SEARCH

SEARCH BY CITATION

Keywords:

  • tropospheric ozone;
  • UV change

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[1] Model studies show that changes in photodissociation rates resulting from changes in ozone column densities induce changes in lower tropospheric ozone, which vary significantly with location and time of the year. The validity of the model results is tested against daily total ozone and ground level ozone at three selected stations (Samoa, Mauna Loa, and Hohenpeissenberg). Observational data for a period of more than 1 decade have been analyzed. Comparisons are made of model-simulated distribution of ozone and its precursors (NOx and CO) at the three stations. Further comparisons are made of observed and model-calculated sensitivity in surface ozone to reduction in ozone column densities. Calculations performed with a global-scale chemical transport model (CTM) with extensive ozone chemistry reproduce well the observed levels and seasonal distribution of NOx, CO, and ozone at remote background stations (Samoa and Mauna Loa) and at stations in more polluted regions (Hohenpeissenberg). A chemical box model is used to demonstrate the chemical link between surface ozone changes and changes in total ozone for different NOx levels. Model studies and analysis of the observational data show that ground level ozone at the remote, low-NOx stations of Mauna Loa and Samoa is correlated positively with total ozone, with an exception at Mauna Loa during winter months. A reduction in ozone column densities, which leads to enhanced photochemical activity, reduces ozone levels at ground level. The sensitivity of surface ozone to changes in total ozone is particularly large in the low-NOx regime at Samoa. An anticorrelation between ground level ozone and total ozone is found at the Hohenpeissenberg station both in the observational data and in the model results during wintertime with high NOx levels. Enhanced photochemical activity leads to enhanced ozone production. There is, however, a disagreement between the observed and CTM-modeled sensitivity in surface ozone to ozone column density during the summer months at Hohenpeissenberg. The strong anticorrelation found in the observations, giving increases in surface ozone at low ozone column densities, is not present in the CTM model studies. It is suggested that a correlation between low ozone column densities and stagnant high-pressure systems is an important cause for the observed anticorrelation.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[2] The decline in the ozone column densities that has been observed during the last two decades affects UVB radiation reaching the troposphere, thereby affecting the oxidation processes [Schnell et al., 1991; Fuglestvedt et al., 1994; Zerefos et al., 2002]. During the first years of the 1990s the ozone decline over the Antarctica accelerated, reaching exceptionally low ozone column densities during winter/spring seasons of 1992, 1993, 1995, 1996, and 1997 [Bojkov et al., 1993; Solomon, 1999]. There are clear indications that reduced ozone levels during the last two decades have led to enhanced flux of short-wave solar radiation. Observations of the UVB (290–320 nm) radiation in the past years both at northern and southern latitudes show significant increases concurrent with the reduced column ozone densities [World Meteorological Organization, 1995, 1999; Zerefos et al., 1997; Zerefos, 2002; Ziemke et al., 2000].

[3] The impact of changes in UV radiation on tropospheric oxidation and on tropospheric ozone trends has been studied in several modeling experiments. Liu and Trainer [1988] studied how tropospheric odd hydrogen and ozone responded to changes in solar UV fluxes. Madronich and Granier [1992], Bekki et al. [1994], and Fuglestvedt et al. [1994] used two-dimensional (2-D) models to calculate long-term OH and ozone changes during the 1980s and early 1990s from the observed reduction in ozone column densities. Similar calculations have been done with a 3-D model [Granier et al., 1996]. These studies suggest that stratospheric ozone depletion, leading to enhanced UVB penetration into the troposphere, has different impact on ozone, depending on season, location, and composition of the chemical compounds (e.g., NOx (nitrogen oxides), H2O, CO, volatile organic carbon (VOC), and CH4) affecting the nonlinear ozone chemistry [Solomon et al., 2003]. In polluted regions, ozone levels could be enhanced because of stratospheric ozone depletion impacting human health and the environment [Solomon et al., 2003].

[4] Although the effects of UV changes on tropospheric chemistry are theoretically well understood from model calculations, there is not enough observational evidence to support the model results. Stratospheric ozone is controlled by dynamical and chemical processes acting on different timescales, and hence the influence on surface UVB and related tropospheric ozone chemistry should also act on different timescales. Stratospheric ozone depletion could lead to tropospheric background ozone decreases either due to reduced influx from the stratosphere or due to enhanced in situ ozone loss. Schnell et al. [1991] reported a negative trend in surface ozone at the South Pole Station possibly induced by enhanced stratospheric ozone depletion. Taalas et al. [1997], analyzing ozonesonde data, concluded that springtime stratospheric ozone loss has a pronounced impact on upper tropospheric ozone (6–8 km) of both hemispheres (−12.8% in Antarctica and −10% in Arctic from 1988 to 1994). Observational evidence of changes in tropical tropospheric ozone associated with the stratospheric ozone changes on a timescale of an 11-year solar cycle has been identified using the tropospheric column ozone data derived from the Total Ozone Mapping Spectrometer satellite data [Chandra et al., 1999].

[5] On a daily basis, there is even less observational evidence due to the long lifetime of ozone and due to the fact that the photochemical link between total and surface ozone can easily be masked by transport effects especially in late winter and at high latitudes [Broennimann and Neu, 1998]. Only under sufficiently large changes of UVB (and total ozone column) and suitable meteorological conditions can such an influence be detected, as presented in a case study on Swiss mountains in late winter [Broennimann and Neu, 1998] and in a case study at Crete, Greece, during the Photochemical Activity and Solar Ultraviolet Radiation campaign [Zanis et al., 2002]. Broennimann et al. [2000] found also positive deviations of ozone peaks clearly connected with positive UVB deviations on a day-to-day basis using a 7 year time series of measurements at the Swiss Alps.

[6] The interaction between chemical processes leads to a strong nonlinearity in the tropospheric net ozone production. Crutzen [1979] showed that the transition between photochemical regimes of ozone formation and destruction in the troposphere is determined by the amount of NOx (NO + NO2) present. In remote background regions, with low NOx, net ozone loss occurs, while in tropospheric regions with high NOx, ozone production dominates. Several studies have demonstrated that ozone loss takes place in the remote unpolluted troposphere [e.g., Liu et al., 1983]. The anticorrelation found in diurnal and seasonal variations of O3 and H2O2 [Ayers et al., 1992] points to net ozone loss during photochemically active time periods. On the basis of observations of key chemical compounds, Ridley et al. [1992] estimated a net ozone loss of 0.5 ppb/d (∼1%/d) in the free troposphere near 3.4 km altitude at Mauna Loa.

[7] In this study we demonstrate through analysis of ozone data, large-scale 3-D CTM studies, and box model simulations that the short-term (days to weeks) variation in stratospheric ozone enforces significant changes in tropospheric photochemistry and in the ozone distribution. Observations from three selected surface stations are analyzed and compared with modeled distribution and changes: Hohenpeissenberg at midnorthern latitudes (48°N), Mauna Loa at low northern latitudes (19°N), and Samoa at low southern latitudes (14°S). Box model studies will be used to demonstrate the relation between atmospheric NOx distribution and ozone formation.

2. Tropospheric Ozone Production and Loss

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

2.1. Tropospheric Ozone Photochemistry and UVB Changes

[8] Stratospheric ozone changes affect tropospheric chemistry through the impact on UVB radiation penetrating into the troposphere. Although all chemical compounds dissociated in the troposphere by radiation in the UVB region are affected by stratospheric ozone changes (e.g., CH2O, H2O2, NO2, and aldehydes), the key reaction affected by stratospheric ozone changes is the photodissociation of ozone, yielding excited state molecular oxygen, leading to the formation of the hydroxyl radical [Fuglestvedt et al., 1994]:

  • equation image
  • equation image

Changes in OH formation are followed by changes in a large number of chemical reactions that can lead to either ozone loss or production, depending on the local chemical composition in the troposphere.

[9] The distribution of NOx in the troposphere is the key to ozone formation. In regions of high NOx, ozone formation occurs through the following sequence of reactions (see Figure 1):

  • equation image
  • equation image
  • equation image

In regions of the troposphere with low NOx, reactions (R1) and (R2) are followed by the reactions below to give ozone loss (see Figure 1):

  • equation image
  • equation image
image

Figure 1. Modeled sensitivity of the ratio (ΔO3/O3)/(ΔTO/TO) for varying NOx concentrations at Hohenpeissenberg for summer and winter conditions.

Download figure to PowerPoint

[10] There are a large number of additional components and chemical reactions included in the chemistry calculations done in the box model and 3-D model simulations presented here. These reactions will modify distributions of key compounds for the ozone chemistry and for the atmospheric oxidation potential. For instance, several other nitrogen oxide compounds are formed (e.g., PAN, N2O5, and HO2NO2) which affect NO and NO2 distributions and therefore the formation and destruction of ozone and species like H2O2 and OH. In the perturbation calculations with reduced ozone column densities, changes in UVB radiation are taken into account in photodissociation rates for all compounds affected by UVB radiation.

2.2. Box Model Runs of the Nonlinear Ozone Chemistry

[11] In addition to the global-scale Oslo CTM2 a photochemical box model has been used in this study to illustrate how the nonlinearity in tropospheric ozone chemistry is related to the NOx chemistry. We will, in particular, demonstrate the sensitivity of the net ozone production rate (production rate minus loss rate) to changes in the total ozone column as we move from a high-NOx environment (moderate to highly polluted regions) to a low-NOx environment (remote pristine regions). Comparisons of the calculated changes using the chemical box model with the observed changes at the three selected sites should give an indication of how the chemistry at the three sites is affected by changes in the ozone column.

[12] The chemical scheme of the box model is identical to the one used in the CTM and includes 51 chemical species in the standard O3-NOx-CO-CH4-NMVOC (nonmethane volatile organic carbon) families, 120 thermal, and 18 photolysis reactions. The box model was constrained by the monthly averaged chemical fields from the CTM at the three sites of interest. To obtain a realistic diurnal evolution, 3 hourly 1997 meteorological data from the CTM were used as input to the box model.

[13] The box model runs were carried out with starting date both in summer (1 July) and winter (1 January). The box model was run for 20 days, and the quantity (ΔO3/O3) was calculated for a 10% change of total ozone, the same perturbation as used in the 3-D CTM study. An example of the calculated changes in (ΔO3/O3)/(ΔTO/TO) (where TO is total ozone) for varying NOx levels for northern midlatitudes, spanning the transition from a net ozone production at high NOx levels to net ozone destruction at low NOx levels, is shown in Figure 1. There is, in general, a larger sensitivity in July when the photochemistry is more active. Furthermore, the transition from net ozone destruction to production occurs at a higher NOx level. The ozone sensitivity is also dependent on the initial values of VOCs and CO that also vary with season.

3. Analysis of Observed Surface and Total Ozone

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[14] The location of the stations used for analyzing total and ground ozone data are Mauna Loa at 20°N, 156°W, 3400 m above sea level (asl); Samoa at 14°S, 171°W, 20 m asl; and Hohenpeissenberg at 48°N, 11°E, 1000 m asl). These stations have been selected because they have long records of surface ozone and they cover different regimes of NOx values, which should demonstrate the dependence of tropospheric ozone perturbation on background NOx levels. Typical NOx levels encountered at the three stations are Samoa (5–100 ppt) [Bradshaw et al., 2000], Mauna Loa (20–140 ppt) [Hauglustaine et al., 1996], and Hohenpeissenberg (1–15 ppb) [Broennimann and Neu, 1998]. Publicly available daily total ozone values and daily mean ground level ozone values for the periods 1975–1992 (Samoa), 1973–1992 (Mauna Loa), and Hohenpeissenberg (1976–1995) have been used in this study (World Ozone Data Center, National Oceanic and Atmospheric Administration (NOAA)/Climate Monitoring and Diagnostics Laboratory (CMDL)). The Samoa and Mauna Loa data are maintained at NOAA CMDL and are available through ftp [Oltmans and Levy, 1994]. The analysis steps in this study were as follows: First, surface ozone and total ozone data were deseasonalized. For deseasonalizing the data, the monthly mean values were used instead of the daily mean values. The deseasonalization of the daily data, both surface and total ozone, was performed by subtracting the corresponding month's mean value. The reason for using monthly mean values is the lack of the daily mean values' homogeneity, since there were missing data. Second, the long-term trend was removed from the deseasonalized data. Lastly, the correlation coefficient between the deseasonalized, detrended time series of total and ground level ozone and the slope of the regression for each season and station was calculated. Results of the analysis for the three stations are depicted in Figure 2. We further define the tropospheric ozone sensitivity as the change in surface ozone (in percent) caused by a change in total ozone (in percent). The sensitivity of the three stations examined is given in Table 1. We find that ground level ozone is anticorrelated with total ozone at Hohenpeissenberg where NOx levels are several ppb, while it is positively correlated at Samoa where NOx levels are in the low-ppt range. At Mauna Loa, total ozone and ground level ozone are positively correlated during summer and fall when NOx levels are low. However, during winter and spring the observed correlation between total and ground level ozone was near zero. The tropospheric ozone sensitivity is clearly most pronounced at Samoa where reduction in total ozone gives reduced surface ozone values. For Mauna Loa the result is less distinct and could be a result of air masses being transported from regions with moderate pollution during these seasons, since the values of CO and most nonmethane hydrocarbons at Mauna Loa in winter and spring are higher than those in summer and fall [Greenberg et al., 1996].

image

Figure 2. Correlation coefficients and number of observations between daily total ozone and mean daily ground level ozone at Samoa, Mauna Loa, and Hohenpeissenberg for each season.

Download figure to PowerPoint

Table 1. Modeled Tropospheric Ozone Sensitivity (ΔO3/O3)/(ΔTO/TO) for Typical Atmospheric Conditions at Hohenpeissenberg, Mauna Loa, and Samoa During Summer and Winter
StationSensitivity (ΔO3/O3)/(ΔTO/TO) Modeled Oslo CTM2Sensitivity (ΔO3/O3)/(ΔTO/TO) ObservedSensitivity (ΔO3/O3)/(ΔTO/TO) Box Model Calculations
Hohenpeissenberg
   January−0.28−0.3−0.17
   July0.09−0.7−0.03
Mauna Loa
   January0.440.00.15
   July0.531.00.09
Samoa
   January1.663.70.48
   July0.792.20.24

[15] The above findings are mainly in agreement with the discussion in section 2 of the dependence of ozone production on NOx levels. The analysis shows that reductions in total ozone lead to reductions of ground level ozone at the more remote stations and increases of ground level ozone at stations affected by pollution. The observations are quite sensitive to transport processes not accounted for in the box model. In particular, this leads to higher sensitivity in clean air where exchange with the free tropospheric reduced ozone leads to increased reductions of ground level ozone. In the free troposphere over clean regions, there are even lower values of NOx than at the surface, and therefore a stronger ozone reduction takes place when the ozone column is reduced.

4. Model Studies of the Distribution and Changes in Ozone and its Precursors

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

4.1. Oslo CTM2

[16] The Oslo CTM2 is used to study the global distribution and seasonal variation in ozone and its precursors, NOx and CO. In particular, we look at how the seasonal distribution of ozone is affected by changes in the ozone column density at the three stations selected for analysis of the ozone column/surface ozone relations.

[17] The three-dimensional Oslo CTM2 is an off-line CTM that uses precalculated transport and physical fields to simulate chemical turnover and distribution in the atmosphere [Sundet, 1997; Jonson et al., 2000; Grini et al., 2002]. The current version of the model focuses on the troposphere, with the model domain reaching from pole to pole and from the ground up to 10 hPa. Horizontal and vertical resolutions are determined by the input data provided. In this paper a data set for 1997 with a resolution corresponding to T42 (2.8° × 2.8°) in the horizontal and with 40 levels in the vertical from the surface up to 10 hPa is used.

[18] Advection is done using the second-order moment method [Prather, 1986]. Convection in the CTM is calculated on the basis of mass fluxes precalculated with the Tiedtke mass flux scheme [Tiedtke, 1989], and vertical transport of species is determined by the surplus/deficit of mass flux in a column. Photodissociation rates are calculated online, following Wild et al. [2000], where clouds, both water and ice, particles, and surface albedo are included in the radiative flux calculations. Emissions are based upon the EDGAR 3.2 database [Olivier and Berdowski, 2001] representing anthropogenic emissions for 1995, and compilations by Müller [1992] are adopted for most natural emissions. These databases have been further refined and developed specifically to represent 1997 emissions for use in the Precursors of Ozone and Their Effects in the Troposphere project (For information on emission database see http://nadir.nilu.no/poet/background.htm). Deposition is based upon the Wesely [1989] scheme, and the boundary layer is treated according to the Holtslag K-profile scheme [Holtslag et al., 1990]. Influx of stratospheric ozone in the Oslo CTM2 has been evaluated with a linearized ozone approach based on tracer correlation.

[19] The distribution of chemically active species is calculated using a chemical scheme that integrates 51 components with ∼120 thermal and 18 photolysis reactions [Berntsen and Isaksen, 1999]. The time evolution is obtained by using the quasi–steady state approximation integrator where the mass conservation is controlled by adjusting NOx [Hesstvedt et al., 1978]. The time step used to integrate the chemical scheme is short enough to control the evolution of the chemical system in each grid box by making small adjustments of NOx. In the simulations done in this study the chemical time step was set to 15 min. The cycling within short-lived families had an even shorter time step.

[20] Influx of stratospheric ozone into the troposphere is controlled by the model transport. The concentration is set in the stratosphere, with values taken from the coupled stratosphere-troposphere CTM at T21 resolution [Gauss et al., 2003], and is updated every day during the whole integration. The transport data are evaluated by McLinden et al. [2000] and show that the integrated flux of ozone is realistic compared to measurements. Also the timing of the flux is good, as is shown by Wild et al. [2003].

[21] Studies of the ozone transport performed in the European Union (EU) project TOPOZ-2 project show that the model does a good job in describing processes in the upper troposphere and lower stratosphere compared to EU project MOZAIC data [Bregman et al., 2001]. A study of atmospheric sea salt shows that the processes taking place in the lower part of the troposphere (rainfall, winds, and boundary layer processes) are well suited for describing the sea salt mass distribution [Grini et al., 2002].

4.2. Calculated Distribution and Seasonal Variation

[22] A 15 month simulation was done with the Oslo CTM2 model using meteorological data for 1997. The run was initialized with chemical distributions from a previous run with a simulation time of 1 year. The calculated monthly average ozone, CO and NOx distributions, near the surface, are shown in Figures 3a, 3b, and 3c, respectively, for January and July. Both CO and NOx show large spatial variations, reflecting the variation in the source distribution strength. CO, with an atmospheric lifetime of 1–2 months in most of the troposphere, has a distribution, which is strongly modified by transport and in situ production from methane oxidation. There are also significant seasonal variations, with higher values during winter than summer in regions with large pollution emissions (middle and high northern latitudes) due to low OH values and thus slow oxidation. The nitrogen oxides distribution shows large regional variations that reflect the emission pattern due to the short lifetime of NOx. High NOx values are found over continents with extensive emissions from pollution sources (North America, Europe, and SE Asia). Regional-scale NOx concentrations range from a few parts per billion during summer to more than 10 ppb during winter when oxidation of NOx (NO2) is slow. In contrast, NOx levels are well below 0.1 ppb over large areas in remote regions far from the NOx sources.

image

Figure 3a. Monthly averaged (with the CTM2) surface layer ozone in ppb for (top) January and (bottom) July.

Download figure to PowerPoint

image

Figure 3b. Monthly averaged surface layer CO in ppb for (top) January and (bottom) July.

Download figure to PowerPoint

image

Figure 3c. Monthly averaged surface layer NOx in ppt for (top) January and (bottom) July.

Download figure to PowerPoint

[23] The net chemical production (NCP) of ozone in the boundary layer over the three stations is shown in Table 2a. NCP of ozone in the planetary boundary layer reflects the NOx distribution. Photochemical production over northern polluted regions is most efficient during summer months with lower values during winter (January). This is a typical feature at middle and high northern latitudes. This picture is in contrast to what is seen at middle and high southern latitudes, where extensive low values of NOx lead to minimum ozone values during summer (January) when the photochemical activity is high and the net photochemical ozone destruction is large. In tropical and subtropical regions not affected by emissions or transport of pollutants, ozone destruction occurs throughout the year, leading to very low ozone concentrations in the boundary layer.

Table 2a. Twenty-Four Hours of Total Ozone Net Chemical Production in the Planetary Boundary Layer Calculated in the Oslo CTM2 for 1 January and 1 Julya
 HohenpeissenbergMauna LoaSamoa
  • a

    Values are in ppb.

January0.70−0.64−1.70
July13.8−0.12−5.52

[24] Comparisons of modeled, monthly averaged surface ozone values with observations from 1997 (World Data Centre for Greenhouse Gases (WDCGG)) are shown in Figure 4a for the sites in discussion. For both Samoa and Hohenpeissenberg the seasonal variations over the year are well simulated in the model calculations. The model tends to slightly underestimate the values at Hohenpeissenberg in wintertime. For Mauna Loa (3400 m asl), except for the observed springtime maximum, the values calculated with the model match the observations reasonably well. The simulated late summer maximum over Hohenpeissenberg is significantly larger than the simulated maximum values over the two other stations.

image

Figure 4a. Comparison of measured and simulated (with the CTM2) monthly averaged surface ozone for Hohenpeissenberg, Mauna Loa, and Samoa.

Download figure to PowerPoint

[25] Comparisons of modeled CO for 1997 with observations for 1997 (WDCGG) are shown in Figure 4b for the three sites. For Mauna Loa the seasonal cycle is influenced by many factors [Rinsland et al., 1999; Novelli et al., 1998]: the seasonal cycle of OH, springtime transport of Asian pollution, timing of biomass burning in the northern tropics, and other sources. The model values are, in general, a little low, especially around the spring peak, which might by attributed to an underestimated transport of Asian pollution during this time of the year. The period September 1997 to October 1998 shows anomalously high values of CO, C2H6, and HCN over Mauna Loa [Rinsland et al., 1999; Novelli et al., 1998]. Back trajectories performed by Rinsland et al. [1999] suggest transport from regions with intense and widespread forest fires in Southeast Asia. The increased emissions from fires are further related to drought conditions prevailing through the El Niño–Southern Oscillation (ENSO) event that started in September–October 1997. Although the model uses satellite data from 1997 to distribute the fires, regional emissions could be underestimated for the ENSO period since the total amount of biomass burnt is based on averages from the 1980s. Increased biomass burning due to ENSO could probably also explain the discrepancy between the modeled and observed CO values at Samoa at the end of 1997. At Hohenpeissenberg, which is mainly influenced by pollution from fossil fuel and industry in Europe, there is, in general, good agreement between calculated and observed CO values. The available emission data lack seasonality of the fossil fuel CO emissions, and this is a likely explanation why the seasonal cycle in the CTM is underestimated.

image

Figure 4b. Comparison of measured and simulated (with the CTM2) monthly averaged CO for Hohenpeissenberg, Mauna Loa, and Samoa.

Download figure to PowerPoint

5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

5.1. Calculated Ozone Perturbations

[26] Several model experiments were performed to study the tropospheric ozone perturbations due to changes in total ozone. Perturbation experiments were performed with stratospheric ozone column densities reduced globally by 10%. Perturbation experiments were run for two selected time periods of 4 months each, one during the summer and one during the winter seasons in the two hemispheres. These time periods are characteristic for column ozone perturbations and at the same time sufficiently long to affect tropospheric ozone distribution. Ozone perturbations over different lengths of periods (1 week to 1 month) were also run to check the sensitivity to the integration period; however, the response of surface ozone was rather similar. These column ozone reductions lead to enhanced UVB fluxes and thereby enhanced photodissociation of chemical compounds and, in particular, ozone, through reaction (R1). A change of 10% is chosen to represent changes in ozone column densities. Such changes are also representative of global total ozone reductions encountered during the last 2 decades due to anthropogenic emissions of halocarbons. Although we realize that the observed ozone reductions are larger at high latitudes than at low latitudes, we have used the same perturbations for all latitudes and all seasons. A test run with a larger perturbation (30%) gave only small changes in the ozone sensitivity. We should notice that the perturbations take into account only the effect of changes in the photodissociation rates and not the effect of stratospheric ozone changes through changes in the influx of ozone, which could be significant. Changes in the net ozone production in the boundary layer at the three sites calculated with the Oslo CTM2 are given in Table 2b for January and July.

Table 2b. Change in 24 Hours of Total Ozone Net Chemical Production in the Planetary Boundary Layer Calculated in the Oslo CTM2 for 1 January and 1 Julya
 HohenpeissenbergMauna LoaSamoa
  • a

    Values are in ppb. Change is the difference between runs with reduced total ozone column (−10%) and the standard ozone column.

January0.21−0.13−0.35
July0.32−0.12−0.13

[27] The difference between the surface ozone distributions calculated with the reduced ozone column and with standard ozone column is shown in Figure 5 for January and July. These are the last months of two 4 month integrations. Negative values mean that surface ozone is reduced when the ozone column is reduced (a positive correlation between column ozone changes and surface ozone response). Similarly, positive values mean that surface ozone is enhanced when the ozone column is reduced (negative correlation). The CTM calculations give reductions in surface ozone in July in the perturbation runs where ozone columns are reduced, even in most regions in the Northern Hemisphere where NOx levels are enhanced. Although the dynamics is the same in the base and perturbation runs, there is more extensive mixing between the free troposphere and the planetary boundary layer (PBL) in July in the Northern Hemisphere and shorter chemical lifetime of NOx in the PBL, leading to lower NOx values in July than in January and a smaller area where ozone is produced. Furthermore, mixing of free tropospheric ozone into the PBL leads to reduced surface ozone levels. In the free troposphere where the NOx levels are lower than in the PBL, the ozone levels are reduced when the UV radiation is enhanced.

image

Figure 5. The difference in surface/lowest model layer ozone in ppb between CTM2 runs with reduced total ozone column (−10%) and the standard ozone column for (top) January and (bottom) July. When the difference is negative, there is a reduction of surface ozone.

Download figure to PowerPoint

5.2. Comparisons of Modeled With Observed Tropospheric Ozone Sensitivity

[28] Table 1 depicts the ozone sensitivity at the three sites, Samoa, Mauna Loa, and Hohenpeissenberg, on the basis of the Oslo CTM2 model runs and on the analysis of ozone observations at the three sites. The sensitivity results using the box model described in section 2.2 are also included to show the effect of chemistry alone.

[29] The largest positive sensitivity is found for the background site, Samoa, both in the model results and in the measurements, as expected from theory, with higher values in the summer season (December–February) than in the winter season (June–August). Here the agreement is good even if the amplitude is larger in the measurements. The CTM shows higher sensitivity than the box model since it includes subsidence from regions above the boundary layer. Many regions of the Southern Hemispheric free troposphere show strong sensitivity due to extensive regions of low NOx. At Mauna Loa the modeled sensitivity is positive for both January and July, 0.44 and 0.53, respectively. The analysis of the measurements shows a change from nearly zero sensitivity in the winter/spring to a positive sensitivity of 1.0 in the summer/autumn. The change in sensitivity from winter to summer, which is particularly pronounced in the measurements, is probably linked to seasonal differences in transport patterns to Hawaii. Comparison of modeled CO with measurements indicates that the chemical sources injected to the airflow or the transport process itself might be underestimated and this might be the reason for the discrepancy between model and measurements in January (Figure 4b). See section 4.2 for a more detailed discussion.

[30] The tropospheric ozone sensitivity is similar in the observations and the model simulations for Hohenpeissenberg during the winter months, with a negative value. However, there are significant differences between the model calculations and the measurements during the summer season. While the CTM gives a small positive sensitivity, observations show a clear negative sensitivity with enhanced surface ozone values when total ozone is reduced. One reason for the discrepancy could be that the NOx levels in the CTM are incorrect, but as can be seen from Figure 6a, the modeled surface level NOx is quite close to the observed. In the CTM model calculations, enhanced photochemical activity due to reduced stratospheric ozone leads to a net ozone loss in the free troposphere over Europe during the summer months and hence to a positive sensitivity of free tropospheric ozone to total ozone changes. In contrast to this the boundary layer exhibits an increased local NCP (Table 2b), hence a negative NCP sensitivity. However, the negative boundary layer NCP sensitivity does not translate into negative ozone concentration sensitivity. As a result of the extensive exchange of ozone between the free troposphere and the boundary layer in the CTM calculations, there is no increase in surface ozone levels in the boundary level induced by a reduction of total ozone, and hence a slight positive correlation between total ozone and surface ozone is calculated (Table 1). The box model calculations where no exchange of ozone with regions outside the boundary layer is included show a small negative correlation between total ozone and surface ozone, significantly smaller than found in the observations. The NOx levels over Hohenpeissenberg during the summer months (∼1 ppb, see Figure 6a) give a small sensitivity of surface ozone changes to changes in ozone column densities in the box model (Figure 1).

image

Figure 6a. Comparison of measured and simulated (with the CTM2) monthly averaged surface NO and NO2 for Hohenpeissenberg.

Download figure to PowerPoint

[31] A more likely explanation for the observed high negative correlation between total and surface ozone changes at Hohenpeissenberg during the summer months is that ozone is controlled by meteorological factors. Low total ozone is observed during situations with stagnant high-pressure systems which favor ozone buildup at the surface. In the perturbation studies such anticorrelations are not considered. Ozone buildup in high-pressure systems is demonstrated in Figure 6b, where surface pressure and observed and modeled surface ozone at Hohenpeissenberg during July and August are given. It is shown that the model can reproduce ozone levels well and that ozone builds up during high-pressure situations. A more detailed discussion of the link with dynamics is given below.

image

Figure 6b. Comparison of daily averaged observed ozone with ozone and surface pressure in Oslo CTM2 for July–August 1997 at Hohenpeissenberg.

Download figure to PowerPoint

[32] The photochemical link between total and surface ozone can be perturbed by dynamical coupling. Weather systems might affect the stratosphere and tropopause height. This, in turn, perturbs the thickness of the ozone column [Galliani et al., 1996; Holton, 1992]. In moving baroclinic systems, the troughs and ridges at 500 hPa lag the surface troughs and ridges. For example, ozone column maximum is often present in the rear side of a cyclone (trough) at the passage of a cold front. Minimum ozone column could be found in the outer sector of the high-pressure (ridge) area close to the warm front. In the mature stage, typical for stagnant high-pressure systems during summer, the 500 hPa and surface tendency is nearly in phase. For such situations a high pressure at the surface is connected with high tropopause and a corresponding low total ozone column. In order to look more closely on a possible bias in the results due to the effects of synoptic weather systems, a further analysis step was performed for the station of Hohenpeissenberg, where surface pressure data are publicly available. To account for this effect, the deseasonalized, detrended data were grouped into five pressure quantiles, and the regression coefficient and slope was again calculated for each pressure quantile. The results from this analysis step are presented in Figure 7 and Table 3. It is evident that the anticorrelation observed before (Figure 2) between total and surface ozone holds in each pressure quantile, which is linked to different synoptic patterns. The range of the slope of the correlation is −0.8 to −1.7 (Table 3). The slope is the sensitivity of tropospheric ozone to changes in columnar ozone, and the observed slopes at Hohenpeissenberg imply that for a given percentage of change in total ozone, surface ozone changes by the same percentage multiplied by −0.8 to −1.7. Nevertheless, Figure 7 shows that for low total ozone, high surface ozone correlates with high pressure. Thus sunny weather and stagnant air with accumulation of air pollution and ozone buildup preferentially occur during periods of low total ozone. The link of ozone values to surface pressure is further illustrated in the model studies shown in Figure 6b, which shows ozone variations over semipolluted regions in Europe (Hohenpeissenberg) during the summer months. The model-simulated ozone variations show good agreement with the observed ozone variations. In both cases there is an ozone buildup following the formation of a high-pressure system.

image

Figure 7. Surface ozone departures from the monthly mean (in ppb) for surface pressure/total ozone quantiles for Hohenpeissenberg.

Download figure to PowerPoint

Table 3. Slope of D (Surface O3)/D(Total O3), Correlation Coefficient (r2), and Number of Observations for the Pressure Quantiles of Summertime (June, July, and August) Hohenpeissenberg Data Plotted in Figure 7a
Pressure QuantileSensitivity (Slope (ΔO3/O3)/(ΔTO/TO))Correlation r2Number of Observations
  • a

    A slope of −1 means that a 1% decrease in total ozone leads to a 1% increase in surface ozone.

1−1.552−0.33280
2−0.828−0.14280
3−1.070−0.20281
4−1.284−0.19280
5−1.698−0.28280

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

[33] In this paper we have presented studies showing that changes in photodissociation rates induced by changes in ozone column densities may significantly affect lower tropospheric ozone. Observational data for three selected stations, which have data for a period of more than 1 decade, have been analyzed and compared with model results from simulations performed with the global-scale Oslo CTM2. The comparison between model studies and analysis of the observational data shows that ground level ozone at the remote, low-NOx stations of Mauna Loa and Samoa is correlated positively with total ozone, with an exception at Mauna Loa during the winter months, when the sensitivity of surface ozone to total ozone changes was close to zero. Hence at these stations, which lie in a net ozone loss regime, a reduction in the total ozone column leads to enhanced photochemical activity that reduces surface ozone levels. Samoa especially showed a particularly large sensitivity of surface ozone to changes in total ozone, both in observations and CTM model calculations. In contrast, at Hohenpeissenberg with higher NOx levels, which lies in a net ozone production regime, an anticorrelation between ground level ozone and total ozone was found, both in the observations and also in the model results during wintertime. Hence the enhanced photochemical activity due to the reduction in total ozone column leads to enhanced ozone production and surface ozone concentrations. During the summertime at Hohenpeissenberg, there is a disagreement between the observed and CTM modeled sensitivity in surface ozone. In the model it is likely that vigorous mixing during summer between the free tropospheric air with the boundary layer air of Hohenpeissenberg results in a slight positive correlation or no correlation between total ozone and surface ozone. However, through sorting the observed data in different surface pressure quantiles linked to different meteorological conditions, it was shown that an anticorrelation between total and surface ozone at Hohenpeissenberg persists for the different pressure quantiles. A more detailed comparison of ozone levels over Hohenpeissenberg during the summer months reveals that the model results agree well with observations during stagnant high-pressure situations with buildup of ozone levels. Since this correlates with low total ozone densities, surface ozone changes during reduced ozone column densities are highly affected by regional emissions of pollutants.

[34] Our study has clear implications for the deduction of surface and free tropospheric ozone trends induced by long-term changes in stratospheric ozone. At clean background stations, reduced stratospheric ozone leads to reduced surface ozone, while at regionally polluted stations (e.g., Europe) the picture is more complex; high NOx levels favor ozone formation when the ozone column is reduced. However, during the summer months at these latitudes, reduced ozone columns are often associated with stagnant high-pressure systems which favor buildup of ozone precursors (NOx, CO, and VOC) and ozone. Analysis of ozone sensitivity to reduced ozone column densities is therefore strongly linked to the large-scale weather situation.

[35] Currently, it is difficult to see a clear picture in trends in tropospheric ozone. It is highly variable from one region to another [Stahelin et al., 2001] depending on the levels of ozone precursors and the spatial and temporal reduction in stratospheric ozone. Furthermore, reductions in stratospheric ozone lead to reduced influx of ozone to the troposphere from the stratosphere. This has been observed to reduce free tropospheric ozone in the free troposphere in polar regions [Schnell et al., 1991; Taalas et al., 1997].

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Tropospheric Ozone Production and Loss
  5. 3. Analysis of Observed Surface and Total Ozone
  6. 4. Model Studies of the Distribution and Changes in Ozone and its Precursors
  7. 5. Comparison of Modeled and Observed Tropospheric Ozone Sensitivity
  8. 6. Conclusions
  9. Acknowledgments
  10. References
  11. Supporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.