non-stagnant day. 4

5. 5 Principal component analysis (PCA) of 8 meteorological variables identifies the dominant meteorological mode driving day-to-day PM 2.5 variability by region: Midwest, Jan 2006 R = -0.54 2 1 0 -2 6 3 0 -3 -6 PC Observed PM 2.5 (µg m -3 ) Transport modes for PM 2.5 :  Eastern US: mid-latitude cyclone a nd cold front passage  Pacific coast: synoptic-scale maritime inflow Jan 28 Jan 30 Tai et al., 2012 Dominant meteorological modes driving PM 2.5 variability.

6. Fluctuations in the period of the dominant meteorological modes can largely explain interannual variability of PM 2.5. Anomalies of annual mean PM 2.5 and period of dominant meteorological mode (cyclone passage) for US Midwest Annual mean PM 2.5 (µg m -3 ) Period Τ (d) Tai et al., 2012 R = 0.76 PM 2.5 cyclone period T  In each region, we identify the dominant meteorological mode whose mean period T is most strongly correlated with annual mean PM 2.5.  In the Midwest: sensitivity dPM 2.5 /dΤ = ~1 µg m -3 d -1

7. 2000-2050 climate change leads to increases in annual mean PM 2.5 across much of the Eastern US, but decreases across the West. Corresponding change in annual mean PM 2.5 concentrations  g m -3 7 We apply observed sensitivity dPM 2.5 /dΤ to model change in period  T in each grid box. There is large variation among model projections. Change in period T of dominant meteorological modes, weighted average for 15 models. day  T period, 2000-2050  PM2.5, 2000-2050 Increased stagnation Increased maritime inflow

8. 8 2000-2050 change in annual mean PM 2.5 (µg m -3 ) Likely responses:  Increase of ~0.1 µg m -3 in eastern US due to increased stagnation  Decrease of ~0.3 µg m -3 in Northwest due to more frequent maritime inflows Models disagree on the sign and magnitude of projected change in annual mean PM2.5, but some patterns emerge. Northeast Midwest Southeast Great Plains South-central Interior NW Interior SW Pacific NW California Eastern US Northwest

9. 9  Overall climate effect on annual PM 2.5 is likely to be less than ±0.5 µg m -3.  Effect of fires on PM 2.5 may be most important impact in future atmosphere, especially on a daily basis. Response of PM 2.5 to 2000-2050 climate change 2000-2050 change in annual mean PM 2.5 (µg m -3 ) Circulation Tai et al., this work Temperature Heald et al, 2008; Pye et al., 2009; Tai et al., 2012a Vegetation Wu et al., 2012 Wildfires Spracklen et al., 2009; Yue et al., 2012 East Northwest Southeast (OC) Southeast (nitrate) Midwest + West (OC) Northwest (OC + BC) Tai et al., 2012

10. 10 Tai, A.P.K., L.J. Mickley, D.J. Jacob, E.M. Leibensperger, L. Zhang, J.A. Fisher, and H.O.T. Pye, Meteorological modes of variability for fine particulate matter (PM 2.5 ) air quality the United States: implications for PM2.5 sensitivity, Atmos. Chem. Phys., 2012a. Tai, A. P. K., L. J. Mickley, and D. J. Jacob, Impact of 2000-2050 climate change on fine particulate matter (PM 2.5 ) air quality inferred from a multi-model analysis of meteorological modes, submitted to Atmos. Chem. Phys., 2012b. Next steps: Investigate health impacts of trends in PM 2.5 air quality and compare to impacts from heatwaves. Proposal submitted to NIH; PI is Francesca Dominici, Harvard. Develop similar tool for assessing climate impact on U.S. ozone air quality, across multiple models and scenarios.

11. 11

12. Multi-model Projection of Synoptic Period and PM 2.5 [Tai et al., in prep] Climatological observation of dPM 2.5 /dΤ dPM 2.5 /dΤ (µg m -3 d -1 ) ∆Τ (d) ∆PM 2.5 (µg m -3 ) Weighted average 2000-2050 change in T (15 IPCC AR4 GCMs) Resulting 2000-2050 change in PM 2.5 × =

13. Project Roadmap: 1.Identify the main meteorological modes controlling observed PM2.5 across the United States (Tai et al., 2010; 2011) 2.Calculate the sensitivity of PM2.5 to the frequency of the dominant meteorological mode. (Tai et al., 2011) Tai, A.P.K., L.J. Mickley, D.J. Jacob, E.M. Leibensperger, L. Zhang, J.A. Fisher, and H.O.T. Pye, Meteorological modes of variability for fine particulate matter (PM2.5) air quality the United States: implications for PM2.5 sensitivity to climate change, submitted to Atmos. Chem. Phys., 2011. 3.Track the changes in these modes using the IPCC AR4 archive of climate projections. 4.Estimate the change in surface PM2.5 concentrations due to climate penalty (or climate benefit). IPCC archive of daily meteorology AQ response to climate change Main meteorological modes driving observed PM2.5 13 AQ management tool

14. Evaluation of present-day meteorological modes in AR4 climate models reveals differences among models. Modeled (2 IPCC models) and observed (NCEP/NCAR) 1981-2000 time series of frequency of dominant meteorological mode for PM 2.5 in U.S. Midwest Frequency (d -1 )  Some models capture both the long-term mean and variability of meteorological mode frequency well.  As a first step, we use only those models that capture present-day mean and variability of frequency to predict future PM 2.5 N42° W87.5° 14 Observed model s