Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
 We present results for multiple Saharan Air Layer (SAL) events and their relationship to tropospheric ozone mixing ratios that were observed in the Eastern Atlantic during the summer of 2010 using ozonesondes. In particular, 5 SAL events are sampled during 2010 in Cape Verde indicting a reduction of ozone mixing ratios throughout much of the SAL layer, except near the base of the SAL. In this layer of enhanced ozone mixing ratio, we find increases of 20–30 ppb in some cases between non-SAL and SAL conditions. In addition, ozone concentrations are enhanced above the SAL layer, with trajectories suggesting enhancements by lightning from the middle/upper troposphere with outflow from Africa. Additional aircraft measurements are required to examine the chemical and aerosol distributions from the Marine Boundary Layer (MBL) through the upper troposphere to determine the heterogeneous chemical processes related to reduced ozone mixing ratio, and further quantify elevated ozone mixing ratios at the base of the SAL and above the SAL.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Tropospheric ozone variability over North Africa is driven largely by biomass burning, dry deposition, biogenic NOx from soils and vegetation in the lower troposphere [Reeves et al., 2010]. In the upper troposphere lightning can produce NOX (LNOx) which can increase ozone mixing ratios [Jenkins et al., 2008; Nalli et al., 2011]. There is strong evidence to support the cross-equatorial transport of enhanced ozone mixing ratios from biomass burning in the Southern Hemisphere to the Guinea region [Real et al., 2010]. During the wet season, June, July and August (JJA), there is a north-south gradient of ozone mixing ratio in the lower troposphere with higher values found in the Sahelian region (North) and lower ozone mixing ratios in the guinea region (south) [Reeves et al., 2010].
 However, the presence of the Sahara desert with annual dust emissions from Africa ranging from 160–1600 Tg yr−1 and major sources of surface dust emissions in Mauritania, Mali and Algeria during JJA [Engelstaedter et al., 2006] is also likely to influence tropospheric ozone during individual dust events. The transport of dust from the African Continent across the Tropical Atlantic during JJA occurs in the lower to middle troposphere within the Saharan Air Layer (SAL) [Carlson and Prospero, 1972]. The SAL is associated with dry, stable air in the lower to middle troposphere along with elevated dust loadings.
 Tropospheric ozone variability over the Sahara and downwind locations over the Eastern Atlantic is largely uncertain because of limited measurements limiting our knowledge of the various chemical processes. The variability of tropospheric ozone mixing ratios related to Saharan dust events have been reported in earlier studies. In particular, ozone mixing ratios are reduced in the presence of Saharan dust [de Reus et al., 2000; Bonasoni et al., 2004; de Reus et al., 2005].
 Ozonesondes were launched from Cape Verde during the summer of 2010 to examine tropospheric ozone variability with a particular focus on the role of the SAL in altering the vertical distribution tropospheric ozone mixing ratios. Surface measurements from Cape Verde show year-round destruction of surface ozone because of halogens, surface deposition and low NOX conditions [Read et al., 2008]. However, entrainment from the free troposphere and elevated NOX mixing ratios that are transported from the African continent can lead to the enhancement of ozone mixing ratios [Lee et al., 2009].
 The objective of this work is to: (a) show how the lower troposphere vertical ozone structure is altered in the presence of the SAL; (b) examine case studies of two SAL events with successive measurements; (c) examine changes in the ozone mixing ratio associated with a large dust outbreak during 6–8 July 2010 [Drame et al., 2011]; (d) examine elevated ozone mixing ratios just above the top of the SAL.
 Ozonesonde measurements were launched from Sao Vicente, Cape Verde (16.84°N, 24.87°W) at the Cape Verde Atmospheric Observatory (CVAO) facility. GRAW DFM-97 radiosondes are combined with ENSCI ECCZ ozonesondes with a 0.5% KI solution were launched at approximately 1400 UTC. A total of 13 ozonesondes were launched from 26 June through 17 July. Ozonesondes were launched on 26, 28 and 30 June; 1–3, 6–8, 10, 11, 13 and 17 July 2010.
 The National Centers for Environmental Prediction (NCEP)/ Global Reanalysis data [Kalnay et al., 1996] and the Worldwide Lightning Location Network (WWLLN) [Dowden et al., 2008] data are used to characterize the wind flow in 400 hPa and identify lightning strikes during the period of 25 June to 8 July 2010. The Moderate Resolution Imaging Spectro-radiometer (MODIS) sensors provide aerosol optical thickness (AOT) on daily basis, which is averaged over the area between 16–18°N and 26–23°W. Ground-based estimates of AOT at 550 nm are provided by Aerosol Robotic Network (AERONET) which is located at Sal, Cape Verde (16.7°N, 22.9°W).
Figure 1a shows AOT from AERONET and MODIS for the period of 25 June through 17 July 2010. The AOT estimates suggest that dust was present during the majority of ozonesonde launches with values greater than 0.6 for 11 out of 13 ozonesonde launches during the period. AOT values were greater than 0.9 for 7 out of 13 ozonesonde launches. Figures 1b, 1c, and 1dshow the vertical profiles of temperature, relative humidity (RH) and ozone mixing profiles for non-SAL (26, 30 June, 10, 13 July), SAL (28 June 1, 2, 3, 6, 7, 8, 11, 17 July) and the average of all profiles. Given that the SAL was present most of the time, the average temperature profile is not much different from the SAL vertical profile. In particular, the SAL profile is warmer (3–4.5°C) than the non-SAL profile in the lower troposphere, and there is a stronger temperature inversion is found (Figure 1b).
 The SAL vertical profile of relative humidity shows that it is drier throughout most of the atmosphere when compared to non-SAL conditions, but especially in the lower troposphere between 925–750 hPa and again in the middle troposphere near 450 hPa (Figure 1c). The top of the SAL is located at approximately 500 hPa, where there is an increase in relative humidity. In these locations, the relative humidity is lower by 20–40%, with the lower tropospheric reduction fitting the conventional model of the SAL [Carlson and Prospero, 1972]. The vertical profile of ozone shows that there are larger ozone mixing ratios (up to 5 ppb) from the surface through 800 hPa followed by smaller ozone mixing ratios through 500 hPa for SAL conditions (Figure 1d).
Figure 2ashows relative humidity and ozone mixing ratio that have been vertically averaged over 500-meter intervals from the surface through 15 km for the 13 ozonesonde cases. In the majority of relative humidity profiles dry air is found in the 925–600 hPa layers and associated with the SAL. The dry air (RH < 40%) is evident in the 9 sounding denoted as SAL. Moist air can be found from the lower through middle troposphere in relative humidity sounding on 26, 28 June, and 13 July. Moist conditions (RH > 75%) are found in the marine boundary layer (MBL)--between the surface and approximately 925 hPa. Consistent with the SAL profiles ofCarlson and Prospero , a moistening at the top of the SAL layer is found near 500 hPa followed by a drying in the upper troposphere for nearly all of the ozonesondes launches.
Figure 2bshows low ozone mixing ratios (<30 ppb) in much of the lower troposphere (surface-500 hPa), especially for 28 June, 1, 2, 3, 6, 7 July. Low near surface ozone values are expected due the ozone destruction near the surface [Read et al., 2008], and the observed low ozone mixing ratios in the presence of the SAL between 850–500 hPa are consistent with aircraft measurements of de Reus et al. . The ozonesondes also show evidence elevated ozone mixing ratios just above the MBL for 28 June, 1, 2, 3, 6, 7, 11, 17 July. The ozone mixing ratios are considerably higher in the middle to upper troposphere for nearly all ozonesondes launches including those when the SAL is not present. This is strongly suggestive of an outside source of ozone such as lightning generated NOX (LNOX). Next we describe 3 case studies to show the daily variability of ozone mixing ratios in the troposphere.
3.1. Case Study 1 (26–30 June)
 During this period, the AOT was lower on 26 and 30 June with evidence of low dust loading from visible images (not shown), and AERONET estimating an AOT of 0.6, 1.1 and 0.4 for 26, 28 and 30 June respectively. Visible satellite images show dust leaving Senegal on 27 June and soundings from Dakar, Senegal show a SAL profile beginning at 0000 UTC through 1200 UTC 27 June. The NCEP reanalysis show that a thermal heat low at 925 hPa moved across Senegal, potentially causing the dust event from 26–28 June (see auxiliary material).
 The vertical profile of temperature at Cape Verde shows a strong temperature inversion on 28 June, which is 4–5°C warmer than the inversion on 26 or 30 June and larger than the average profile (Figure 3a). The warm air associated with the SAL is also considerably drier on 28 June relative to 26 and 30 June. The driest air associated with the SAL event is near 900 hPa with RH values less than 10%. In fact the entire 925–600 hPa column on 28 June is drier than 26 June as the SAL enters the region. A moistening of the atmosphere near the top of the SAL is found with RH values recovering to near 70% at 550 hPa, followed by another dry layer above 450 hPa (Figure 3b). The SAL is replaced by a humid air mass on 30 June, with relative humidity values in the 60–95% range from the surface through approximately 500 hPa.
 The ozone mixing ratio distribution shows a well-mixed MBL followed by a decrease in ozone at the top of the MBL where clouds may be present for 26 and 30 June (Figure 3c). On the other hand, the SAL event of 28 June shows an increase in ozone mixing ratio with a peak at the base of the SAL. Ozone mixing ratios increase from 20 to 45 ppb between 26 and 28 June in this layer. A decrease in ozone mixing ratios is found on 30 June in this layer when a Cape Verde is under the influence of a cleaner air mass. In the layers between 850 and 500 hPa lower ozone mixing ratios are found on 28 June relative to the average, 26 and 30 June, respectively in Figure 3c. This suggests that heterogeneous chemistry may act to limit ozone within the SAL layer as suggested by de Reus et al. [2000, 2005]. In the middle and upper troposphere ozone mixing ratios approach 100 ppb on 28 June near 400 hPa with lower ozone-mixing ratio values of 60–65 ppb found on 26 and 30 June.
3.2. Case Study 2 (30 June–3 July)
 In the second case study, relatively clean air on 30 June was replaced by dusty air that visible satellite images show leaving the coast of Africa on the morning of 1 July. During this period, the AERONET AOT increased from 0.4 on 30 June to approximately 0.9 during 1–3 July (Figure 1a). Soundings from Dakar, Senegal show a SAL profile beginning at 0000 UTC at through 1200 UTC on 30 June. A stronger temperature inversion is present for 1–3 July relative to 30 June in Figure 2d. The intrusion of dry air is present from 925 through 500 hPa for the period of 1–3 July with the largest difference noted between 925 and 850 hPa (Figure 3e). Reductions in RH of 30–70% are found within the 950–500 hPa layer when comparing 30 June to the 1–3 July SAL intrusion.
 Similar to Case 1 an increase in ozone mixing ratio is found near 925 hPa with values of 35–40 ppb in this shallow layer. Ozone mixing ratios double from 15 ppb on 30 June to 30 ppb on 1 July and further increase to 37 ppb on 3 July. In contrast ozone mixing ratios from 900–500 hPa ozone mixing ratios decrease from 25–30 ppb 30 June to near 15 ppb on 2 July once again implying ozone destruction (Figure 3f). Above the SAL layer, ozone mixing ratios values increase from approximately 40–50 ppb on 1,2 July and 60–70 ppb on 3 July.
3.3. Case Study 3 (6–8 July)
 In the third case, a large dust event initiated by a thermal low over Algeria on 4 July with AOT values observed in Dakar between 3 and 4 on 7 July. Calipso images and model simulations show the dust flowing over the monsoon reaching heights of 2–5 km [Drame et al., 2011]. This SAL event is associated with very strong winds near the surface and may have several source regions (Mauritania, Mali) in addition to the primary source region in Algeria. Visible satellite images show the heavy dust layer reaching Cape Verde during the afternoon of 8 July. Figure 4a shows the vertical profile of temperatures from Cape Verde, with the temperature inversion being lifted from near 950 hPa on 6 July to 850 hPa on 8 July.
 The dry air associated with the SAL intrusion is evident on 6 July near 950 hPa with relative humidity values near 20 percent but decreasing each day while the minimum is located at lower pressure/higher altitude (Figure 4b). Overall, the 900–500 hPa column is increasingly drier each day with the RH values reaching a minimum on 8 July. The 8 July RH profile is approximately 10–30% lower than the average RH profile in the 850–500 hPa layer. An increase in relative humidity is found near the top of the SAL followed by low values in the middle/upper troposphere.
 Similar to the earlier cases where the SAL was present, there is an increase in ozone mixing ratios near the base of the SAL, with values increasing from 25 ppb on 6 July to near 50 ppb on 7 July. On 6 and 7 July there is a decrease in ozone mixing ratio above the base of the SAL to 500 hPa with lower values on 7 July, relative to the average ozone mixing ratio profile. However, unlike the earlier case studies, there is a large increase in ozone mixing ratio on 8 July from 850–500 hPa relative to the prior two days when the SAL was present. This increase in ozone mixing ratio corresponds to a very high dust loading which was elevated in height as seen in Figure 1a and noted by Drame et al. . Elevated levels of ozone mixing ratios of approximately 75–85 ppb are also found between 400–300 hPa during the 6–8 July period.
 As shown in Figure 2b and for the 3 case studies, high ozone mixing ratios are found at approximately 400 hPa. The source of elevated ozone mixing ratios is most likely due to LNOX that is associated with lightning over the continent. Figure 5shows the lightning locations as identified by the WLLUN network for the period of 25 June through 8 July and overlain with 400 hPa streamlines from the NCEP reanalysis. The flow-pattern for the first nine ozonesondes show the mean flow coming from West Africa where lightning is abundant (seeauxiliary material). The anticyclone in the centered over Algeria is responsible for the easterly/southeasterly flow that arrives to Cape Verde.
4. Discussion and Conclusion
 We present the first published ozonesondes study of tropospheric ozone variability associated with the SAL from Cape Verde. We present 3 case studies where AOT values indicate that dust was present within the SAL layer. Our findings show that the vertical profiles of ozone are significantly altered in the presence of the SAL with lower tropospheric flow coming from Africa. In particular we find: (a) an increase in ozone mixing ratios at the base of the SAL in each case study, which was not present under non-SAL conditions. (b) A reduction in ozone mixing ratio from 850–500 hPa under SAL conditions when compared to non-SAL conditions, except for 8 July when there is an increase in ozone mixing ratio between 900 and 500 hPa when compared to 6–7. The loss of ozone may have occurred through heterogeneous chemical reactions on aerosol surfaces, the formation of nitrate particles leading to a loss of nitrogen or the reduction of important radicals associated with ozone production [Zhang et al., 1994; de Reus et al., 2000; Tang et al., 2004]. (c) Above 500 hPa, the ozone mixing ratios rapidly increased with a range of 60 to nearly 100 ppb in the 400–300 hPa layer which is most likely linked to production of LNOX over West Africa and transported westward towards Cape Verde. The production of LNOX leading to an increase in downstream O3 mixing ratios over the tropical Atlantic is consistent with Global Chemical Transport Model (GCTM) results [Barret et al., 2010].
 High dust loadings associated with elevated ozone mixing ratios at the base of the SAL and also present during the large dust event of 8 July was also found with SAL profiles at Dakar, Senegal during 2008 [Jenkins et al., 2012]. They suggest the possibility of denitrification by water stressed microbes on Saharan and Sahelian aerosols in the presence of higher water vapor amounts. Higher water vapor amounts could be vertically exchanged between the top of the moist MBL and the base of the SAL, through diffusion or turbulent fluxes, leading to an increase in NOX and the production of O3. There is evidence of higher surface NOX and O3 when during the winter season when the air masses are coming from Africa [Carpenter et al., 2010] although it is not clear what role individual dust events in day-to-day variability.
 This work implies that additional tropospheric aircraft observations are required to characterize and quantify the distribution of aerosols, O3, NO and other O3 precursors over the Sahara and Eastern Atlantic. Such measurements will help to define the processes that influence tropospheric ozone variability as it relates to the SAL, dust aerosols and lightning. The variability in tropospheric mixing ratios also implies that important factors must be taken into account for GCTMs and regional chemistry models. These factors include: surface sources of dust, heterogeneous chemistry, halogen chemistry and lightning parameterizations.
 We thank the Cape Verde Atmospheric Observatory for providing facilities for the research. This work was funded by National Science Foundation under the grant AGS 1013179.
 The editor thanks Anoop Mahajan and an anonymous reviewer for assistance evaluating this paper.