Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection

4.2.4 Tropospheric O3

Tropospheric O3 is a direct greenhouse gas. The past increase in tropospheric O3 is estimated to provide the third largest increase in direct radiative forcing since the pre-industrial era. In addition, through its chemical impact on OH, it modifies the lifetimes of other greenhouse gases, such as CH4. Its budget, however, is much more difficult to derive than that of a long-lived gas for several reasons. Ozone abundances in the troposphere typically vary from less than 10 ppb over remote tropical oceans up to about 100 ppb in the upper troposphere, and often exceed 100 ppb downwind of polluted metropolitan regions. This variability, reflecting its rapid chemical turnover, makes it impossible to determine the tropospheric burden from the available surface sites, and we must rely on infrequent and sparsely sited profiles from ozone sondes (e.g., Logan, 1999). The total column of ozone is measured from satellites, and these observations have been used to infer the tropospheric ozone column after removing the much larger stratospheric column (e.g., Fishman and Brackett, 1997; Hudson and Thompson, 1998; Ziemke et al., 1998). The current burden of tropospheric O3 is about 370 Tg(O3), which is equivalent to a globally averaged column density of 34 DU (Dobson Units, 1 DU = 2.6871016 molecules/cm-2) or a mean abundance of about 50 ppb, see Table 4.9.

Table 4.9: Estimates of the change in tropospheric ozone since the pre-industrial era from various sources compared with the values recommended in this report.
Current climatology of tropospheric ozone (Park et al., 1999):
  Global mean tropospheric O3 : 34 DU = 370 Tg(O3) content in the Northern Hemisphere = 36 DU, in the Southern Hemisphere = 32 DU.
SAR recommendation:
  "50% of current Northern Hemisphere is anthropogenic" gives pre-industrial global mean content = 25 DU.
Increase = +9 DU
19th & early 20th century observations:a
  Assume Northern Hemisphere tropospheric ozone has increased uniformly by >30 ppb.
Increase = +10 to +13 DU
Survey of CTM simulated change from pre-industrial:b
  DU increase Model Reference
  9.6 UIO Berntsen et al. (1999)
  7.9 GFDL Haywood et al. (1998)
  8.9 MOZART-1 Hauglustaine et al. (1998)
  8.4 NCAR/2D Kiehl et al. (1999)
  9.5 GFDL-scaled Levy et al. (1997)
  12.0 Harvard/GISS Mickley et al. (1999)
  7.2 ECHAM4 Roelofs et al. (1997)
  8.7 UKMO Stevenson et al. (2000)
  8.0 MOGUNTIA VanDorland et al. (1997)
Increase = +7 to +12 DU (model range)
TAR recommendation:
  Pre-industrial era global mean tropospheric O 3 has increased from 25 DU to 34 DU. This increase, +9 DU, has a 67% likely range of 6 to 13 DU.
Increase = +9 DU (+6 to +13 DU)

The troposphere is defined as air with O3 <150 ppb, see Logan (1999). The Dobson Unit is 1 DU = 2.6871016 molecules of O3 per square centimetre; globally 1 DU = 10.9 Tg(O3) and 1 ppb of tropospheric O3 = 0.65 DU. The change in CH4 alone since pre-industrial conditions would give about +4 DU global increase in tropospheric O3 alone (see Table 4.11).

a Early observations are difficult to interpret and do not provide coverage needed to derive the tropospheric burden of O3 (see Harris et al., 1997). The change in burden is derived here by shifting tropospheric O3 uniformly in altitude to give 10 ppb at the surface in Northern Hemisphere mid-latitudes and 20 ppb at surface in Northern Hemisphere tropics (implies 10 DU), or by additionally reducing Southern Hemisphere tropics to 20 ppb and Southern Hemisphere mid-latitudes to 25 ppb at the surface (13 DU).
b From a survey of models by Hauglustaine and Solomon and Chapter 4. Except for Kiehl et al., these were all CTMs; they used widely varying assumptions about pre-industrial conditions for CH4, CO, NOx , and biomass burning and they did not all report consistent diagnostics.

The sources and sinks of tropospheric ozone are even more difficult to quantify than the burden. Influx of stratospheric air is a source of about 475 Tg(O3)/yr based on observed correlations with other gases (Murphy and Fahey, 1994; McLinden et al., 2000). The in situ photochemical sources are predicted to be many times larger, but are nearly balanced by equally large in situ chemical sinks (see discussion on CTM modelling of tropo-spheric O3 in Sections 4.4 and 4.5, Table 4.12). Photochemical production of ozone is tied to the abundance of pollutants and thus varies widely over a range of spatial scales, the most important of which (e.g., biomass burning plumes, urban plumes, aircraft corridors, and convective outflows) are not well represented in most global CTMs and cannot be quantified globally with regional models. The dominant photochemical sinks for tropospheric O3 are the catalytic destruction cycle involving the HO2 + O3 reaction and photolytic destruction by pathways involving the reaction of O(1D), a product of O3 photodissociation. The other large sink, comparable in magnitude to the stratospheric source, is surface loss mainly to vegetation. Another loss of O3 is observed under certain conditions in the polar marine boundary layer, notably at the end of Arctic winter. It indicates reactions involving halogen radicals and aerosols (Oum et al., 1998; Dickerson et al., 1999; Impey et al., 1999; Platt and Moortgat, 1999; Prados et al., 1999; Vogt et al., 1999). The contribution of these processes to the global budget is not yet quantified, but is probably small.

Atmospheric measurement campaigns, both at surface sites and with aircraft, have focused on simultaneous observations of the many chemical species involved in tropospheric O3 production. Primary areas of O3 production are the mid-latitude industrialised and tropical biomass burning regions. For example, the North Atlantic Regional Experiment (NARE) and the Atmosphere Ocean Chemistry Experiment (AEROCE) showed that the prevailing westerly winds typically carry large quantities of ozone and precursors from the eastern USA over the North Atlantic, reaching Bermuda and beyond (e.g., Dickerson et al., 1995; Penkett et al., 1998; Prados et al., 1999). The Pacific Exploratory Missions (PEM: Hoell et al., 1997, 1999) measured extensive plumes of pollution including ozone and its precursors downwind of eastern Asia. Convective transport of emissions from biomass burning affect the abundance of O3 in the mid- and upper troposphere (Pickering et al., 1996). Emissions by tropical fires in South America and southern Africa have been identified as the cause of enhanced O3 over the South Atlantic (Thompson et al., 1996), and the effects of biomass burning were seen in the remote South Pacific in PEM Tropics A (Schultz et al., 1999; Talbot et al ., 1999). Due to the widely varying chemical environments, these extensive studies provide a statistical sampling of conditions along with a critical test of the photochemistry in CTM simulations, but they do not provide an integrated budget for tropospheric O3. An example of such model-and-measurements study is given in the Section 4.2.6 discussion of tropospheric OH.

Figure 4.8: Mid-tropospheric O3 abundance (ppb) in northern mid-latitudes (36°N-59°N) for the years 1970 to 1996. Observations between 630 and 400 hPa are averaged from nine ozone sonde stations (four in North America, three in Europe, two in Japan), following the data analysis of Logan et al. (1999). Values are derived from the residuals of the trend fit with the trend added back to allow for discontinuities in the instruments. Monthly data (points) are shown with a smoothed 12-month-running mean (line).

Recent trends in global tropospheric O3 are extremely difficult to infer from the available measurements, while trends in the stratosphere are readily identified (Randel et al., 1999; WMO, 1999). With photochemistry producing local lifetimes as short as a few days in the boundary layer, the local measurement of tropo-spheric O3 does not reflect the abundance over the surrounding continent, and a surface measurement is not representative of the bulk of the troposphere above. Thus it is not contradictory for decadal trends in different atmospheric regions to be different, e.g., driven by the regional changes in pollutants, particularly NOx. Ozone sondes offer the best record of O3 throughout the troposphere, although measurements at many stations are made only weekly (infrequently for a variable gas like O3). Weekly continuous data since 1970 are available from only nine stations in the latitude range 36°N to 59°N (Logan et al., 1999; WMO, 1999). Different trends are seen at different locations for different periods. Most stations show an increase from 1970 to 1980, but no clear trend from 1980 to 1996. A composite record of the mid-tropospheric O3 abundance from 1970 to 1996 from the nine stations is taken from the analysis of Logan et al. (1999) and presented in Figure 4.8. There is no obvious linear increase in O3 abundance over this period, although the second half of this record (about 57 ppb) is clearly greater than the first half (about 53 ppb). Of the fourteen stations with records since 1980, only two, one in Japan and one in Europe, had statistically significant increases in mid-tropospheric O3 between 1980 and 1995. By contrast, the four Canadian stations, all at high latitudes, had significant decreases for the same time period. Surface O3 measurements from seventeen background stations also show no clear trend, even in the northern mid-latitudes (Oltmans et al., 1998; WMO, 1999). The largest negative trend in surface O3 was -0.7 ± 0.2%/yr at the South Pole (1975 to 1997), while the largest positive trend was +1.5 ± 0.5%/yr at Zugspitze, Germany (1978 to 1995). This ambiguous record of change over the past two decades may possibly be reconciled with the model predictions (see Section 4.4) of increasing tropospheric O3 driven regionally by increasing emissions of pollutants: the growth in NOx emissions is expected to have shifted from North America and Europe to Asia.

The change in tropospheric O3 since the pre-industrial era is even more difficult to evaluate on the basis of measurements alone. Since O3 is reactive, atmospheric abundances cannot be retrieved from ice cores. Recent evaluations of surface measurements in the 19th and early 20th century in Europe (Volz and Kley, 1988; Staehelin et al., 1994, 1998; Harris et al., 1997) indicate much lower O3 abundances than today, yet the scaling of these data to a tropospheric O3 burden, even for northern mid-latitudes, is not obvious. In the SAR, these data were used to make a rough estimate that O3 abundances in the Northern Hemispheric troposphere have doubled since the pre-industrial era. A similar difference, of 10 to 13 DU when globally averaged, is obtained using the climatology given by Park et al. (1999) for tropospheric O3 today and a parallel one with abundances adjusted to match the 19th century measurements in the Northern Hemisphere. CTM calculations predict that current anthropogenic emissions of NOx, CO, and VOC, as well as the increase in CH4 should have increased tropospheric O3 by a similar amount, primarily in the Northern Hemisphere. A recent survey of CTM studies gives global average increases ranging from 8 to 12 DU, although this small range does not adequately represent the uncertainty. These results are summarised in Table 4.9. Based on measurements, analyses, and models, the most likely increase in tropospheric O3 was 9 DU globally averaged, with a 67% confidence range of 6 to 13 DU. For some of the emissions scenarios considered here, tropospheric O3 is expected to increase even more in the 21st century as emissions of its precursors - NOx, CO and VOC - continue to grow (see Section 4.4).

Other reports in this collection