Trends and variability in surface ozone over the United States

Abstract

We investigate the observed trends and interannual variability in surface ozone over the United States using the Global Modeling Initiative chemical transport model. We discuss the roles of meteorology, emissions, and transport from the stratosphere in driving the interannual variability in different regions and seasons. We demonstrate that a hindcast simulation for 1991-2010 can reproduce much of the observed variability and the trends in summertime ozone, with correlation coefficients for seasonally and regionally averaged median ozone ranging from 0.46 to 0.89. Reproducing the interannual variability in winter and spring in the western United States may require higher-resolution models to adequately represent stratosphere-troposphere exchange. Hindcast simulations with fixed versus variable emissions show that changes in anthropogenic emissions drive the observed negative trends in monthly median ozone concentrations in the eastern United States during summer, as well as the observed reduction in the amplitude of the seasonal cycle. The simulation underestimates positive trends in the western United States during spring, but excluding the first 4 years of data removes many of the statistically significant trends in this region. The reduction in the slope of the ozone versus temperature relationship before and after major emission reductions is also well represented by the model. Our results indicate that a global model can reproduce many of the important features of the meteorologically induced ozone variability as well as the emission-driven trends, lending confidence to model projections of future changes in regional surface ozone.

Tropospheric Ozone Assessment Report: Assessment of global-scale model performance for global and regional ozone distributions, variability, and trends

“…For the U.S, the positive bias ranges from 5-15 nmol mol -1 over the eastern U.S. but exceeds 15 nmol mol -1 over North American coastal regions compared with observations. The mean seasonal cycle is generally reproduced with correlation coefficients for monthly mean values greater than 0.6 (Figure 6d), although the multi-model mean tends to peak later in the year over the eastern U.S., consistent with previous model evaluations ( Murazaki and Hess 2006;Fiore et al, 2009;Reidmiller et al, 2009;Lamarque et al, 2012;Naik et al, 2013;BrownSteiner et al, 2015;Strode et al, 2015;Travis et al, 2016). In Europe, the annual multi-model mean performs better over northern Europe (biases range from -2 to +10 nmol mol .…”

“…Pinatubo (after Lin et al, 2015b). Similar hindcast simulations with a different model (but with similar chemistry scheme) captured some of the observed springtime western U.S. ozone interannual variability, but with a correlation coefficient of 0.4 (Strode et al, 2015).…”

“…Model skill in simulating mean surface ozone distribution varies by the type of model and simulations, evaluation approach (e.g., individual sites versus regional averages), and the availability and quality of observations (Fiore et al, 2009;Reidmiller et al, 2009;Lamarque et al, 2012;Doherty et al, 2013;Naik et al, 2013;Strode et al, 2015;Brown-Steiner et al, 2015;Monks et al, 2015;Schnell et al, 2015;Colette et al, 2015;Tilmes et al, 2015). Overall, current generation global CCMs reproduce the spatial patterns in annual mean surface ozone based on available observations but are generally biased high in the NH and, except for eastern Australia and New Zealand, biased low in the SH.…”

“…CTMs and SD-CCMs (and nudged CCMs) are often used for performing process-oriented analysis, including interpretation of short-term field measurements (e.g., Law et al, 1998;Liang et al, 2007;Zhang et al, 2008;Telford et al, 2010;Lin et al, 2012a;Wespes et al, 2012) and understanding the causes of ozone variability and long-term trends in observational records, by isolating the roles of emissions and meteorology (Koumoutsaris and Bey, 2012;Lin et al, 2014Lin et al, , 2015Lin et al, , 2017Strode et al, 2015). These models are also used to make chemical forecasts as part of flight planning for field missions (e.g., Fast et al, 2007).…”

“…A small number of CCM studies have evaluated tropospheric ozone and STT using simulations that include interactive stratospheric and tropospheric chemistry and aerosols (e.g., Neu et al, 2014;Lin et al, 2015a,b;Strode et al, 2015). A recent study applied observational constraints from the Tropospheric Emission Spectrometer (TES) and Microwave Limb Sounder (MLS) on-board NASA's Aura satellite over the 2005-2010 period to assess the ability of one CCM to reproduce the tropospheric ozone response to stratospheric circulation changes (Neu et al, 2014).…”