GFDL - Geophysical Fluid Dynamics Laboratory

Skip to main content

Arlene's Research Page

My research applies global tropospheric chemistry models in conjunction with observations to investigate interactions among pollutant emissions (both natural and anthropogenic), regional air pollution, global atmospheric chemistry, and climate.

Please scroll down for descriptions of specific research areas.

MITIGATING OZONE POLLUTION AND CLIMATE CHANGE VIA METHANE CONTROL

Earlier work in the GEOS-Chem model showed that methane emission controls reduce both global warming and air pollution via decreases in background tropospheric ozone [Fiore et al., 2002], and a preliminary cost-benefit analysis indicated that methane controls were feasible [West and Fiore, 2005]. More recently, we have conducted multi-decadal full chemistry transient simulations in the MOZART-2 tropospheric chemistry model to show that neither the air quality nor climate benefits depend strongly on the location of the methane emission reductions, implying that the lowest cost emission controls can be targeted. With a series of future (2005-2030) transient simulations, we demonstrate that cost-effective methane controls would offset the positive climate forcing from methane and ozone that would otherwise occur (from increases in nitrogen oxide and methane emissions in the baseline "CLE" scenario; Figure a) and improve ozone air quality (Figure b).

The figure shows (a) the adjusted radiative forcing (W m-2) in 2030 vs. 2005 from changes in tropospheric methane (blue) and ozone (red) calcualted with the GFDL AM2 radiative transfer model, and (b) the percentage of model grid-cell days in the GFDL MOZART-2 model with daily maximum 8-hour average ozone above 70 ppb in summer (June-July-August) over the United States and Europe, under the baseline scenario (CLE; global emissions of CH4, NOx, CO, and NMVOC change by +29%, +19%, -10% and +3%, respectively) and with decreases in anthropogenic CH4 emissions by 2030 of 75 (A), 125 (B; cost-effective with available technology), and 180 (C; requires development of new control technologies) Tg yr-1, and in a simulation with pre-industrial CH4 concentrations (700 ppb). Please see Fiore et al. [2008] for details.

INTERCONTINENTAL TRANSPORT OF POLLUTION

In recent years there has been a growing body of literature documenting evidence for air pollution transport from Asia to North America, North America to Europe, and Europe to Asia [e.g., TF HTAP, 2007 and earlier summaries in Holloway et al. [2003] and Fiore et al. [2003]. Understanding the surface ozone response over a "receptor" region to emission changes over a foreign "source" region is key to evaluating the potential gains from an international approach to abate ozone pollution. Under the Task Force on Hemispheric Transport of Air Pollution, we applied an ensemble of global and hemispheric chemical transport models to estimate the spatial average surface ozone response over east Asia (EA), Europe (EU), North America (NA), and south Asia (SA) to 20% decreases in anthropogenic emissions of ozone precursors, NOx, NMVOC, and CO (individually and combined), from each of these regions.

The figure above shows the decrease in monthly mean surface ozone over the receptor regions (one per panel) resulting from simultaneous 20% decreases in all anthropogenic ozone precursor emissions in the three foreign source regions combined (ALL; black) and individually: NA (red), EU (green), EA (dark blue), and SA (cyan). We also compared the role of 20% decreases in regional anthropogenic emissions of methane (blue bars in Figure below) with that of the "traditional" ozone precursors (NOx+NMVOC+CO; red bars in Figure below) that are subject to regulations to abate regional ozone pollution.


The top panel shows the influence of each source region on surface O3 within the same region (termed "domestic"). The bottom panel shows the sum of the surface O3 decreases to emission changes wtihin the three foreign source regions; these foreign contributions (and that from anthropogenic methane) are generally considered to be part of a region's "background" O3. For more details, please see Fiore et al. [2009]. This work represents one activity under the much larger TF HTAP effort aimed at estimating intercontinental transport of air pollution and uncertainties in those estimates; publications from this effort can be found here. In my graduate work [Fiore et al., 2002], we illustrated the potential for future increases in global emissions (including methane) to thwart domestic efforts to abate ozone pollution and extend the ozone pollution season into spring and fall. Specifically, the figure (below, right) shows the number of GEOS-Chem model grid-square days per month in the United States with afternoon (1-5 p.m. local time) ozone concentrations in surface air in excess of a 70 ppbv threshold for a 1995 base case simulation (black bars) and a simulation of the 2030 atmosphere based upon emissions projections from the IPCC [2001] A1 scenario (red bars). In the 2030 A1 simulation, global emissions of ozone precursors increase, but the distribution shifts from the developed world to the developing world. In the United States, emissions of ozone precursors decline by 20-40% relative to 1995. A 70 ppbv threshold is used here as a metric to gauge changes in the frequency of ozone pollution events in our simulations; results are similar for 60 and 80 ppbv thresholds.

Ongoing efforts include examining the impacts of climate change on intercontinental transport and placing those in the context of projected emission changes, as well as the potential for observations to help gauge which model estimates of intercontinental pollutant transport are most accurate.

 

UNCERTAINTIES IN ISOPRENE EMISSIONS AND CHEMISTRY: IMPLICATIONS FOR U.S. SURFACE OZONE

Uncertainties in isoprene emissions and chemistry confound efforts to quantitatively address the role of isoprene in contributing to surface ozone pollution, and how air quality will respond to a global change (e.g., changes in climate or land-use practices). The upper two panels in the figure show the standard isoprene and nitrogen oxide emissions for July 2001 in the GEOS-CHEM model for a 1x1 degree (latitude by longitude) simulation over North America. We implemented an isoprene inventory [Purves et al, Global Change Biology, 2004] (bottom left panel). This inventory is similar to the BEIS2 inventory used in many U.S. regional models. The difference in the two isoprene emission inventories yields local differences of up to 15 ppbv (the color scale saturates in the blue). Differences in the representation of organic isoprene nitrates (their yield and fate) in chemical mechanisms substantially influence estimates of surface ozone over the eastern United States, locally by up 12 ppbv, with the largest uncertainties in the high isoprene-emitting southeastern United States. During the summer 2004 ICARTT aircraft campaign over the eastern United States, isoprene and a suite of its oxidation products were measured, including total alkyl nitrates (which are dominated by isoprene nitrates over the eastern United States); these measurements provide constraints on some of the uncertainties in isoprene nitrate chemistry. More details are available in Fiore et al. [2005] and Horowitz et al. [2007] and on the GFDL tropospheric chemistry web page.

 

VARIABILITY IN ATMOSPHERIC COMPOSITION AND LINKS WITH CLIMATE: CH4 and O3

The figure at left [Fiore et al., 2006] shows the global annual mean methane abundances measured at the NOAA cooperative network of surface stations (black circles), and MOZART-2 simulations with constant methane emissions (red crosses), and time-varying anthropogenic (blue triangles) and wetland (green crosses) emissions. We find that the flattening of the methane trends post-1998 can be explained in the model by an increase in lightning NOx, by influencing OH concentrations (reaction with OH is the major sink for methane). This result highlights a potential feedback on the methane lifetime, in addition to other factors (temperature, humidity, photolysis rates, emissions from wetlands and fires) that may change with climate.

As an undergraduate researcher in Daniel Jacob's modeling group, I analyzed trends in summer afternoon ozone concentrations from 1980-1995 measured at EPA/AQS (formerly AIRS) stations. Cynthia Lin extended this work to include an analysis ofr trends in 8-hour and 1-hour average ozone concentrations, as well as trends in background ozone over the United States. More details are available in Fiore et al., 1998, Lin et al., 2000, and Lin et al. 2001.

Current work involves exploring the role of changes in climate on atmospheric composition and air quality with the newly developed GFDL CM3 model.
  

 

POLICY-RELEVANT BACKGROUND OZONE OVER THE UNITED STATES

Background ozone is considered by U.S. EPA when reviewing the scientific criteria upon which the national ozone air quality standard is based. The 2006 EPA Air Quality Criteria for Ozone and Related Photochemical Oxidants summarizes the current state of the science, and additional documents prepared during the review of the ozone standard, completed in 2008, are available here. This latest review resulted in a tightening of the National Ambient Air Quality Standard (NAAQS) for ozone from 84 to 75 ppbv.

In the most recent review of the NAAQS for ozone, "policy-relevant background" is defined to be the ozone concentrations that would be present if North American anthropogenic emissions of ozone precursors were turned off. Policy-relevant background thus consists of any naturally produced ozone, plus any ozone produced from anthropogenic precursor emissions outside of North America, that is present in surface air over the United States. A quantitative estimate of policy-relevant background is needed by EPA for two reasons: (1) to ensure that the standard is not set too close to background levels (since it must be attainable with domestic anthropogenic emission controls), and (2) to calculate the health risk associated with exposure to the increment of ozone above the background (i.e., exposure to ozone produced from North American ozone precursor sources). Since observations only exist for total ozone, models are needed to estimate the contribution to observed surface ozone that is produced from North American anthropogenic precursor emissions.

My graduate work involved estimating policy-relevant background ozone in surface air over the United States with the GEOS-CHEM model. In the past, the EPA used observational statistics to define a 25-45 ppbv range of surface ozone background during the U.S. ozone pollution season. A 40 ppbv level was then adopted for use in risk assessments. The figure shows our best estimate for the U.S. surface ozone background (green diamonds) that would exist if North American anthropogenic emissions were turned off. We classified ozone data from the Clean Air Status and Trends Network into low-lying sites (generally below 1.5 km) and elevated sites (above 1.5 km; all in the western U.S.). We then aggregated our results to construct the cumulative probability distributions shown in the figure, for the 58 surface sites and the 12 elevated sites, for the three seasons, for the observations (black asterisks) and the model (red triangles). The corresponding distribution of background ozone concentrations is shown as green diamonds. The figure indicates that an appropriate background for use in risk assessment should vary as a function of season, altitude, and total ozone level. In particular, the depletion of the background during high-ozone events should be taken into account as the 40 ppbv value previously used by EPA would underestimate the risk posed by ozone concentrations above the background under these circumstances. The highest observed ozone concentrations at all altitudes in all seasons are associated with pollution from North American anthropogenic emissions, as seen by the difference between the green diamonds and red triangles. The background ozone concentrations shown here include the contribution from hemispheric pollution and would be even lower if international emission controls could be considered as part of a broader strategy to improve U.S. air quality. Details can be found in Fiore et al. [2003].


APPLICATION OF EOFs to EVALUATE AIR QUALITY SIMULATIONS

As a graduate student, I used an empiricial orthogonal function analysis to evaluate and compare the global GEOS-CHEM model at both 4x5 and 2x2.5 degrees (latitude x longitude) horizontal resolution with observations from the EPA/AQS (formerly AIRS) network and with the regional Multiscale Air Quality SImulation Platform (MAQSIP) regional model, which was the prototype for the CMAQ model currently being used in regulatory applications. This work is described in Fiore et al., [2003]. The figure shows the first three Varimax-rotated empirical orthogonal functions (EOFs; left and middle columns) as derived from daily afternoon average (13-17 local time) surface ozone concentrations for the summer of 1995 in the observations averaged on the regional MAQSIP model grid (left) and in the MAQSIP model (middle). The percentage of total variance explained by each individual EOF is shown. The right column shows the principal components (daily variation in the strength of the EOFs; solid lines) derived from observations and the corresponding time series (dotted lines) obtained from projecting the simulated daily summer afternoon ozone concentrations onto the observationally-derived EOFs (left column). The east-west EOF identifies a longitudinal gradient as a major factor of variability for ozone in the eastern United States (upper panels). When the principal component time series (solid line; upper right panel) peaks, this east- west EOF is strongly expressed and ozone concentrations are elevated along the east coast (where the EOF is shown in red) and low in the central part of the country (where the EOF is shown in blue). The pattern is reversed when the time series is minimum. The second EOF highlights the industrial Midwest and the Northeast (middle panels) as a region where ozone concentrations vary coherently; the third EOF isolates the Southeast (bottom panels). The MAQSIP model reproduces all three EOFs (center column) and the associated time series (dotted lines in right column), indicating that it captures the synoptic-scale meteorological processes that modulate the observed day-to-day spatial and temporal variability in surface ozone concentrations.