WRF-Chem model estimates of equatorial Atlantic Ocean tropospheric ozone increases via June 2006 African biomass burning ozone precursor transport

Abstract

Long-range horizontal and local vertical transport of biomass burning ozone precursors (i.e. carbon monoxide and nitrogen oxides) from Central Africa are simulated for June 2006. Twenty-kilometer resolution combined meteorological and chemical simulations examine transport pathways, spatial distribution, and quantities of ozone precursors and ozone. Results suggest that due to biomass burning, ozone mixing ratios increase by 28–33 parts per billion by volume in the lower troposphere (850 hecto-Pascals) over the Atlantic Ocean west of Central Africa during June. The inter-hemispheric transport of biomass burning emissions from Central Africa subsides over the Gulf of Guinea with a northward extent of approximately 2–5°N. In the lower troposphere, ozone mixing ratio increases decrease from 28 parts per billion by volume in the southern Gulf of Guinea to 2–3 parts per billion by volume on the Gulf of Guinea Coast. There is middle and upper tropospheric ozone enhancement of 6–12 parts per billion over the Equatorial Atlantic Ocean which is the result of convective detrainment of ozone precursors from deep convection on the Gulf of Guinea Coast followed by transport that propagates around a broad anticyclone. The model ozone produced by biomass burning emissions is less than the observed implying that lightning-induced nitrogen oxide emissions, which are not included in this simulation, are a significant tropospheric ozone source for the eastern Equatorial Atlantic Ocean.

1 Introduction

Abundant tropospheric ozone is produced over the eastern Equatorial Atlantic Ocean by biomass burning generated ozone precursors (carbon monoxide, CO; nitrogen oxides, NO_x=NO+NO₂) transported northward and westward from the Southern Hemisphere (SH) of Central Africa (i.e. Angola, Congo, the Democratic Republic of Congo (DRC), Gabon, and Zambia) during the boreal summer (June, July, and August). The ozone precursors and resultant ozone are transported northward across the Equator or remain in the SH. The inter-hemispheric pathways are northeastward (Jenkins et al. 1997; Mari et al. 2008; Williams et al. 2010) and northwestward towards the Gulf of Guinea (GoG) from the source region (Real et al. 2010; Williams et al. 2010. Once in the Northern Hemisphere (NH), along the GoG Coast, it is possible for ozone precursors and resultant ozone to be vertically transported and subsequently detrained by deep convection. This convection can be within the Inter-tropical Convergence Zone (ITCZ) or associated with the West African Monsoon (WAM, Janicot et al. 2008). Once in the upper troposphere (UT, 400 hPa to tropopause) ozone precursors and resultant ozone are transported southward and southeastward via a large upper tropospheric anticyclone over continental Africa and the Southern Hemispheric Atlantic Ocean (Barret et al. 2008). The ozone and precursors that are convectively transported have climatic importance as ozone in the upper troposphere has maximum effectiveness for radiative forcing.

The intra-hemispheric pathways occur through: (1) local convective transport followed by convective detrainment over the immediate biomass burning source region (Pickering et al. 1996; Halland et al. 2009; Huang et al. 2012) or (2) horizontal transport that is due west and west-northwest in the lower troposphere (LT, surface to 700 hPa) over the Equatorial Atlantic Ocean. Sauvage et al. (2007b) and Mari et al. (2008) have identified the active phase of the southern branch of the African Easterly Jet (AEJ-S) as significant to long-range transport and the spatial distribution of ozone and ozone precursors in the SH. The active phase contains easterly winds with speeds greater than a threshold of 4 m s⁻¹ at 700 hPa (Nicholson and Grist 2003).

During the summer of 2006, there were two field experiments conducted in the eastern Equatorial Atlantic Ocean and West Africa. They were the second cruise of the Aerosols and Ocean Expeditions (AEROSE II) and the African Monsoon Multidisciplinary Analysis (AMMA, Redelsperger et al. 2006). Measurements were taken from ship, ground, and aircraft platforms to constrain trace gas estimates in the region. These measurements depict distinct ozone mixing ratio pulses in the LT and the ozone mixing ratios exceeding 100 ppbv in the UT. The ozonesonde measurements were confined to the eastern Equatorial Atlantic Ocean (30°W-30°E, 5°S-5°N) over and along a transect at 23°W during June on the NOAA R/V Ronald H. Brown Ship during AEROSE II (Morris et al. 2006; Hawkins 2007; Jenkins et al. 2008; Nalli et al. 2011; Smith 2012). There were also ground-based ozonesondes launches at Cotonou, Benin (6.23°N, 2.21°E, Thouret et al. 2009) and Ascension Island (United Kingdom, 14.42°W and 7.98°S) as part of Southern Hemisphere Ozonesonde Network (SHADOZ, Thompson et al. (2003; 2012). The French-F20, German Falcon D20, Advanced Test Reactor, and British Aerospace-146 aircrafts performed in-situ measurements over West Africa in August 2006 (Reeves et al. 2010).

Ozonesonde measurements from AEROSE II depict ozone enhancement from the surface to the tropopause at 23°W (Morris et al. 2006; Hawkins 2007; Jenkins et al. 2008; Nalli et al. 2011; Smith 2012). While biomass burning is more intense and produces more ozone in July, August, and September (Ziemke et al. 2009) than June, the AEROSE II observations are exclusive to June. With the ozonesonde site at this longitude, it is possible to examine horizontal long-range transport and inter-hemispheric transport west of Ascension Island. Together with the SHADOZ sites (Thompson et al. 2003; 2012) at Contonou and, Ascencion Island, the AEROSE II ozonesondes comprise a network of three sounding sites in the eastern Equatorial Atlantic Ocean region. Data for the eastern Equatorial Atlantic Ocean is not available for July or August 2006 and over recent decades; it is rare to find multiple ozonesonde observation sites in this region. Hence, it is not feasible to examine the long-range transport of ozone precursor emissions and subsequent ozone production outside of June 2006 making it an optimal study period.

Aircraft measurements during August 2006 show high mixing ratios of CO greater than 300 ppbv and ozone mixing ratios greater than 100 ppbv in the 800–500 hPa layer with the highest values at 650 hPa near 5°N (Reeves et al. 2010). The higher ozone mixing ratios have their origins in Central Africa and are associated with biomass burning emissions and long-range transport (Mari et al. 2008). Ozone mixing ratios of 40–80 ppbv are found in the UT but CO mixing ratios of 100–150 ppbv were considerably lower than in the middle troposphere (MT, 700 to 400 hPa). Mari et al. (2008), using the Lagrangian Flexible Particle Diffusion (FLEXPART) model, suggests some of the upper tropospheric ozone enhancement is from biomass burning emissions from Central Africa.

The Williams et al. (2010) study used the AMMA Version of the Tracer Model (TM4_AMMA) calculated that biomass burning ozone precursor transport from Central Africa increases JJA 2006 mean ozone mixing ratios over the Tropical Atlantic Ocean by approximately 30 ppbv in the LT and increases ozone by 5 ppbv in the MT and UT. The Dynamic Meteorology Lab zoom Interactive Chemistry and Aerosols (LMDz_INCA) GCTM found ozone enhanced in the GoG by 15 ppbv in the LT, and over the eastern Equatorial Atlantic Ocean by 20–40 and 4 ppbv in the LT (825–750 hPa layer) and UT, respectively (Williams et al. 2010; Bouarar et al. 2011).

It is known that biomass burning emissions do not fully explain the tropospheric ozone mixing ratio increases observed over the eastern Equatorial Atlantic Ocean. In fact, Jenkins et al. (2008), suggest that lightning induced nitrogen oxides (LNO_x) make a large contribution to the enhanced ozone mixing ratios of 60–110 ppbv in the MT and UT downstream over an Equatorial transect at 23°W through westward transport of aged ozone enriched air. Measurements in 2008 and 2010 suggest LNO_x is responsible for elevated ozone in the MT and UT at Sao Vicente, Cape Verde (16.84°N, 24.87°W) and Dakar, Senegal (14.49°N, 17.49°W, Jenkins et al. 2012; 2013). June, July, and August 2006 ozonesonde measurements at Cotonou, Benin show relatively low ozone mixing ratios (20–40 ppbv) in the 0–2 km range but several relatively high ozone mixing ratio episodes at 3–5 km where they exceed 100 ppbv, most notably in August. Higher ozone mixing ratios (60–100 ppbv) are found in the UT (Thouret et al. 2009).

This study will use the combined meteorology and chemistry mesoscale model, the Chemistry version of the Weather Research and Forecasting Mesoscale Model (WRF-Chem, Grell et al. 2005). The model predicts synoptic and mesoscale meteorological features that affect the horizontal and vertical transport of chemical constituents and aerosols. The WRF-Chem model is suitable for simulations in the tropical latitudes (see Jenkins et al. 2012; Zhang and Chen 2012). Hence, this study uses two simulations to make conclusions specifically for June 2006: 1) a simulation without biomass burning emissions called a control simulation and 2) and one with biomass burning emissions (BB simulation) at 20 km grid spacing. These simulations will 1) determine first-order mean June 2006 increases of CO, NO_x, and resultant ozone due to biomass burning transported over the eastern Equatorial Atlantic Ocean to include the locations of the GoG, Cotonou, transect at 23°W, and Ascension Island; 2) determine where in the troposphere (LT, MT, UT) the largest mean increases in CO, NO_x, and resultant ozone occur; and 3) determine if the model increases of tropospheric ozone are in agreement with limited ozonesondes from Cotonou, 23°W, Ascension Island, and recent GCTM studies. The LNO_x emissions are not included in these simulations. The effects of these emissions on NO_x and ozone in the study region will be documented in a separate paper.

Section two of this study describes chemistry observational datasets, the WRF-Chem model and simulation framework, and the emissions datasets employed in this study. Section three will present the WRF-Chem simulation results and Section four discusses the implications of the study.

2 Observations and model description

This study examines several remote sensing datasets of fire locations, CO, NO_x, and ozone for June 2006. The fire area is determined using the Moderate Imaging Spectrometer (MODIS) data, which provide longitude and latitude of fire locations in June 2006 on continental Africa (Giglio et al. 2006). The National Center for Environmental Prediction (NCEP)/National Center for Atmospheric Research (NCAR) Reanalysis 925 hPa winds are interpolated to a uniform 2.5° grid (Kalnay et al. 1996) and are overlaid on the fire area data. Carbon Cycle Greenhouse Gas Group (CCGG) surface flask point CO mixing ratios for Ascension Island are reported for June 2006.

Tropospheric column ozone (TCO) is estimated from Ozone Monitoring Instrument (OMI) retrievals on the Aura Satellite using seven of the twenty-four 2.5 km deep layers and the NCEP/NCAR tropopause height records. The retrieval errors for the 2.5 km-deep layers vary from 6 to 35 % (~2–5 Dobson Units, DU, Liu et al. 2010). For tropospheric nitrogen dioxide (NO₂) columns, we used Level 3 (L3)-Version 3 OMI NO₂ product at 0.25 × 0.25° resolution data for June 2006 (Bucsela et al. 2013). Monthly averaged CO mixing ratios are derived from the L2 retrievals of the Tropospheric Emissions Spectrometer (TES, Beer et al. 2001; Halland et al. 2009) for 215.44, 464.16, and 681.291 hPa pressure levels and L3, version 6 1.00 × 1.00° retrievals from the Monitoring Pollution in the Troposphere (MOPITT) at the 900 hPa pressure level. The MOPITT CO mixing ratio retrievals have a greater sensitivity to the LT than TES (Boyard et al. 2012).

The numerical simulations are performed with the WRF-Chem Model (Grell et al. 2005) Version 3.2. The meteorological initial and lateral boundary conditions are provided by the 1 × 1° 6-hourly National Centers for Environmental Prediction et al. (1999) Global Forecasting System Final Analyses output. The WRF-Chem simulation uses the Lin microphysics scheme (Lin et al. 1983) and the Grell convective scheme (Grell and Devenyi 2002). The WRF-Chem simulations are integrated for the period: 0000 UTC 20 May to 0000 UTC 5 July. The simulations have a 20 km resolution, a spatial domain of 15°S-15°N and 30°W-30°E, a model top pressure level of 50 hPa, and 26 vertical sigma levels. The WRF-Chem model simulation allows feedback of the chemistry to the meteorology. Therefore, changes in ozone and aerosols result from biomass burning produce slightly different meteorology in the BB simulation than in the control simulation.

The initial and lateral chemistry boundary conditions include 6-hourly output for 0000 UTC 20 May to 0000 UTC 5 July 2006 from the Fourth Version of the Model for Ozone and Related Tracers (MOZART-4, Emmons et al. 2010). The MOZART-4 output includes species such as NO_x, CO, hydrocarbons, and ozone and has a 2.8 × 2.8° resolution. The MOZART-4 is driven by meteorological fields from the NCEP/NCAR Reanalysis (Emmons et al. 2010). The WRF-Chem simulations apply the Carbon Bond-Zaveri Chemical Mechanism (CBM-Z, Zaveri and Peters 1999), Fast-J photolysis scheme (Wild et al. 2000), and four bins of Model for Aerosol Interactions and Chemistry (MOSAIC) aerosols (Zaveri et al. 2008). Aerosol feedback and wet scavenging are turned on for these experiments.

There are two simulations in this study; the control simulation without fire emissions and the second includes observed fires from June 2006, which is denoted as BB simulation. Both of these simulations include biogenic emissions from the Model of Emissions of Gases and Aerosols from Nature (MEGAN, Guenther et al. 2006). It is a 1.0 × 1.0° modeling system that estimates aerosol and volatile organic compounds (VOC) emissions using constant temperature, leaf area index, and photosynthetically active radiation. Both WRF-Chem simulations include year 2000 anthropogenic emissions from the Reanalysis for Tropospheric Chemical Composition (RETRO) at 0.5 × 0.5° (Schultz et al. 2007).

The BB simulation includes 1.0 × 1.0° fire emissions of carbon, CO, carbon dioxide, NO_x, deuterium, nitrous oxide, non-methane hydrocarbons, methane, organic carbons, black carbon, 2.5 μm and total particulate matter, dry matter, and sulfur dioxide from the Global Fire Emissions Database (GFED v2, van der Werf et al. 2006), which inputs MODIS data. The amount of the constituents emitted per grid cell is computed from a combination of fuel load, burned fraction, combustion completeness, and a fraction of emissions from combustion. These emissions are averaged for every 8-Julian day period. The uncertainties (i.e. interannual variability in fire emissions, ejection heights for biomass burning emissions in global chemistry transport models, bottom-up and top-down emissions estimates, and CO burden) in GFED v2 versus newer versions of GFED are addressed in van der Werf et al. (2006) and Williams et al. (2012).

In addition to the June 2006 AEROSE II ozonesonde observations along 23°W (Morris et al. 2006; Hawkins 2007; Jenkins et al. 2008; Nalli et al. 2011; Smith 2012), simulation results are compared to observations from the SHADOZ Network. These locations include Cotonou, Benin (Thouret et al. 2009) and Ascension Island (Thompson et al. 2003).

3 Results

3.1 Overview of June 2006 observations

This section presents remote sensing observations, in situ observations, and increases of CO, NO_x, and ozone from the WRF-Chem BB simulation versus control simulation. The MODIS fire area product depicts many biomass burning areas and suggests high emission rates in the northern half of Angola, the southern third of the DRC, and northern Zambia during the month of June 2006 (Fig. 1a). In the region of 15°S-15°N and 30°W-30°E, there are 12,432 fire locations during June 2006. The NCEP/NCAR Reanalysis 925 hPa streamlines superimposed on the fire areas implies there is inter-hemispheric transport of CO and NO_x from the biomass burning source region (Barret et al. 2008; Mari et al. 2008). Figure 1b and c show the mean June 2006 CO emissions are 165–195 g CO m⁻² (8 days)⁻¹ and NO_x emissions are 2.5–3.0 g NO_x m⁻² (8 days)⁻¹ for northern Angola and southern DRC. Table 1 displays the total surface emissions anthropogenic, biogenic, and fire CO and NO_x for the month of June 2006 for the entire WRF-Chem domain. Carbon monoxide emissions from fires dominate over NO_x emissions and all the other CO emissions sources for the eastern Equatorial Atlantic Ocean. The emission rate for CO over June 2006 is 9.06 × 10⁸ mol km⁻².

Fig. 1

Maps depicting a) fire locations (red dots) observed by MODIS and the NCAR/NCEP Reanalysis mean 925 hPa streamlines in the area of 30°W-30°E and 15°S-15°N, b) GFED v2 mean CO emissions in g CO m⁻² (8 days)⁻¹(pink, orange and red squares), and c) GFED v2 mean NO_x emissions in g NO_x m⁻² (8 days)⁻¹ (pink, orange, and red squares) with the locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), 2.5°N, 2.5°E Gulf of Guinea location, and Ascension Island (7.98°S, 14.42°W), during June 2006

Table 1 Contains the June 2006 WRF-Chem model spatial and vertical tropospheric CO and NO_x emissions (mol km⁻²) from anthropogenic, biogenic, and fire sources

Since the MOPITT product is sensitive in the LT, this study uses the 900 hPa lower tropospheric level (Fig. 2a). The axis of westward transport is at approximately 3°N in the MOPITT product. The TES CO mixing ratios are gridded to six tropospheric pressure levels (Herman et al. 2013; Fig. 2) At 681.291 hPa (Fig. 2b), mean CO mixing ratios exceed 120 ppbv over the source in Central Africa and the GoG. The TES product shows CO mixing ratios greater than 120 ppbv occur over land and to the north of the source indicative of west-northwestward transport to the GoG Coast. There is CO mixing ratio enhancement greater than 150 ppbv over Nigeria at 900 hPa (see Fig. 2a). Compared to the 681.291 hPa TES CO (see Fig. 2b) and 900 hPa MOPITT CO (see Fig. 2a). There are mean TES CO mixing ratios of 120 ppbv directly over the source and the GoG at 464.16 hPa (Fig. 2c). The mean June 2006 CO mixing ratios exceed 105 ppbv over and off the coast of Gabon at 215.444 hPa (Fig. 2d). Indicative of vertical transport of CO to the UT.

Fig. 2

Contour plots of CO mixing ratio (ppbv) at a) 900.00 with the locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), Gulf of Guinea location (2.5°N, 2.5°E), and Ascension Island (7.98°S, 14.42°W), b) 681.291 hPa, c) 464.16 hPa, and d) 215.444 hPa, averaged for June 2006 from TES. Black contour lines represent CO mixing ratios equal to 50, 70, 90, 100, 120, and 150 ppbv. For Fig. g, white space denotes topography higher in altitude than the 900 hPa hPa level

Figure 3a shows that the highest values of mean June 2006 tropospheric column NO₂ (x 10¹⁵ molecules per cm²) occur in the biomass burning region. The column amounts exceed 2 × 10¹⁵ molecules per cm² over southern DRC and northern Angola. The largest NO₂ column amounts are along 8–7°S and between15 and 20°E and exceed 7 × 10¹⁵ molecules per cm².

Fig. 3

Contour plot of mean June 2006 OMI tropospheric column a) NO₂ (x 10¹⁵ molecules cm⁻²) with the white-marked locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), Gulf of Guinea location (2.5°N, 2.5°E), and Ascension Island (7.98°S, 14.42°W) and b) ozone (DU)

Mean June 2006 TCO increases are more widely distributed than NO₂ and CO, as ozone is photochemically produced downwind of biomass burning. Mean June TCO is in the range of 39–42 DU (Fig. 3b) off of the Gabon, DRC, and northern Angola coastlines and exceeds 42 DU on the Congo Coast to the northwest of the most intense biomass burning. Considering that 1 DU is approximately 1.0 ppbv, the OMI TCO noted here is comparable to the mean June 2006 tropospheric column mixing ratio of 50 ppbv in this vicinity calculated by the Global Modeling Initiative (GMI) model in Ziemke et al. (2009).

3.2 Carbon monoxide, nitrogen oxides, and ozone changes with biomass burning

For the month of June 2006, the WRF-Chem control simulation mean mixing ratios are subtracted from the BB simulation mean mixing ratios to determine changes in ozone precursors and ozone. At 925 hPa, the inclusion of fire emissions leads to mean CO increases of 300 ppbv (Fig. 4a) and NO_x exceeding 1800 pptv (Fig. 5a) in southwestern DRC and northern Angola. Mean ozone increases of 28 ppbv are found downstream of Central Africa over the eastern Equatorial Atlantic Ocean (Fig. 6a). At 925 hPa (see Figs. 1a and 4a), the CO and NO_x enhancement are found as far north as Cameroon as a result of northeastward inter-hemispheric flow (Barret et al. 2008; Mari et al. 2008) and at 850 hPa the southeast trade winds carry the ozone precursors over the central and southern GoG (north of the Equator and south of 2°N). At 850 hPa, average CO increases up to 300 ppbv over coastal DRC (see Fig. 4b). Due to its shorter lifetime, NO_x enhancement declines more rapidly with distance from the source than CO (see Fig. 5b). The abundant CO is collocated with NO_x, where the simulation shows there is widespread NO_x enhancement greater than 175 pptv and CO enhancement greater than 150 ppbv is over most of Angola and the southern DRC. The northward extent of CO enhancement greater than 25 ppbv, NO_x enhancement greater than 25 pptv, and ozone enhancement greater than 6 ppbv is limited to 2–3°N over the GoG (see Figs. 4b, 5b, and 6b). This is consistent with June 2006 observations showing limited ozone enhancement from the SH (Thouret et al. 2009).

Fig. 4

Contour plot of BB simulation streamlines and CO mixing ratio (ppbv) control versus BB simulation differences (the difference relative to the simulation without fire) at a) 925 hPa with the green-marked locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), Gulf of Guinea location (2.5°N, 2.5°E), and Ascension Island (7.98°S, 14.42°W), b) 850 hPa, c) 700 hPa, d) 400 hPa, and e) 200 hPa from WRF-Chem. White space denotes topography higher than the altitude of the 925 hPa and the 850 hPa pressure levels, Figs. a and b respectively

Fig. 5

Contour plot of BB simulation streamlines and NO_x mixing ratio (pptv) control versus BB simulation differences (the difference relative to the simulation without fire) at a) 925 hPa with the red and green-marked locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), Gulf of Guinea location (2.5°N, 2.5°E), and Ascension Island (7.98°S, 14.42°W), b) 850 hPa, c) 700 hPa, d) 400 hPa, and e) 200 hPa from WRF-Chem. White space denotes topography higher than the altitude of the 925 hPa and the 850 hPa pressure levels, Figs. a and b respectively

Fig. 6

Contour plot of BB simulation streamlines and ozone mixing ratio (ppbv) control versus BB simulation difference (the difference relative to the simulation without fire) at a) 925 hPa with the red and green-marked locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), 2.5°N, 2.5°E Gulf of Guinea location, and Ascension Island (7.98°S, 14.42°W), b) 850, c) 700, d) 400 hPa, and e) 200 hPa , and f) BB-control difference vertically-integrated tropospheric column ozone (DU) from WRF-Chem. White space denotes topography higher than the altitude of the 925 hPa and the 850 hPa pressure levels, Figs. a and b respectively

As expected, there is CO, NO_x, and subsequent ozone increases at 700 hPa (Figs. 4c, 5c, and 6c) which are smaller than at 925 and 850 hPa. The June 2006 mean CO increases exceed 100 ppbv and mean NO_x exceeds 40 pptv. The combination of these ozone precursors lead to 12 ppbv ozone enhancement over northern Angola. Smaller CO and NO_x increases of 25 and 10 pptv respectively extend westward to Ascension Island and 23°W and are associated with 7–12 ppbv mean increase in ozone mixing ratios over the Atlantic Ocean between 12°S and the Equator. Simulated CO and NO_x increases at 400 and 200 hPa are also relatively small (50 ppbv and 10 pptv) compared with lower altitudes (Figs. 4d-e and 5d-e).

Average ozone increases are 6–8 ppbv at 400 and 200 hPa just south of the Equator and 15°W-0°W (Fig. 6d-e) and correlate with small ozone precursor enhancement. There is an anticyclonic gyre at 400 hPa that directs the southward movement of these ozone precursor increases. However, at 200 hPa (see Fig. 6e), the slight CO enhancement is carried southward counterclockwise around a much broader upper tropospheric anticyclone that covers much of the Equatorial Atlantic Ocean and Central Africa (Barret et al. 2008). The small CO enhancement is transported toward Ascension Island and 23°W where average increases of 25 ppbv are indicated in Fig. 4 for the month of June 2006. Moreover, the June 2006 WRF-Chem model difference of the vertically integrated TCO for the two simulations is depicted in Fig. 6f. The difference of the BB and control simulation TCO suggests a 6–12 DU increase in TCO south of 5°N and extending from the biomass burning source region on the African continent to 30°W to include the region of the AEROSE II ozonesonde launches. Figure 6f also shows minimal TCO enhancement in the vicinity of Cotonou but robust westward transport. Hence, not only is the average CO, NO_x, and ozone increasing from biomass burning emissions in the LT but possibly from detrainment from the deep convection linked to the WAM and ITCZ.

3.3 June 2006 time evolution of ozone precursors and ozone

This study presents June 2006 vertical profiles of mean CO, NO_x, and ozone mixing ratios for the GoG (2.5°N, 2.5°E, Fig. 7a-c) and 2, 6, 9, 12, 16, 20, 23, and 27 June 2006 mean CO, NO_x, and ozone for Cotonou (Fig. 7d-f). At the GoG, there are mean CO increases of 70 ppbv in the LT, but above 800 hPa it is uniformly increased by 20–25 ppbv. There is NO_x enhancement of 30 pptv at the surface and virtually none above 650 hPa. The largest average ozone mixing ratio increases are 10–15 ppbv and occur near the surface. The increases are then reduced to near zero in the 700–500 hPa layer, but increase to near 5 ppbv in the UT. The surface increases of ozone are the result of biomass burning emissions from the source region. The small ozone enhancement of 1–2 ppbv above 450 hPa which is the result of CO detrained out of the deep convection along the GoG coast. At Cotonou, CO is increased by 15–20 ppbv with minor increases found in NO_x and ozone. In Fig. 7f, observed mean ozone mixing ratios for 2, 6, 9, 12, 16, 20, 23, and 27 June 2006 match BB simulation ozone mixing ratios averaged over the same period in the lower 100 hPa of the troposphere, but observed mean ozone mixing ratios increase monotonically with increasing altitude during the month of June. There are differences in the observed ozone and BB simulation of up to 70 ppbv above 200 hPa.

Fig. 7

WRF-Chem control (black-open circle) and BB (green-dashed) simulation a) CO mixing ratio (ppbv), b) NO_x (pptv), and c) ozone mixing ratio (ppbv) mean vertical profiles at 2.5°N and 2.5°E for the month of June 2006. Mean WRF-Chem d) CO mixing ratio (ppbv), e) NO_x (pptv), and f) mean WRF-Chem and observed ozone mixing ratio (light-blue solid, ppbv) vertical profiles at Cotonou, Benin (6.23°N and 2.21°E) averaged for 2, 6, 9, 12, 16, 20, 23, and 27 June 2006

Figure 8a-c show that the biomass burning emissions reach the vicinity of the AEROSE II Cruise Track due to the west-northwestward pathway from Central Africa and the AEJ-S. Carbon monoxide and NO_x increases and the subsequent ozone increases are 25–30, 20–30, and 8–10 ppbv respectively throughout the troposphere. Carbon monoxide has a lifetime in the troposphere of about 1–2 months. Furthermore, Hawkins et al. (2007) reported CO mixing ratio measurements of 145–170 ppbv from 5 to 10 June and 165–196 ppbv from the non-dispersive infrared gas filter correlation Thermo Environmental Instrument for the period of 25 June-4 July from 5°S-15°N at 23°W. In Fig. 8c, observed mean ozone mixing ratios for 11, 14, 15, 26, 29a, and 29b June 2006 match BB simulation ozone in the lower 150 hPa of the troposphere, but observed mean ozone mixing ratios increase with increasing altitude over the month of June. There are notable observed ozone mixing ratio peaks at 650 and 375 hPa. There are differences in the observed ozone and BB simulation is up to 30 ppbv in the MT and 70 ppbv just below the tropopause at approximately 160 hPa.

Fig. 8

WRF-Chem control (black) and BB (green) simulation mean a) CO mixing ratio (ppbv), b) NO_x (pptv), and c) ozone mixing ratio (ppbv) vertical profiles at 0°N and 23°W, averaged for 11, 14–15, 26, 29a, and 29b June 2006

At Ascension Island (Fig. 9a-c), the mean simulated CO enhancement for 2, 14, 20, and 28 June 2006 is 5–11 ppbv below 300 hPa with a peak of 15 ppbv at 900 hPa. There is little NO_x enhancement with the exception of up 30 pptv in the 750–550 hPa layer. There is a 5–10 pptv NO_x mixing ratio increase that corresponds with the CO mixing ratio increase at 900 hPa. At 200 hPa, there are CO and NO_x mixing ratio increases of 30 ppbv and 15–20 pptv, respectively in the UT at 175 hPa as a result of detrainment from deep convection on the GoG Coast. This corresponds with mean ozone mixing ratio increases up to 5 ppbv in the LT with a maximum at 700 and 200 hPa where it is increased 5–7 ppbv. The longer lifetime of NO_x in the UT coupled with the presence of CO leads to the 5–7 ppbv ozone enhancement at 200 hPa.

Fig. 9

WRF-Chem control (black-open circle) and BB (green-dashed) simulation a) CO mixing ratio (ppbv), b) NO_x (pptv), c) mean WRF-Chem ozone mixing ratio (ppbv) and observed ozone mixing ratio (light blue) vertical profiles at Ascension Island averaged for June 2006, and d) 2 June (red), 28 June (blue), mean 2, 14, 20, 28 June 2006 (black) observed ozone mixing ratio vertical profiles at Ascension Island

The vertical profile of the mean ozone mixing ratio on 2, 14, 20, and 28 June 2006 at Ascension Island depicts distinct layers of elevated ozone mixing ratios throughout the troposphere (Fig. 9c). This profile shows an ozone mixing ratio peak at 800 hPa that exceeds the BB simulation ozone mixing ratios by 10 ppbv. In the UT, the mean observed ozone mixing ratio peaks are up to 38 ppbv higher than BB simulation. In Fig. 9d, we focus on 2 days, 2 and 28 June where ozone mixing ratios greater than 60 ppbv are observed near 850 and 650 hPa respectively. When comparing the observations to the simulated mixing ratios, the 2 June observations are more than 50 ppbv higher at 850 hPa (see Fig. 9d). To examine these two cases further, CCGG CO mixing ratio observations at Ascension Island are plotted with the BB simulation and BB simulation increases.

Figure 10 shows CCGG surface observations at Ascension Island from 1 to 30 June indicating higher CO values early and late in the month. The BB simulation shows three peak periods of CO, for Ascension Island on 3 June, 8–11 June, and 21–26 June. During these periods, simulated mean CO mixing ratios have increased to the 30–80 ppbv range relative to the control simulation. The simulated CO mixing ratios are coherent with the observations except for the 8–11 June period when observed mixing ratios are closer to 60 ppbv rather than the 120 ppb simulated mixing ratios.

Fig. 10

Evolution of BB simulation CO mixing ratio (ppbv, blue) and control vs. BB simulation difference (light green, the difference relative to the simulation without fire), and CCGG average CO mixing ratio (ppbv) observations (red-open circle) at Ascension Island (7.98°S and 14.42°W) for June 2006 from WRF-Chem

Figure 11a-b show the 2 June NO_x and ozone simulated increases at 850 hPa along with streamlines (Fig. 12a) indicating the direction of the flow in the BB simulation. In Fig. 11a-b, there is a pulse of elevated NO_x, CO (not shown), and ozone transported westward toward Ascension Island from Central Africa. Simulated NO_x increases exceeding 175 pptv along with mean ozone increases of 28–33 ppbv are found on 2 June (see Fig. 11a-b). The BB simulation timing of the ozone precursors and ozone pulses from Central Africa are not exactly coherent with the observations. However, the BB simulation ozone vertical profile (Fig. 12b) for 2 June is similar to the observations (see Fig. 9d). The model simulated the layers of ozone increases (900–800 and 600–300 hPa). The magnitude is comparable to the observations in the 600–300 hPa layer (10–20 ppbv), but underestimated in the 900–800 hPa layer (15–20 ppbv as opposed to the observed 40–45 ppbv).

Fig. 11

Contour plot of BB versus control simulation and streamlines for a) NO_x (pptv) mixing ratio differences and ozone (ppbv) mixing ratio differences (the difference relative to the simulation without fire) at a-b) 850 hPa at 1400 UTC 2 June. c-d), 675 hPa at 14 Z 28 June, and e-f) 675 hPa at 15 Z 29 June from WRF-Chem. Coastlines and political borders are outlined in green. Black contour lines represent NO_x mixing ratio differences equal to 25, 100, and 300 pptv and ozone mixing ratio differences of −6, 0, 6, 20, and 30 ppbv. Fig. a contains the red and green-marked locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), Gulf of Guinea location (2.5°N, 2.5°E), and Ascension Island (7.98°S, 14.42°W) and white spaces that denote topography higher in altitude than the 850 hPa level

Fig. 12

WRF-Chem BB simulation streamlines and ozone mixing ratio vertical profiles for a-b) 850 hPa at 1400 UTC 2 June 2006, c-d) 675 hPa at 14 Z 28 June 2006, and e-f) 675 hPa at 15 Z 29 June 2006. Figure a contains the red and green-marked locations of the AEROSE II Ship track along 23°W, Cotonou, Benin (2.21°E, 6.23°N), Gulf of Guinea location (2.5°N, 2.5°E), and Ascension Island (7.98°S, 14.42°W)

Figures 11c-d and 12c-d provide insight on the 28 June 675 hPa pulse at Ascension Island and to the observed elevated ozone mixing ratios (>90 ppb) between 700 and 600 hPa at 3.4° N and 23°W on 29 June (Hawkins 2007; Jenkins et al. 2008; Smith 2012). The source of the 28–29 June pulse is Central Africa. Easterly winds along the Equator to 3°S (Fig. 12c) are carrying the pulse of ozone mixing ratios westward to the Ascension Island. At 1400 UTC on 28 June, ozone increases is 15–25 ppbv between 25°W and 0° longitude near 3°S. The simulation depicts 10–25 pptv NO_x increases (see Fig. 11c) and a 7–12 ppbv ozone increases over Ascension Island (see Fig. 11d). The vertical profile of BB simulation ozone mixing ratios has a 65 ppbv peak at 500 hPa and shows increases from 800 to 275 hPa (Fig. 12d). The relative peak in the Ascension Island ozonesonde observation on 28 June is at 675 hPa (see Fig. 9d). However, model ozone peak does match the observed 65 ppbv.

The 28 June ozone pulse depicted at 675 hPa in the Ascension Island observations, (see Fig. 9d) and 29 June at 23°W (Figs. 11e-f and 12e-f) suggest long-range transport. On the north side of this gyre (see 675 hPa flow, Fig. 12f), along the Equator, the simulation shows NO_x enhancement of 10–40 pptv and ozone increases of 2–6 ppbv. The observed ozone increases at 1454 UTC 29 June 23°W ozonesonde is 90 ppbv in the 675–650 hPa layer (Morris et al. 2006; Hawkins 2007; Jenkins et al. 2008; Nalli et al. 2011; Smith 2012). The BB simulation vertical profile shows the layer of model ozone mixing ratio enhancement at 23°W on 1500 UTC on 29 June is similar in depth (800–275 hPa) as Ascension Island (see Fig. 11d) on 28 June. The similar depth and west-northwestward flow (see Fig. 12f) movement suggests that this is the same ozone pulse.

Figure 13a-c depict the simulated time-evolution of CO increases over the Cotonou, 23°W, and Ascension Island. The highest amounts of CO transport occur at 23°W and Ascension Island and are distinct episodes in the simulation (see Fig. 13b-c). Carbon monoxide enhancement greater than 25 ppbv occurs throughout the troposphere during the periods of 12–13, 17–18, and 26–29 June at 23°W (see Fig. 13b). At Ascension Island, increases in CO in the UT are found during the second half of June and in the LT during 1–4 June, 8–11 June and 22–28 June (Fig. 13c) corresponding with the Ascension Island CCGG surface CO enhancement. Carbon monoxide enhancement exceeds 80 ppbv in the LT during the period of the 24–25 June, which is related to the detrainment and subsequent transport around the broad upper tropospheric anticyclone over the African continent and the Atlantic Ocean. At Cotonou, the transport of CO occurs between 1000 and 900 hPa with little evidence of middle or upper tropospheric transport from Central Africa. Hence, there is more zonal transport of the biomass burning emissions towards the Atlantic Ocean as compared to meridional transport towards the GoG coast (see Figs. 4, 5, 6, 7, 8, 9, 10, 11 and 12).

Fig. 13

Time–height evolution of CO mixing ratio differences (ppbv), the difference relative to the simulation without fire) from 1000 to 100 hPa for June 2006 from WRF-Chem at a) Cotonou, Benin (2.21°E, 6.23°N), b) Equator and 23°W, and c) Ascension Island. Black contour lines represent CO mixing ratio differences equal to 35 ppbv

Figures 14a-c depict the simulated time evolution of ozone differences over Cotonou, 23°W, and Ascension Island. As suggested from the small increases in CO, only minor changes in ozone are found at Cotonou during the month of June (see Fig. 13a). The largest changes of 6–18 ppbv of ozone are found at 23°W and Ascension Island during the month of June in the LT, MT, and UT through long-range transport (see Fig. 14b-c). Similar to CO, the results point to greater transport leading to average increases in ozone mixing ratios through zonal flow as compared to meridional flow during June 2006.

Fig. 14

Time–height evolution of ozone mixing ratio differences (ppbv, the difference relative to the simulation without fire) from 1000 to 100 hPa for June 2006 from WRF-Chem at a) Cotonou, Benin, b) Equator and 23°W and c) Ascension Island. Black contour lines represent ozone mixing ratio differences equal to 6 ppbv

The AEJ-S has been proposed as a critical mechanism in the transport of biomass burning emissions north and northeastward into northern DRC and the Central African Republic (Mari et al. 2008; Real et al. 2010) and westward over the GoG and Equatorial Atlantic Ocean. Mari et al. (2008) and Real et al. (2010) noted the presence of active phases of the AEJ-S in late June, July, and August 2006 related to the increased inter-hemispheric transport of CO and ozone mixing ratios. With the AEJ-S active phase wind speed threshold of 4 m s⁻¹ at 700 hPa (Nicholson and Grist 2003; Mari et al. 2008), the WRF-Chem Model depicts two active phase time periods: 16–19 and 25–29 June 2006 (Fig. 15). The 16–19 June active phase corresponds with CO mixing ratio increases of 70–150 ppbv between 6 and 3°S and averaged for 0–10°E during the 16–19 June period. The CO enhancement during 25–29 June exceeds 100 ppbv which is slightly smaller in magnitude than the 16–19 June period. During the inactive phase of the AEJ-S, there are small CO increases of 25–70 ppbv at 300 hPa. The increases in CO correspond with mean ozone increases of 2–12 ppbv. On 29 June, ozone mixing ratio increases of 16 ppbv are found.

Fig. 15

Time–latitude evolution of a) 700 hPa CO mixing ratio difference (ppbv, the difference relative to the simulation without fire), b) 700 hPa zonal winds (m s⁻¹), c) 300 hPa CO mixing ratio difference (ppbv), and d) 300 hPa ozone mixing ratio difference (ppbv) averaged for 10 to 0°E for June 2006 from WRF-Chem. Black contour lines in a) and c represent CO mixing ratio differences equal to 25, 70, and 100 ppbv

4 Discussion

A combined meteorology and chemistry model with a resolution of approximately 0.20° latitude by 0.20° longitude (20 × 20 km) is used for simulating June 2006 biomass burning emissions transport and mean tropospheric ozone precursors and resultant ozone increases. The results suggest meteorological synoptic-scale dynamics have a critical role in the long-range horizontal transport and localized vertical transport of ozone precursors and ozone over the eastern Equatorial Atlantic Ocean and Central Africa, respectively. In June 2006, the largest increases in CO and NO_x occurred over the Central Africa and the eastern Equatorial Atlantic Ocean. This study found mean ozone increases of 24–33 ppbv in the LT (925, 850 hPa) for June 2006 over the Atlantic Ocean waters adjacent to Central Africa which are comparable to the 20–40 ppbv increases in the 825–750 hPa layer in GCTMs for JJA 2006 (Williams et al. 2010; Bouarar et al. 2011). Moreover, the GCTM studies suggest the largest increases in ozone would occur over Central Africa during JJA 2006 at the surface. From a three-dimensional view, this implies that in the LT, the maximum ozone increases slope downstream and to the west with decreasing pressure over the eastern Equatorial Atlantic Ocean.

Moreover, biomass burning emissions are transported northwest over the GoG from the source region and northeast into Central Africa as noted by Real et al. (2010); however, the simulation shows the northern extent of 6 ppbv or greater mean ozone increases are limited to 2–4°N in June 2006. While our results are for an earlier period, there is reason to believe this pathway remains suppressed through August. For example, the Trajectory Model Klimaat Seismology (TRAJKS) Model study in Williams et al. (2010) shows the majority of the trajectories from biomass burning emissions extending to the west of GoG and reaching 23°W.

Ozone transported to or produced over the GoG as a result of northward transport of ozone precursors could be reduced by vertical transport in WAM and ITCZ deep convection. In a GCTM study, Bouarar et al. (2011) suggests that for JJA 2006, the northern extent of ozone enhancement from biomass burning emissions is approximately 6°N. Robust ozone enhancement eventually reaches this far north because the ITCZ deep convection moves north and biomass burning emissions have a more expansive areal coverage in July and August. In addition to the westward and occasional northwest biomass burning emission pathways, there is the northeast pathway in the LT for ozone precursors and resultant ozone. This continental pathway of ozone seems to be confined to south of 5°N. Hence, this minimal northern extent of the biomass burning emissions suggests a southern spatial bias in the WRF-Chem mesocale model. For example, the TES and MOPITT products suggest slightly higher CO mixing ratios reaching to more northern latitudes in the MT and LT. This southern bias in this study aligns with other summer 2006 trajectory studies such as Mari et al. (2008) and global chemistry transport model studies: Williams et al. (2010) and Bouarar et al. (2011). In all these cases, more sensitivity testing beyond the scope of this study is needed to discern whether these differences are from inaccuracies in the biomass burning emissions inventory or a result of pollution transport from other regions such as eastern Africa or India (Barret et al. 2008).

Local vertical transport to the UT increases during the inactive phases of the AEJ-S as there tends to be more convection (Mari et al. 2008) over central Africa. Over central Africa, particularly in northern and eastern DRC (downwind of the Virunga Mountains along 30°E), southern Congo, Gabon, and northern Angola, there is overlap of deep convection and fires that cause “mixing and cooking” (Chatfield and Delaney 1990). There is evidence of this occurring in the simulations with higher mean CO mixing ratios (25–70 ppbv) found at 300 hPa during 16–22 June 2006 leading to ozone increases of 2–12 ppbv. The ozone precursors are transported downwind and contribute slightly to ozone increases over the Equatorial Atlantic Ocean. Huang et al. (2012) concluded this is the most efficient transport mechanism for CO, but in JJA it does not contribute as much as horizontal transport pathways.

The portion of the biomass burning emissions that reach the northern GoG and GoG Coast and then are entrained into deep convection are subsequently transported vertically to the UT via updrafts. Then they are detrained and carried in an anticyclonic direction. The detrainment results in a 6–12 ppbv mean ozone increase at 400 and 200 hPa over the eastern Equatorial Atlantic Ocean. At 200 hPa, mean ozone increases are transported as far south as 15°S and as far east as the biomass burning emission source region (~18°E) over northern Angola. The mean increases in ozone mixing ratios are higher than Williams et al. (2010) and Bouarar et al. (2011) whose GCTM simulations suggest a 4–5 ppbv ozone increase. This difference may be due to the finer grid scale in the WRF-Chem Model, leading to stronger vertical updrafts and chemical tracer detrainment rates in the UT (Hoyle et al. 2011) compared to coarser GCTMs. The easternmost average ozone increases at 200 hPa are the result of anticyclonic recirculation that likely contributes to the observed OMI TCO maximum over northern Angola, coastal DRC Congo, and Gabon (see Fig. 3b).

The WRF-Chem simulations show LT mean ozone increases of 2–12 ppbv extend far westward to Ascension Island and the AEROSE II Cruise at 23°W where ozone measurements are taken. The model shows mean ozone increases of 6–18 ppbv are found in the LT and MT, especially during 21–30 June 2006. But these increases in the mean ozone mixing ratio from biomass burning emissions over the background (control simulation) are insufficient to match the 60–95 ppbv observed ozone mixing ratios at 23°W (Morris et al. 2006; Hawkins 2007; Jenkins et al. 2008; Nalli et al. 2011; Smith 2012). The enhancement from WRF-Chem is much lower than the 20–40 ppbv ozone enhancement generated by biomass burning suggested by Williams et al. (2010) and Bouarar et al. (2011).

There are frequent pulses of ozone precursors and resultant ozone simulated by WRF-Chem and observed by OMI that emerge off of the coast of Central Africa and move westward. In the MT, these pulses may take a more northwestward direction due to an anticyclonic gyre in the southwestern portion of the model domain at 675 hPa. As the summer season evolves, the increased frequency of northwestward pulses in the GoG as observed during August 2006 (Thouret et al. 2009; Reeves et al. 2010) may be associated with the passage of stronger African Easterly Waves (AEWs, Janicot et al. 2008) which can increase southerly flow in their wake and hence the poleward transport of ozone precursors and resultant ozone. The results presented here suggest most of the OMI enhanced TCO in the SH (see Fig. 3) is related to biomass burning. Even though we have quantified the biomass burning emission contribution here, observations of high ozone mixing ratios in the MT and UT in the SH, specifically at Ascension Island, over the Equatorial Atlantic Ocean (Morris et al. 2006; Hawkins 2007; Jenkins et al. 2008; Nalli et al. 2011; Smith 2012) and in the Sahel of NH Africa (Jenkins et al. 2012; 2013) are not fully explained from biomass burning alone. Other tropospheric ozone sources exist in the eastern Equatorial Atlantic Ocean (i.e. LNO_x, soil NO_x and stratospheric/tropospheric exchange), however it is likely that for much of the year LNO_x is the most significant among them and will explain much of the difference between ozone produced by model biomass burning emissions and the ozonesonde observations. The ozone generated from LNO_x in this region and for this time period will be demonstrated in a manuscript to quickly follow this study.

Figures 7f, 8c, and 9c show that mean observed June 2006 ozone mixing ratios are 38–70 ppbv higher than the BB simulation ozone mixing ratios above the 200 hPa level and 7–38 ppbv higher in the MT over Cotonou, 23°W, and Ascension Island. The discrepancy in ozone between the model and observed ozone suggests that the vertical transport of ozone precursors to the UT is present, but relatively small (CO of 25–50 ppbv at 23°W and Ascension Island; see Fig. 13a-b), hence the generation of ozone in the MT and UT is small. Moreover, the model and observed differences suggests that while LNO_x may not be a perfect comparison, it is a significant source that must be accounted for. Additionally, several GCTM studies show significant ozone production over the eastern Equatorial Atlantic Ocean. For example, Sauvage et al. (2007b) shows that ozone production over the eastern Equatorial Atlantic Ocean via LNO_x emissions represents 37 % of the year 2000 total column ozone. Barret et al. (2009) and Bouarar et al. (2011) noted average upper tropospheric ozone increases in August 2006 and JJA 2006 respectively, are 10–20 ppbv, and that 2004–2005 satellite observed mean TCO increases by 15 DU (Martin et al. 2007) when LNO_x is considered.

5 Conclusion

This study presents mean ozone precursor mixing ratios and ozone increases simulated by the combined meteorology and chemistry model WRF-Chem with a resolution of approximately 0.18° latitude by 0.20° longitude (20 × 20 km) exclusively for June 2006. Even though biomass burning remains primarily in the Southern Hemisphere throughout the Northern Hemispheric summer season (June, July, and August) and the spatial transport mechanisms are similar (Suavage et al. 2007b; Mari et al. 2008; Real et al. 2010), the WRF-Chem model findings presented here do not apply to the entire summer, but only June 2006. In June 2006, modeling results show that ozone precursor and ozone enhancement remains primarily south of the Equator due to frequent westward moving pulses of biomass burning emissions emerging from the coast of Central Africa. These results show ozone increases by 24–33 ppbv in the LT (925and 850 hPa) over the Atlantic Ocean waters adjacent to Central Africa. These emissions are then transported to Ascension Island, 23°W, and beyond resulting in the classic S-shaped ozone vertical profile [Hawkins 2007; Jenkins et al. 2008; Nalli et al. 2011; Smith 2012]. The 16–19 June and 25–29 June AEJ-S active phases induce horizontal transport at 700 hPa and mitigate vertical transport of ozone precursors and ozone. In the MT (700 and 400 hPa), the mean ozone increases are 6–12 ppbv over much of the eastern Equatorial Atlantic Ocean. Although minimal ozone precursors and ozone are horizontally transported northward to the GoG Coast, those that reach the coast are vertically transported to the UT by WAM and ITCZ convection. The ozone precursors and ozone are then recirculated around an upper tropospheric anticyclone causing a mean ozone enhancement of 6–12 ppbv as far south as 15°S and as far east as northern Angola.

There are several related issues that should be explored further: 1) the refined quantification of the LNO_x emissions contribution to the ozone budget over the West Africa, adjacent Atlantic Ocean waters, and the SH Atlantic Ocean ozone maximum, 2) the 7–70 ppbv discrepancy between observed ozone mixing ratios and the BB simulation ozone mixing ratio estimates at Cotonou, 23°W, and Ascension Island, and 3) the slow northward propagation of the biomass burning emissions to the GoG Coast during the summer of 2006, particularly at Cotonou. A manuscript is in preparation to address some of these issues with the WRF-Chem over an identical domain. The two additional simulations will isolate LNO_x emissions and combine fire and LNO_x emissions over the eastern Equatorial Atlantic Ocean domain using ground-based lightning locations.