Core version of a general overview talk presented (with some variations) at the Chinese Cultural University, and at National Taiwan University, National Central University and Academia Sinica, Taiwan, 15-19 December 2003.


Modelling Ozone over East Asia: Pollution, Chemical Weather and Climate

Oliver Wild

Frontier Research System for Global Change, Yokohama, Japan


I'd like to describe some aspects of the current state of research on ozone in the troposphere, focusing in particular on our understanding of the field as encapsulated in state-of-the-art computer models. I will focus on the principal environmental effects of ozone, namely air pollution, its regional variability, and its global impacts as represented by its effects on climate. I will focus on the East Asian region, both because it will be of greatest interest here in Taiwan, and because it is of particular scientific interest to me, which I shall describe in a moment.

Why are we interested in tropospheric ozone? Ozone in the troposphere was until recently thought to originate principally in the stratosphere, where it is formed by the photolysis of molecular oxygen, and to be carried down into the troposphere across the tropopause. However, it may also be formed in the troposphere directly by oxidation of carbon monoxide (CO) and hydrocarbons (HC) in the presence of nitrogen oxides (NOx). It is now realized that this is the principal source of ozone in the troposphere, although there is still some controversy over the relative importance of these sources. Ozone is an important constituent of photochemical smog, and is harmful to the health of humans and animals, and is known to cause damage to plants and crops even at relatively low levels of 30-40 ppb. As the principal source of short-lived OH radicals, it controls the oxidizing capacity of the troposphere, and hence directly affects the lifetimes of many other trace gases. In addition, it is a significant greenhouse gas and contributes to a global warming of climate. Finally, the abundance of ozone has increased by a factor of three over the past century, now clearly seen as a direct result of increased emissions of precursors due to human activities.

Why is East Asia important? The countries along the western shores of the Pacific contain almost 25% of the world's population, and have recently been experiencing a period of unprecedented economic and industrial growth that is expected to be continued into the foreseeable future. This has been accompanied by a large increase in emissions of trace gases and ozone precursors, particularly from China, which still has much lower per-capita emissions than most of the developed world. For NOx, Asian emissions have recently overtaken those of Europe and North America, and pollution from the region is expected to have a larger effect on the global environment in the future. A further reason for studying the region is that its more southerly location compared to Europe and North America lead to faster chemical processing of pollution, and greater deep convection lofts this higher into the atmosphere, where transport is faster and the ozone lifetime is longer. The coastal location of most of the major source regions in East Asia, close to the start of the Pacific storm track, make export of pollution from this region particularly efficient.

In this talk, I shall focus on ozone over East Asia. I plan to work backwards, first considering export of pollution from the region, its global impacts, and how these compare with the equivalent source regions of Europe and North America. I will then spend much of the time discussing regional ozone production and its variability over the region, using the Transport and Chemical Evolution over the Pacific (TRACE-P) measurement campaign as a case study to examine what we know about this, and how well we can simulate it in current models. This will also highlight the strengths and weaknesses of models, and I will suggest ways of improving the simulations in future. I will conclude by looking at pollutant import to the East Asian region from upwind sources over the Eurasian continent and beyond, and examining their impacts on air quality over Asia.

I first describe the tool I have used in these studies - the Frontier Research System for Global Change (FRSGC)/University of California, Irvine (UCI) global chemical transport model (CTM). This is an off-line model driven by meteorological fields from a global circulation model (GCM), either run in native mode or forced by assimilated meteorological measurements. I have implemented a tropospheric chemistry scheme, including HOx-O3-NOx and methane chemistry, and including treatment of the oxidation of higher hydrocarbons (NMHC). Emissions from surface fossil fuel and natural sources are provided from the EDGAR v2.0 datasets for 1990, supplemented by mid-tropospheric emissions from lightning and aircraft sources. Deposition at the surface and in clouds and rainfall are also provided. A simplified stratospheric chemistry is provided for ozone, allowing the stratospheric concentrations to vary in a realistic way, and ensuring that the flux across the tropopause is close to that estimated from observations of about 500 Tg per year. In the model runs described here, I've generally used wind fields from the ECMWF Integrated Forecast System (IFS) generated specifically for chemical modelling by the University of Oslo, either at T63 (1.9°) or at T21 (5.6°) resolution. For some of the earlier studies I have used climate model winds from the GISS GCM at 4°x 5° resolution.

A comparison of the annual cycles of CO and O3 at surface sites over Eurasia suggests that the model captures the general features of these trace gases, driven by dynamical and chemical processes, with both GISS and ECMWF meteorology. Most sites are remote from major sources, and show a spring peak, except for Zugspitze in Europe, which shows a summer maximum in ozone from surrounding European sources. Elsewhere a summer minimum is typical, as chemical lifetimes are shortest in this season, and air frequently originates from more southerly latitudes. The variability in concentrations is also well captured, providing additional support for the validity of the model.

We look first at the major industrial emission regions of the northern hemisphere, North America, Europe and East Asia, which dominate global surface emissions of NOx. These regions lie in a similar latitude belt where transport in the mid troposphere is dominated by westerly flow. Typical transport times between the regions are about a week, typically shorter than the chemical lifetime of ozone in the mid-troposphere of a few weeks. These regions may therefore directly influence each other, and pollution from one region may affect air quality over another. This intercontinental transport of ozone has been studied with the model by applying 10% changes to emissions over each region in turn, and then examining the different impacts.

We find that while the emission region is clearly the one most greatly affected, ozone from each source region is transported north and westwards, and may have significant impacts on both of the other two regions, and over the northern hemisphere as a whole. European sources have the smallest global impacts due to the more northerly location, slower chemical formation and less efficient transport processes, although they have a greater effect on the Arctic than those from the other regions. East Asian sources have the greatest impacts, as lifting of pollutants by convection and frontal systems is more efficient, and leads to impacts higher in the troposphere where the the chemical lifetime of ozone is longer. Ozone in this region has a greater impact on radiative forcing than that lower in the troposphere, suggesting the emissions from East Asia may play a greater role in global climate warming than those over the other regions. Considering the surface impacts on downwind continents, we find the greatest effects from North American sources over Europe, where relatively short-distance trans-Atlantic transport occurs in the lower and mid-troposphere, but that East Asian impacts on North America are also significant. The mean impacts of 2-3 ppb O3 are relatively small, but are sufficient to substantially affect air quality standards which are usually defined on a period of time exceeding some threshold concentration. The potential importance of intercontinental transport is thus large, and it merits further study; a number of measurement campaigns are in planning and underway to address these issues further.

I will next focus on regional ozone production over East Asia, and will use results from the NASA TRACE-P measurement campaign held over the Western Pacific region in Spring 2001 to illustrate and evaluate the model simulations. The campaign involved extensive in-situ and lidar measurements from two aircraft (a DC-8 and a P-3B), a network of ozonesondes over the Pacific region, a number of sites over China, Korea and Japan with both surface measurements and remote-sensing, and data from satellite instruments, in particular CO from MOPITT, NOx and O3 from GOME and O3 from TOMS.

I will first explore the variability of ozone in the springtime over East Asia by showing selected ozonesonde profiles over the TRACE-P period against model results. We see that ozone varies very widely over the region, with variable tropopause heights, strong stratospheric intrusions reaching into the mid-troposphere, layers of polluted air with high ozone from tropospheric sources, and clean or polluted boundary layer air. Despite this variability, we find that in most cases the profiles are reasonably well reproduced, with layers in appropriate places. By separating the tropospheric and stratospheric components of ozone in the model, we can attribute layers to tropospheric or stratospheric sources, which reveals some interesting features such as descent of intrusions to within 2 km of the surface over Taiwan. However, we can also see significant discrepancies in some places, such in the upper troposphere over Taiwan and Hong Kong where there are layers of high O3, probably from biomass burning sources over South and Southeast Asia, which are not captured.

To evaluate the model performance in more detail, we compare the altitude of the 150 ppb isopleth of O3 on the ozone soundings, representative of the tropopause height or of significant stratospheric intrusions below this, and compare with the model. In most cases the agreement is excellent, with the strong latitudinal variation in tropopause height clearly visible, and a relatively small spread indicating that the magnitude and timing of intrusions is generally captured well, with just a few exceptions at Tateno and Cheju, close to the tropopause break. The correlation is very good, 0.99, with a small bias of about 500 m in the CTM indicating that the tropopause is too low, although this is smaller than the vertical layer resolution of about 1 km at this altitude. This comparison demonstrates that the combination of high-quality ECMWF meteorology and proper treatment of stratospheric chemistry for O3 provides a good simulation of the impacts of the stratosphere on tropospheric ozone.

To evaluate regional production, we compare O3 below 800 hPa on the soundings with those from the model. The agreement here is rather less good, with a 9% overestimate of the slope and a 12 ppb bias too high, despite a similar level of variability. Cleaner sites such as Hilo and Cheju are well reproduced, but sites closer to source regions such as Tateno and Taipei are clearly overestimated by as much as 20 ppb. This is a consequence of the relatively course model resolution used, and the consequent assumption that concentrations are well-mixed within a model grid box; however, it may also suggest that the timescales for O3 production in polluted conditions is too short.

We make a more detailed comparison of ozone profiles by comparing with the upward and downward-pointing lidar profiles from the DC-8 aircraft following its flight track. On flight 5, between Hawaii and Guam, the aircraft intercepted a large plume of ozone from East Asia at a low altitude over the Pacific, sandwiched between a clean marine boundary layer and cleaner subsiding air in the mid-troposphere. This plume became separated from the main westward flow of pollution across the Pacific by the action of a strong frontal system, and the plume was recirculated at more southerly latitudes, stagnating in a region of high pressure. These features are well reproduced by the model, and allow us to identify the source and evolution of this plume in greater detail.

On flight 18, returning back across the Pacific from Tokyo to Hawaii, the profiles look entirely different. The tropopause was much lower, about 10 km, and the aircraft intercepted two frontal systems. In each case these was substantial stratospheric influence behind the cold front, and cleaner conditions ahead; between the fronts was a region substantially influenced by Asian emissions. Ahead of the more mature front the air was principally of clean, Pacific origin, and a number of stratospheric intrusions are clearly visible and are well reproduced by the model. Substantial cloud cover in the vicinity of the fronts, clearly visible by the absence of lidar data, are also reproduced in the ECMWF meteorology.

How well can does the model capture the meteorological variability seen here? Springtime over East Asia is characterized by the sequential cyclogenesis and the frequent passage of frontal systems, and the CTM needs to capture the general features of these systems to correctly model the outflow and evolution of ozone. The visible image from the GOES satellite on March 4, 2001, shows a large swirl of cloud over the Western Pacific associated with a frontal system extending from the north of Japan down to Taiwan. The cloud cover and humidity associated with the warm sector of this system is clearly picked out in the ECMWF data, and the differing pre- and post-frontal flow patterns in the surface wind can readily be distinguished. Considering the distributions of O3 and CO, we find that there is strong continental outflow at low altitudes in the cold post-frontal flow, with significant ozone production throughout the region, and a strong banding of high CO behind the cold front. In the upper troposphere, we find ascent of CO in the rising warm conveyor belt ahead of the cold front, bringing Asian CO into the prevailing westerly flow, and strong descent behind the cold front, bringing high levels of stratospheric ozone into the troposphere. Convective outflow of CO from southeast Asian biomass burning sources is also clearly visible. These features provide a good illustration of the major pathways controlling export of Asian pollution to the global troposphere. The variability seen here leads to the concept of "Chemical Weather" by analogy with the meteorological variability with which we are more familiar.

The instantaneous ozone production rate along the aircraft flight tracks cannot be measured directly, but can be derived by running a photochemical steady-state box model driven by the aircraft observations. This can then be compared with ozone production along the flight tracks derived from the CTM. This novel comparison allows a better assessment of the strengths and weaknesses of the CTM simulations. We find that the net destruction in the marine boundary layer south of 27$deg;N is well reproduced, as is the rapid production over Japan. We find significant additional ozone production throughout the upper troposphere, but find that this is somewhat lower than in the measurement-derived values. This suggests that either very variable sources such as biomass burning are poorly represented, or that the chemical timescales for ozone production in the model are too short. Comparison of a wide range of species with the in-situ aircraft measurements suggests that while CO is reasonably well modelled, NOx is too low. This provides additional support for the hypothesis that O3 production is too fast in source regions, leading to reduced transport of NOx into the free troposphere, and hence less subsequent production. This shift in location of production is important for assessing the global impacts of regional emissions, and hence this discrepancy will be investigated further in future.

Finally, I show a number of dynamic figures examining the variation of ozone for a single week in March 2001. I first compare the model-derived total O3 column with that from the TOMS satellite instrument. For better representation of the variability and evolution of the column, I show the data from individual orbits of the satellite against the model-derived column at 2-hour intervals. The mean columns are well represented, although the model tends to over-predict columns at high latitudes and under-predict them at low latitudes, in common with previous stratospheric model simulations. The variability in the column is well represented, and allows the evolution of individual features in the TOMS columns to be explored in more detail. In a second movie, I show the variation in tropopause height against the TOMS O3 column, demonstrating how it matches up, and showing the lower tropopause at the leading edge of regions with high total column. In a third movie, we study an elevation of the 35 ppb isopleth of stratospheric ozone, which now clearly shows how the intrusions into the troposphere match up with the columns seen by TOMS. We also demonstrate the greater importance of intrusions over the Eastern than the Western Pacific during this period, although large intrusions still affect East Asia. In a fourth movie, we show hoe the meteorology of the region affects the export of CO from East Asia; outflow is predominantly at low levels behind cold fronts, but there is also substantial lifting by both convection and warm conveyor belts, creating distinct pulses of pollution in the mid-troposphere which are rapidly transported eastwards. The concept of chemical weather is west illustrated here. Finally, we show the outflow and evolution of O3 from East Asia, showing many of the same features as CO, and compare it with the stratospheric intrusions viewed earlier. It is clear from this that stratospheric and tropospheric influences intertwine and ultimately mix over the eastern Pacific, revealing why the attribution of particular sources based on measurement and trajectory studies alone is often inadequate.

We conclude by looking at the influx of O3 and CO to East Asia from sources upwind over Eurasia. Earlier trajectory-based studies have clearly demonstrated that much of the air arriving at the 100E meridian, which may be considered the western boundary of East Asia, has crossed Europe, and may therefore be influenced by European emissions. We go one stage further, and demonstrate that not only European but also North American O3 makes a significant contribution to air arriving over East Asia. Boundary layer outflow from Europe typically occurs at high latitudes, particularly in the winter when the Siberian High causes a northward deviation of flow over Central Asia. Much of this flow turns southwest-wards over East Asia, bringing European air to Japan. We find the maximum impact from European sources on CO over Japan occurs in February and March in the boundary layer; for O3 the greatest impacts are in April-May in the mid troposphere. Interestingly, we find that North American impacts on O3 are greater than those of Europe, particularly in the free troposphere. We find that this is principally due to efficient lifting of North American pollution by frontal systems over the western Atlantic; European impacts, by contrast, are dominated by outflow in the boundary layer, where the chemical lifetime of ozone is rather shorter. While the impacts of these sources on background ozone over East Asia is not large, it will be expected to affect air quality in the region in spring.

In summary, I hope that I have provided some insight into the features and characteristics of ozone over East Asia, and into some of the successes and failures in current attempts to model them. The location and meteorology of the region make it an important but challenging region to study. Continued improvement in our understanding is expected through the development of models and their application to periods when comprehensive measurement date is available. I would like to stress here the importance of high-quality measurements in providing a "ground truth" with which models can compare, and in highlighting deficiencies in current understanding that will lead to better models in future. A number of major challenges remain for modelling tropospheric ozone, in particular the need to improve treatment of very variable or episodic emission sources such as biomass burning, and the need to consider the impacts of aerosol particles such as dust, soot and sulphate particles, which may substantially effect the chemistry of ozone through both direct and photolytic processes. More detailed treatment of the feedbacks between the ozone and climate, and between ozone and the biosphere, are also required.

Looking to the future, I would like to highlight the advances that may be expected with improved computing power, which will allow improvements in understanding to be made through more detailed treatment of the fundamental processes affecting ozone. The Japanese government's Earth Simulator Project was finally completed 2 years ago, and we now have access to a powerful supercomputer (ranked first in the semi-annual Top 500 supercomputer listings for the fourth time in Nov 2003) which is dedicated entirely to atmosphere, ocean and solid earth research. This computer will provide better simulations of tropospheric ozone by allowing higher resolutions and more comprehensive chemistry, bringing the spatial scales involved closer to those required by the strongly non-linear chemistry of polluted emission regions. This will ultimately improve confidence in our understanding of tropospheric chemistry, and will allow more confident development of the simplified chemistry schemes that will be required in the next generation of Earth System models designed to study climate and global change.



Oliver Wild   (Frontier Research System for Global Change)
30 Dec 2003