2014 Progress Report: Air Quality Impacts of Extreme Weather Events: Historical Analysis and Future ProjectionEPA Grant Number: R835204
Title: Air Quality Impacts of Extreme Weather Events: Historical Analysis and Future Projection
Investigators: Wang, Yuhang , Deng, Yi , Zhang, Henian
Current Investigators: Wang, Yuhang , Deng, Yi , Loadholt, Jay , Park, Taewon , Song, Yongjia , Zeng, Tao , Zhang, Henian , Zhang, Yuzhong , Zou, Yufei
Institution: Georgia Institute of Technology
Current Institution: Georgia Institute of Technology , Georgia Environmental Protection Division
EPA Project Officer: Chung, Serena
Project Period: June 1, 2012 through May 31, 2015 (Extended to May 31, 2016)
Project Period Covered by this Report: June 1, 2014 through May 31,2015
Project Amount: $749,859
RFA: Extreme Event Impacts on Air Quality and Water Quality with a Changing Global Climate (2011) RFA Text | Recipients Lists
Research Category: Air Quality and Air Toxics , Water Quality , Climate Change , Air , Water
We have identified major research areas in the first 2 years of this project. The focus of this year’s research effort is two fold: (1) more in depth and detailed work to substantiate the major findings of the research, and (2) organize the research results for publications. We recognized that many results that we will want to publish do not conform to some “norms” of the air quality-climate research community. Part of the reason is that extreme air quality and weather events all differ greatly in their physical, chemical, and dynamical processes, which demands new approaches and these new analyses often lead to new conclusions. In anticipation of potential objections in the review process, we have worked very hard this year to build strong evidence to support our findings. This process inevitably delayed the submission of the journal papers we had planned. Because major journals require that research findings are not announced before paper publications, we also limited meeting presentations, which will resume after these papers are submitted.
(1) Relationship of extreme air quality and extreme weather events
Here we explore the relationships between extreme weather and air quality events. The major issue we have been dealing with is that the definitions of weather extremes by meteorologists and climatologists vary greatly depending on the problems they are investigating. These definitions also differ greatly for the usual air quality extremes that are commonly known. In this year, we worked to distinguish clearly single day extremes from multi-day extremes and to examine how air quality and weather extremes, when defined in a consistent manner, relate to one another. The seasonal changes also are accounted for.
(2) Statistical analysis of ozone extremes in summer
Summer is the season where clear ozone decrease trends are seen in various regions of the United States. Here we examine the shape change of ozone distribution and how it relates to low and high extremes (as well as non-attainment of the ozone standard). We show that, for all regions, we see a consistent shape change in the last 30 years and its implication for a “natural range” of ozone. Here “natural” means the bound of ozone range likely in nature given the fact that anthropogenic or natural emissions cannot go to zero (which is usually done in model simulations). This will provide help inform whether potentially new EPA standards, which will probably be lower than 70 ppbv, can be realistically achieved. At some point, we probably need to separate regulatory standards from a public advisory standard (such as 50 ppbv ozone). The effort this year is to try to understand how we can attribute the changes on the high (and low) extremes to emissions and climate. It is very difficult based on statistics, but we think we now have a path forward.
(3) Climate attributions of summer ozone and PM2.5 trends
We use advanced EOF analysis to link ozone and PM2.5 variations to regional circulation patterns. The method we used targets the trend rather than the regional patterns targeted by standard EOF analysis. This year, we changed the original analysis from 1980-2012 to 1980-2014 and started working on similar analyses for the spring and fall. We show in this analysis that recent climate change is a significant contributor to ozone decreases (although not as much for PM2.5) in the summer.
(4) Cluster analysis and SOA in the Southeast
We applied cluster analysis to group ozone and PM2.5 observations into coherent regions of the United States. Initially we wanted to do this, so we can evaluate model simulations more effectively. Now, we find a more interesting and significant problem. SOA in the summer currently is believed to be the largest in the Southeast due to isoprene emissions. We then asked if we could use these self-coherent cluster regions to demonstrate that the observations of OC and EC in the past decade show more of an SOA signal in the Southeast than in other regions. To our surprise, we could not see a clear difference. The Southeast is not that different from the Northwest, where there are very little isoprene emissions. My colleagues at Georgia Tech convinced me that AMS data did show the summertime SOA formation. We have been working very hard this year to figure out how we can reanalyze the data to align with the up-to-date SOA mechanisms (i.e., dependence on sulfate and nitrate for example). We now have a quantitative measure to determine if the SE SOA is different from other regions in the United States on the basis of the long-term EPA observations.
(5) Validity of ozone-temperature relationship
Many air quality-climate studies have focused on ozone-temperature relationship and how it can be applied to climate projections. However, we find that vapor pressure deficit (VPD), commonly used by ecologists to quantify water stress of plants, also correlates with ozone. In October, it correlated much better with ozone than temperature. The apparent contradiction is reconciled when we applied a statistical covariance analysis to separate ozone relationship to T, VPD, and the correlation of T and VPD. We show that ozone-T relationship is strongest in summer, but in the Southeast summer, almost all of the relationship is due to correlated T and VPD (not T itself). In other regions in the summer, ozone is more correlated with T itself. In October in all regions, however, ozone does not show a dependence on T itself; any apparent correlation is due to correlated T and VPD. In the Southeast in October, ozone-VPD correlation is even larger than that due to correlated T and VPD.
(6) Extension of the ozone season to the fall in the Southeast
Despite the large emissions reduction of NOx and VOCs in the United States, October 2010 ozone concentrations in the Southeast are higher than July 2010, reaching the 30-year ozone average in July. In that sense, the ozone season was extended to October in 2010. We applied regional and climate model simulations to understand the reasons for extreme ozone in October 2010 and its implications for ozone extension into the fall in the future. We show that the extreme is due to VPD-driven ozone increases; the mechanism is due to an increase of isoprene emissions when plants are under water stress. Model simulations for 2008, 2009, and 2010 are applied to support our findings. Looking at GISS and GFDL CMIP5 projections, the extension into the fall will not occur on the basis of climate mean state. However, the frequencies of VPD extremes in the Southeast in October will likely increase, effectively increasing the occurrence frequency of high ozone in October and providing support that the ozone season likely will extend into the fall season in the Southeast in the future due to climate change. The detailed regional model simulations for 2008 and 2009, and additional model evaluation were conducted this year.
(7) Feasibility of long-range forecast of ozone and PM2.5 in the United States
We applied the linear inverse model to investigate the potential of long-range (1 month) forecast of ozone and PM2.5 in the United States. We show on a hindcast basis that July ozone and PM2.5 can be reasonably well predicted at the end of June in the last 5 years (the limitation is that we need a long record of observations and the PM2.5 data record is short). The analysis suggests that with reliable ensemble long-range meteorological forecasts, air quality forecasts also can be extended into long range, providing useful guidelines for longer-range planning. We worked on the PM2.5 forecast this year and extended all the analysis results to 2014.
(8) Extreme PM2.5 in January 2013 in China
Due to large emission reductions, extreme PM2.5 cases in the United States tend not to be regional or last for a long period of time. In contrast, PM2.5 episodes are frequent in China. The January 2013 case generated a lot of publicity after PM2.5 data tweeted by the U.S. Embassy in Beijing were picked up by the New York Times. We show that there is an extreme of climate-driven pollutant accumulation beyond any year since 1980 (i.e., not due to abnormal emissions in that month). We further show that the extreme was driven in part by melting ice in the Arctic and recent changes in Eurasia snow cover. The CESM model ensemble simulations were used to show the impact, which was done this year. We also looked at why CMIP5 simulations significantly underestimate the climate impact, i.e., the cobenefit of greenhouse gas control in China is much larger than estimated by the models. The work was primarily funded by an NSF project. A small portion of a student and my time is supported by this project this year.
(9) Uncertainty quantification using ensemble CESM simulations
We have completed an ensemble of 1x1 CESM simulations this year to examine the responses of extreme ozone and PM2.5 to weather/climate variations driven by atmospheric internal modes. The equivalence of 100-year simulations for the present atmosphere will be sufficient to quantify the related uncertainties in how weather/climate variations affect air quality extremes.
We have completed nearly all planned analysis tasks except the last one. Two of the research papers are completed and will be submitted to high-impact journals soon. Work is ongoing on several other papers. While most of the remaining work involves writing and revising the papers, we do anticipate some additional data and modeling analyses to resolve any remaining research issues. If time permits, we will examine the observed air quality and weather extreme relationships in light of simulated uncertainties, through which we investigate how climate projections can be applied in air quality extreme projections.