A pilot study of gaseous pollutants ’ measurement ( NO 2 , SO 2 , NH 3 , HNO 3 and O 3 ) in Abidjan , Côte d ’ Ivoire : contribution to an overview of gaseous pollution in African cities

This work is part of the DACCIWA FP7 project (Dynamics-Aerosol-Chemistry-Cloud Interactions in West Africa) in the framework of the Work Package 2 “Air Pollution and Health”. This study aims to characterize urban air pollution levels through the measurement of NO2, SO2, NH3, HNO3 and O3 in Abidjan, the economic capital of Côte d’Ivoire. Measurements of inorganic gaseous pollutants, i.e. NO2, SO2, NH3, HNO3 and O3 were performed in Abidjan during an intensive campaign within the dry season (15 December 2015 to 16 February 2016), using INDAAF (International Network to study Deposition and Atmospheric chemistry in AFrica) passive samplers exposed in duplicate for 2week periods. Twenty-one sites were selected in the district of Abidjan to be representative of various anthropogenic and natural sources of air pollution in the city. Results from this intensive campaign show that gas concentrations are strongly linked to surrounding pollution sources and show a high spatial variability. Also, NH3, NO2 and O3 gases were present at relatively higher concentrations at all the sites. NH3 average concentrations varied between 9.1± 1.7 ppb at a suburban site and 102.1± 9.1 ppb at a domestic fires site. NO2 mean concentration varied from 2.7± 0.1 ppb at a suburban site to 25.0± 1.7 ppb at an industrial site. Moreover, we measured the highest O3 concentration at the two coastal sites of Gonzagueville and Félix-Houphouët-Boigny International Airport located in the southeast of the city, with average concentrations of 19.1± 1.7 and 18.8± 3.0 ppb, respectively. The SO2 average concentration never exceeded 7.2± 1.2 ppb over all the sites, with 71.5 % of the sampling sites showing concentrations ranging between 0.4 and 1.9 ppb. The HNO3 average concentration ranged between 0.2 and 1.4 ppb. All these results were combined with meteorological parameters to provide the first mapping of gaseous pollutants on the scale of the district of Abidjan using geostatistical analysis (ArcGIS software). Spatial distribution results emphasize the importance of the domestic fires source and the significant impact of the traffic emissions on the scale of the city. In addition, in this work we propose a first overview of gaseous SO2 and NO2 concentrations on the scale of several African cities by comparing literature to our values. The daily SO2 standard of World Health Organization (WHO) is exceeded in most of the cities reported in the overview, with concentrations ranging from 0.2 to 3662 μg m−3. Annual NO2 concentrations ranged from 2 to 175 μg m−3, which are lower than the WHO threshold. As a conclusion, this study constitutes an original database to characterize urban air pollution and a first attempt towards presenting a spatial distribution of the pollution levels at the scale of the metropolis of Abidjan. This work should draw the attention of the African public authorities to the necessity of building an air quality monitoring network in order to (1) to define national standards and to better control the pollutants emissions and (2) to investigate the impact on the health of the growing population in developing African countries. Published by Copernicus Publications on behalf of the European Geosciences Union. 5174 J. Bahino et al.: A pilot study of gaseous pollutants’ measurement in Abidjan


Introduction
For several years, urban areas have experienced a deterioration in air quality and an increase in health and environmental impacts.Several scientific studies have shown that a large number of premature deaths, as well as respiratory and cardiovascular diseases, are related to air pollution (Brook et al., 2004;Pope III et al., 2002;Pope III and Dockery, 2006;WHO, 2006).According to the World Health Organization (WHO) estimation in 2012, 11.6 % of the deaths in the world were associated with outdoor and indoor air pollution.It represents nearly 6.5 million of deaths per year, among which 3 million deaths were attributable only to ambient air pollution.A total of 88 % of these deaths occurred in low-and middle-income countries (WHO, 2016).Also, according to the Organisation for Economic Co-operation and Development (OECD) Centre, 700 000 premature deaths were linked to air pollution in Africa in 2013 (Roy Rana, 2016).
Anthropogenic activities are considered to be responsible for the main sources of gaseous and particulate pollutants emissions into the air and their concentrations measured in the urban atmosphere (Kampa and Castanas, 2008).Several studies have been carried out in major cities of Europe, Asia and northern America.These studies underlined the emergency of taking action to reduce the emission of pollutants into the atmosphere in order to mitigate health and environmental impacts.Air quality networks have been set up in major cities of the United States and Europe with the purpose of informing the public authorities as well as the population in real time.However, in Africa, where researchers have found that air pollution causes more premature deaths per year than either unsafe drinking water or malnutrition, an air quality monitoring network is almost non-existent (Roy Rana, 2016).Very few studies on air pollution have been carried out over the African continent; this is particularly true for the western African region (Norman et al., 2007).
The western African region has experienced an economic upturn during the last few years characterized by an economic growth estimated at 2.7 % in 2017 with a prospect of 3.5 % in 2018 (IMF, 2017).Most of the economic activities of African countries (e.g.industries, trade, transport and real estate) are concentrated in cities.Therefore, this favoured a strong population explosion in cities due to a massive rural exodus and significant migration of western African populations (Denis and Moriconi-Ebrard, 2009).This intense economic activity, associated with a rapid population growth, strong urbanization and the perpetual uncontrolled expansion of cities, causes increasingly significant anthropogenic emissions of gaseous and particulate pollutants.This causes a significant deterioration of general air quality that may alter the health of populations and damage ecosystems (Fourn and Fayomi, 2006;Norman et al., 2007).
Very few qualified datasets related to the levels of pollutants' concentrations are available for African cities.Through some research programs and projects such as AMMA (African Monsoon Multidisciplinary Analysis), IN-DAAF (International Network to study Deposition and Atmospheric chemistry in AFrica) and POLCA (POLlution des Capitales Africaines), some experimental studies have been carried out in certain African capitals and rural areas such as in Bamako, Dakar, Yaoundé, and Amersfoort (Adon et al., 2016;Conradie et al., 2016;Liousse and Galy-Lacaux, 2010;Lourens et al., 2011;Val et al., 2013).The DAC-CIWA (Dynamics-Aerosols-Chemistry-Clouds Interactions in West Africa) European program follows the framework of the above-mentioned programs.The measurements implemented within the framework of DACCIWA are mainly focused on the Southern western African (SWA) region.DAC-CIWA has conducted extensive fieldwork in the SWA to collect high-quality observations, spanning the entire process chain from surface-based natural and anthropogenic emissions to impacts on health, ecosystems and climate (Knippertz et al., 2015).Work Package 2 (WP2) of the DAC-CIWA program aims to link emission sources, air pollution and health impacts in terms of lung inflammation and related diseases over representative differentiated urban sources in SWA (i.e.traffic, domestic fires and waste burning in Abidjan, Côte d'Ivoire, and two-wheel vehicle traffic in Cotonou, Benin).The strategy in WP2 is based on a multidisciplinary approach (e.g.physics, chemistry, toxicology, epidemiology, modelling).Experimental results from WP2 rely both on intensive short campaigns and also on mid-term monitoring measurements at the different selected urban sites.Three urban supersites to represent the major urban sources mentioned above have been selected for the mid-term monitoring from December 2014 to March 2017.
Abidjan is located close to the Ebrié Lagoon along the Gulf of Guinea on the south-east coast of Côte d'Ivoire.Abidjan, the economic capital of Côte d'Ivoire is a cosmopolitan city in sub-Saharan Africa and the second most populated western African city.According to the National Institute of Statistics, the population of Abidjan was estimated to be 4.7 million in 2014, with an average annual population growth of 2.6 % (INS, 2015).The autonomous district of Abidjan is composed of 10 municipalities in the city (Abobo, Adjamé, Attécoubé, Cocody, Koumassi, Marcory, Plateau, Port-Bouët, Treichville and Yopougon) and three neighbouring municipalities (Anyama, Bingerville and Songon).The cumulative area of these municipalities is 2119 km 2 (i.e.0.66 % of the country) (Yao-Kouassi, 2010) compared to 580 km 2 in 1990 (Kopieu, 1996).Abidjan is a port city and a dynamic economic centre, not only for the country, but also for the entire western African subregion.The port of Abidjan is of the first rank in western Africa in terms of merchandise traffic, with 21.5 million tons in 2013 and a prospect of 27 million tons in 2018.Abidjan hosts 70 % of the country's industries and constitutes the main maritime facade for the hinterland countries such as Burkina Faso, Niger and Mali.In addition, Abidjan represents 60 % of the gross domestic product (GDP) of Côte d'Ivoire, which is estimated to be USD 31.76 billion (Country Economy, 2016).The annual growth rate of GDP in Côte d'Ivoire is expected to be 9 % at the end of the third quarter of 2017.The main sectors that support the economic growth of Côte d'Ivoire are real estate, road transport, road building, manufacturing and mining industries.Abidjan has experienced strong economic growth in recent years that contributes to different environmental issues such as air pollution.This is one of the main issues in western African capitals, driven by explosive unplanned growth of urban conglomerations and contributing to unregulated emissions.
This study presents results of an intensive measurement campaign performed from 15 December 2015 to 16 February 2016.This work, part of the WP2, aims to characterize urban air pollution in Abidjan through the measurement of the concentrations of gaseous pollutants (NO 2 , SO 2 , NH 3 , HNO 3 and O 3 ).Twenty-one sampling sites distributed throughout the district of Abidjan were selected to be instrumented with gaseous passive samplers in order to obtain the spatial distribution of each gaseous pollutant.This study allows the contribution of each source of pollution to be established as well as the concentrations of inorganic gaseous species in Abidjan to be compared with those of other cities in Africa and other developing countries.

Sampling sites
For air pollution assessment in a city, it is important to consider most of the anthropogenic activities that may constitute sources of pollution.For the purpose of obtaining a good spatial coverage of the Abidjan district, several measurement sites have chosen across the different municipalities.
In this work, a network of experimental measurement sites was built.Measurement sites were chosen based on criteria of land use, orographic effects and the objectives of the DAC-CIWA program (i.e.(1) population density of each municipality is considered as a factor highly impacting gaseous pollutants' sources, (2) spatial coverage of the 13 municipalities of the district of Abidjan and (3) documentation of identified air pollution main sources).Figure 1 presents the mapping of the 21 selected sites referenced from A 1 to A 21 and their identified major sources of pollution nearby in the district of Abidjan, while Table 1 presents details of each site.So, 10 sampling sites are classified as traffic sites (A 1 -A 8 , A 10 , A 14 ) of which 5 of these sites present a combination of traffic and another pollution source (A 4 , A 7 , A 8 , A 10 , A 14 ).A 1 is also the DACCIWA supersite representative of traffic.Three sites are considered to be representative of industrial areas (A 9 to A 11 ).Industrial sites are often influenced by traffic.Four sites are representative of the residential district (A 7 , A 8 , A 12 and A 13 ).Four sites (A) have been selected as suburban and one as a green area (A 17 to A 21 ).Finally, two DAC-CIWA supersites studied over a 28-month period (December 2014-March 2017) in terms of local emission sources were considered in this study to be representative of domestic fires (A 16 ) and waste burning (A 15 ).These two supersites as well as A 1 (traffic site) and A 6 (Airport FHB) are shown in Fig. 2. Mid-term measurement sites are operational.The seasonal variation of gaseous pollutants showed that the highest concentrations are measured during the dry season (December-February).This information led to the choice of the study period.

Sampling procedure
Ambient concentrations of gaseous pollutants (NO 2 , SO 2 , HNO 3 , NH 3 and O 3 ) were determined using the INDAAF (http://indaaf.obs-mip.fr,last access: 3 April 2018) passive sampler method.These passive samplers developed by the Laboratory of Aerology (LA) in Toulouse (France) using the work of Ferm (1991) were tested and qualified in the framework of the INDAAF project for major African ecosystems over the past 15 years (Adon et al., 2010).Validation and inter-comparison studies have been performed according to comparisons with active analysers.Results presented in Adon et al. (2010) assure the quality and the accuracy of the measurements.Passive samplers have also been used for the measurement of gaseous pollutants' concentration in urban areas during the POLCA program (Adon et al., 2016).Passive samplers can be used for indoor and outdoor pollution monitoring in all environments (Salem et al., 2009) and provide good results (Gorecki and Namiesnik, 2002).Moreover, they are suitable to investigate the spatial distribution of gaseous pollutants in an air quality monitoring network (Carmichael et al., 2003;Cox, 2003;Cruz et al., 2004).Also, INDAAF passive samplers have the advantage of being small, silent and reliable, and they do not need electricity.The sampling technique is based on the molecular diffusion of gas molecules into the passive sampler, where they are quantitatively collected on an impregnated filter (He, 2014;Galy-Lacaux et al., 2001).The calculation of gaseous pollutants' concentrations in the air depends on the physical characteristics of the passive samplers, the duration of exposure and the meteorological parameters.We calculate the gaseous pollutant average concentration in ppb using the following formula given by Adon et al. (2010) Eq. (1): where C amb (ppb) represents the concentration of the considered gaseous pollutant in ppb, X is the number of gas molecules trapped on the cellulose filter (µmol) and R (constant gas) = 0.08206 L atm mol −1 K −1 .T is the average temperature during the sampling period (K), P is the mean atmospheric pressure during the sampling period, D is the  molecular diffusion coefficient of the specific gas in the atmosphere, t is the exposure duration time (s) and the ratio L/A = 47.5 m −1 is the air resistance coefficient for the IN-DAAF passive sampler (Adon et al., 2010).The passive samplers were prepared by the LA in Toulouse and dispatched in Abidjan (Côte d'Ivoire) from 15 December 2015 to 16 February 2016 during the dry season.For each set of passive samplers, blank samples were kept sealed in the laboratory and others sent to the sites.Passive samplers were exposed bi-monthly at the 21 measurement sites of the Abidjan district to represent integrated 2-week periods over the 2-month study period.Passive samplers were mounted on stainless steel rails in duplicate and placed at a height of at least 2.5 m above the ground, corresponding to the height of human exposure.On each rail, we installed duplicate passive samplers for each gaseous species measured (Fig. 2).Passive samplers were then exposed for 2 weeks before being replaced by new ones.Once removed, passive samplers were kept refrigerated before being shipped to the LA for analysis.Over the study period, 672 passive samplers were exposed and 2 were damaged.A total of 670 samples and 48 blanks (12 for each gas) were analysed (24 blanks kept in the laboratory and 24 other blanks sent in the same condition with exposed passive samplers).

Chemical analysis
To determine and calculate the concentration of studied gaseous pollutants, ionic concentrations of ammonium (NH + 4 ), nitrate (NO − 3 ), sulfate (SO 2− 4 ) and nitrite (NO − 2 ) are determined by ion chromatography (IC) after desorption of the filters in 10 mL (5 mL for the NH 3 ) of 18 M deionized water by ultrasonic stirring (15 min).The IC system used for this study has been extensively described in previous studies (Adon et al., 2010;Galy-Lacaux and Modi, 1998;Hodgkins et al., 2011).The reliability of the IC analytical results are assessed by the chemistry laboratory of the LA in Toulouse by performing a quality control inter-comparison program organized by the World Meteorological Organization/Global Atmospheric Watch (WMO/GAW) twice a year (Laouali et al., 2012).Results of the WMO/GAW quality assurance program for the year 2016 show that analytical precision is estimated to be ±5 % for all ions (http://www.qasac-americas.org/).The reproducibility of the INDAAF passive samplers was found to be 20, 9.8, 14.3, 16.6 and 10 % for HNO 3 , NO 2 , NH 3 , SO 2 and O 3 respectively.Detection limits for each trace gas were calculated from field blanks' filters and found to be 0.07 ± 0.03 ppb for HNO 3 , 0.2 ± 0.1 ppb for NO 2 , 0.7 ± 0.2 ppb for NH 3 , 0.05 ± 0.03 ppb for SO 2 and 0.1 ± 0.1 ppb for O 3 (Adon et al., 2010).For our present study, the detection limit was found to be 0.09 ± 0.00 ppb for HNO 3 , 0.37 ± 0.02 ppb for NO 2 , 1.4 ± 0.18 ppb for NH 3 , 0.01 ± 0.00 ppb for SO 2 and 0.08 ± 0.04 ppb for O 3 .Potential interference between HNO 3 and NO 2 passive samplers was estimated and found to be negligible.For the traffic site A 1 , our results presented in Supplement Table S1 show that nitrate (NO − 3 ) concentrations are very low compared to nitrite (NO − 2 ) ions, with a ratio around 2 %.Other NO y species such as peroxyacetyl nitrate (PAN) and nitrous acid (HONO), which can be deposited as nitrite and oxidize to nitrate, may www.atmos-chem-phys.net/18/5173/2018/Atmos.Chem.Phys., 18, 5173-5198, 2018 interfere with the determination of NO 2 and HNO 3 concentration (Cape, 2009).The choice of the coating solution in the INDAAF passive sampler aims to reduce this interference.
The average concentration of two samples is used in all the cases except when contamination of one sample is suspected and when the average concentration is under the detection limit.Thus, for the present work, 4 samples (0.6 %) were removed from the whole database for contamination and/or detection threshold, making a database of 666 samples representing about 99.4 % of the total analysed samples.

Meteorological data
Meteorological parameters used for this study were collected at the site (A 6 ), which is located at the weather station of the Félix Houphouët-Boigny International Airport (Airport FHB, Fig. 2).This weather station is managed by ASECNA (Agency for Safety of Air Navigation in Africa and Mada-gascar) and provides meteorological data for air navigation in the district of Abidjan.The meteorological data used in this study included wind direction and velocity, temperature and relative humidity.Daily variations in temperature and relative humidity during the study period (15 December 2015 to 16 February 2016) are presented in Fig. 3. Daily average temperature ranged between 25 and 29 • C, while mean relative humidity was about 78 %, with a maximum of 92 % during the study period.In addition, this period was particularly dry with only two rainy days and a total cumulative rainfall of 32 mm.The wind rose presented in Fig. 4 shows the predominance of southwest (SW) and north-northeast (NNE) wind directions.Wind velocity presents values ranging between 1 and 9 m s −1 , with an average of 2.75 ± 0.6 m s −1 .These observed meteorological parameters are characteristic of the major dry season in Côte d'Ivoire (Konate et al., 2016).

Spatial analysis
In this study, we interpolated the inorganic gaseous species' concentrations measured at the 21 sites to obtain the spatial distribution of gaseous concentrations of the entire city of Abidjan using geostatistical analysis (GIS).The mapping tool used was ArcGIS software version 10.2.2.The features were projected using World Geodetic System (WGS_1984) UTM_Zone_30N (Universal Transverse Mercator).Two interpolation methods are frequently used to study spatial distribution of pollutants.These are inverse distance weighting (IDW) and kriging.The general formula for both interpolators is formed as a weighted sum of the data Eq.( 2): where Ẑ (S 0 ) is the predicted concentration at S 0 location and is calculated as a linear weighted sum of N observations surrounding the predicted location.Z (S i ) is the measured concentration at the S i location, λ i is an unknown weight for measured concentration at the S i location, S 0 is the prediction location and N is the number of measured concentration.
In IDW, the weight λ i only depends on the distance between the measurement points and the prediction location.IDW computes predicted concentrations at the location as a weighted average of neighbouring measured concentrations (Ibrahim et al., 2012;Rivera-Gonzalez et al., 2015).This method provides a spatial distribution of pollutants much closer to reality.The IDW method showed better similarity between measured and interpolated concentrations of sulfur dioxide (SO 2 ) and nitrogen dioxide (NO 2 ) (Jha et al., 2011).
For the kriging method, the weight λ i is not only based on the distance but also on the overall spatial arrangement of the sampling sites.Kriging is the method of interpolation derived  (Oliver and Webster, 1990).

Data analysis
The empirical semivariograms provide information on the spatial autocorrelation of datasets.The semivariograms necessary for kriging were estimated for each pollutant and are presented in Fig. 5.

Cross-validation
Ordinary kriging with a Gaussian semivariogram model and IDW was used to perform the spatial interpolation.To validate the interpolation model, error analysis was carried out between measured and interpolated values.The free statistical software R was used to determine the mean absolute error (MAE), the root mean square error (RMSE) and the coefficient of correlation (R 2 ).The RMSE and MAE are regularly employed in model evaluation studies (Chai and Draxler, 2014).Lower MAE and RMSE values and a higher value of R 2 are used for better comparison (Jha et al., 2011;Rivera-Gonzalez et al., 2015).The critical value of the Pearson's correlation coefficient at the threshold of 5 % is R 2 = 0.42.Regression between interpolated data and observed data using IDW and the kriging method is presented in Fig. 6, while statistical parameters calculated for each interpolation method are presented in Table 3. Results indicate that the lowest MAE and RMSE values and the highest R 2 value are always obtained for the IDW interpolation method.Thus, the IDW was chosen to interpolate data and represent the spatial distribution of all gaseous pollutants.

Results and discussion
The average concentration measured every 2 weeks for each gaseous pollutant is shown in Table S2.The average concentrations calculated from the four 2-week periods were used to  4. The differences in the spatial distribution of the sites are examined in the following discussion.

Spatial distribution of gaseous pollutants in Abidjan
In order to present the spatial distribution of gaseous pollutants investigated in this work, the centre of the district of Abidjan is divided into a rectangle of 35 km × 30 km between 5  Besides the road traffic, the industrial sector seems to be another great contributor to the highest measured NO 2 concentration.Significant concentration levels around 20.9 ppb are measured over the three industrials sites (A 11 , A 9 and A 10 ).Abidjan currently has three industrial zones covering 885 ha, and the fourth is under construction.The biggest one is located in the municipality of Yopougon (645 ha), followed by the industrial areas of Vridi in the municipality of Port-Bouët (120 ha) and of Koumassi (120 ha) (MI, 2015).The industrial area of Vridi (A 11 ), where the highest NO 2 concentrations were measured (25 ppb), includes several activities related to the harbour area, shipyards, oil terminals, chimneys, a thermal power station and the processing and storage of petroleum product factories.With the exception of oil and shipping industries, most of these activities are also found in the other industrial areas.Other activities such as cement manufacturing, chemical industries and plastic production are also present.These industrial activities are well known to be the main source of anthropogenic nitrogen oxide emissions into the atmosphere (Krzyzanowski and Cohen, 2008).

Sulfur dioxide (SO 2 )
Figure 8 shows the spatial distribution of the SO 2 concentrations in Abidjan.SO 2 concentrations varied from a minimum of 0.4 ppb in the suburban site (A 20 ) to a maximum of 7.2 ppb measured at the A 3 traffic site (Table 2).The highest concentrations were generally obtained at the five traffic sites (A 3 , A 2 , A 4 , A 19 , A 5 ) and at the A 1 suburban site, with respective values of 7.2, 6.8, 4.7, 3.4, 3.1 and 2.6 ppb.We notice a decrease of SO 2 concentrations from the northwest of the city where the main roads (North Highway and East Source Road) and the industrial zone of Yopougon (A 9 ) are located to the southeast of Abidjan near the Atlantic Ocean.The majority of the sampling sites (71.5 %) show concentrations lower than 2 ppb.Similarly to NO 2 , the highest SO 2 concentrations are observed in areas of dense traffic and in industrial zones.It should be noted that refining companies located in the industrial zone of Vridi (A 11 ) do not yet have desulfurization units.This kind of production unit may constitute an important source of SO 2 (Rodriguez and Hrbek, 1999;Tam et al., 1990).The sulfur content of the fuel produced and marketed in Côte d'Ivoire is higher than 2000 part per million (ppm) (CCAC and UNEP, 2016).This content is more than 40 times higher than the standards applied in most developed countries (50 ppm or below).There is also intensive charcoal production through biomass burning in the northwest of the city.The large rubber plantations located in this area are regularly reconverted into charcoal using traditional ovens.According to the study NAMA (Nationally Appropriate Mitigation Action) on sustainable charcoal in Côte d'Ivoire, charcoal production increased by 22 % between 2003 (400 850 t) and 2012 (488 128 t).During the same pe-  (Bates et al., 1992;Andreae, 1991).The northeastern gradient could be also attributed to air masses' circulation in the dry season, with north-northeast harmattan winds shown by the wind diagram (Fig. 4).

Ammonia (NH 3 )
The main source of NH 3 emissions is biomass burning including biofuel combustion, agriculture, animal husbandry, NH 3 -based fertilizer, industrial processes and vehicular emission (Paulot et al., 2017;Whitburn et al., 2015;Behera et al., 2013;Akagi et al., 2011).In urban areas, domestic fires, traffic and industrial activity are generally considered as major sources of NH 3 (Sun et al., 2017;Teng et al., 2017;Perrino et al., 2002;Sapek, 2013;Sutton et al., 2000;Whitehead et al., 2007).In western African capitals, domestic fires and biomass burning are the main source of NH 3 emission (Liousse et al., 2014;Adon et al., 2010).Anthropogenic emissions of NH 3 from domestic fires in Bamako (Mali) and Dakar (Senegal) have been estimated for 2005 to be 1.6 × 10 −3 and 7.0 × 10 −4 TgNH 3 yr −1 respectively (Adon et al., 2016).In Abidjan, NH 3 concentrations were much higher compared to other gaseous pollutants.Figure 9 shows the spatial distribution of NH 3 in Abidjan.The domestic fires site (A 16 ) has the highest concentration of NH 3 (102.1 ± 9.1 ppb).This site is a mid-term instrumented monitoring site of the DACCIWA project located near a traditional smokehouse of food products.It was selected to investigate the concentrations of pollutants emitted by domestic fires and their contributions to this source.The second most significant concentration of NH 3 was measured at the site A 14 located on the east of the city and characterized by domestic   to poultry dejections (Gac et al., 2007).In the global NH 3 anthropogenic source budget, Bouwman et al. (2002) indicated that the animal excreta source represents 21.7 Mt yr −1 of the total 53.6 Mt yr −1 .In addition to breeding, this urban area combines large industrial and individual plantations of hevea and banana.These plantations often make excessive use of chemical fertilizers rich in urea, which therefore lead to ammonia emissions (Degré et al., 2001).A high NH 3 concentration was also measured at the waste burning site (A 15 ).This site is located in the village of Akouedo and classified as an uncontrolled dump site of garbage (Kouame et al., 2006).With a size of 153 ha, it has received all the vegetable, animal and industrial waste of the city of Abidjan since 1965.This huge waste is estimated at about 550 000 t per year (Kouame et al., 2006).Human waste and its management is a problem in Abidjan that can also explain the concentration levels obtained.To summarize, NH 3 concentrations are very high, which can be attributed to domestic fires, as well as bacterial decomposition associated with agriculture (pig and poultry farming) and waste dumps.

Nitric acid (HNO 3 )
HNO 3 is believed to be a major end product in the oxidation of gaseous nitrogen compounds.It is one of the mineral acid contributors to acid rain (Cogbill and Likens, 1974).
Figure 10 shows the spatial distribution of HNO 3 in Abidjan during the dry season.The study of the spatial distribution of HNO 3 in Abidjan showed that the concentrations ranged between 0.2 ± 0.0 and 1.4 ± 0.3 ppb.Approximately half of the sampling sites have concentrations between 1.1 ± 0.3 and 1.4 ± 0.3 ppb.These sites are mainly lo-cated in the centre and in the west of the city.The highest concentrations of HNO 3 were measured at a suburban site, A 17 (1.4 ± 0.3 ppb), a waste burning site, A 15 , and an industrial area, A 9 (1.3 ± 0.2 ppb), a residential site, A 13 , a domestic fires site, A 16 , and a residential district, A 7 (1.2 ± 0.3).
Most of these sites are located in the east of the city.This could be explained by the fact that gaseous HNO 3 is the result of several chemical transformation processes of NH 3 and NO x (NO 2 + NO) guided by the presence of oxidants such as H 2 O 2 , O 3 and OH radicals (Seinfeld and Pandis, 2016;Kumar et al., 2004;Hanke et al., 2003).The prevailing SW winds can also transport the precursor pollutants from the city centre in an easterly direction, and ageing in the air masses leads to the production of HNO 3 far from their source.

Ozone (O 3 )
O 3 is often associated in urban areas with adverse effects on human health and natural environment (Hagenbjörk et al., 2017).It is well known that this secondary pollutant depends on photochemical interactions of gaseous precursors composed of nitrogen oxides (NO x ) and volatile organic compounds (VOCs) (Duan et al., 2008;WHO, 2006).At a specific location, the level of O 3 concentration depends on emission of its precursors (VOCs and NO x ), the long-range transport of O 3 and meteorological parameters (Hagenbjörk et al., 2017).Two chemical regimes are associated with the rate of ozone production, the NO x -saturated (or VOC-limited) regime and the NO x -limited regime.In urban areas with high NO x concentrations (NO x -saturated) and a low ratio of VOCs / NO x , the ozone production rate is generally low and  Sadanaga et al., 2012).It ensures the conversion of NO to NO 2 and is therefore responsible for the net production of ozone (Camredon and Aumont, 2007).We also measured significant concentrations at two other peri-urban sites in the east of the city: the waste burning site of the village of Akouedo (A 15 ) and the suburban site of Bingerville (A 17 ), with O 3 concentrations of 17.55 and 16.76 ppb respectively.As in the case of HNO 3 , the highest concentrations measured in the east of the city may be due to the ageing of precursor pollutants transported in an easterly direction by SW prevailing winds.At the traffic sites (A 1 , A 2 , A 7 , A 9 ) located near the Banco forest, ozone concentration was found to be between 11.3 and 12.9 ppb.The lowest O 3 concentration in Abidjan was measured in the north of the city at the site A 4 (5.09ppb).

Mean gaseous concentrations of the main anthropogenic activities in Abidjan and comparison with other cities of developing countries
The study of the spatial distribution of gaseous pollutants in Abidjan reveals different anthropogenic contributions to the measured levels of pollution.Therefore, we have tried to investigate the relation between the concentration levels measured for each primary pollutant (NO 2 , SO 2 , NH 3 ) and the potential emission of each anthropogenic activity.Sampling sites were grouped by the activity sectors' source of pollution.Six different groups were formed.The main activities' sources of pollution have been identified in Sect.3.1 to be traffic, industrial, residential, domestic fires, waste burning and suburban.We calculated the mean concentration of each gaseous pollutant associated with each sector (Fig. 12).

Average concentration of primary gaseous pollutants associated with anthropogenic activities
Nitrogen dioxide (NO 2 ) The analysis shows that the industrial sector and traffic mainly explain the highest measured NO 2 concentrations, with an average concentration of 20.9 ± 2.8 and 17.8 ± 4.7 ppb respectively.However, domestic fires and res-  idential sites present very comparable levels of NO 2 concentrations of 17.9 ± 0.3 and 16.1 ± 2.4 ppb respectively compared to industrial and traffic sites.It should be noted that residential and domestic fires sites are strongly influenced by road traffic in Abidjan.The identified residential and domestic fires sites are never more than 10 m far from roadways.This could explain the near-similar concentration levels observed between traffic, domestic fires and residential sites.Waste burning (12.0 ± 0.0 ppb) and suburban (7 ± 2.4 ppb) sites present lower concentrations of NO 2 .Nevertheless, these concentrations are not negligible and can be due, on the one hand, to the limited traffic of the garbage collection vehicles and also on the other hand to the circulation of minibuses called "gbaka", which connect the city centre to the suburbs.

Sulfur dioxide (SO 2 )
The road traffic sector is the main contributor to the SO 2 pollution levels measured in Abidjan (4.2 ± 2.2 ppb).This result seems to be a good indicator of the quality of fuels used in Abidjan.In countries without stringent fuel policies such as Côte d'Ivoire, diesel sulphur content is generally significant.The average concentration of SO 2 in suburban and industrial sites is 2.0 ± 1.4 and 1.7 ± 0.1 ppb, respectively.
Ammonia (NH 3 ) NH 3 concentrations are significantly higher at domestic fires sites, with an average concentration 2 to 4 times higher than for other sites (84.9 ± 17.2 ppb).This very high value highlights the use of biomass burning (firewood and charcoal) as a source of energy by most households of Abidjan.The high standard deviation is due to the location of two of the domestic fires sites near the source, while the other domestic fires sites are much further away from the source.The other groups of sites representative of waste burning (39.1 ppb) and traffic sources (27.2 ± 3.1 ppb) present NH 3 concentrations lower than 40 ppb.Peri-urban agriculture and intensive livestock in the Abidjan suburbs are mainly responsible for the suburban areas' NH 3 concentration (22.6 ± 10.3 ppb).This value is close to concentrations measured at residential (21.9 ± 5.2 ppb) and industrial (20.7 ± 0.7 ppb) sites.Ammonia concentrations measured in Abidjan are very high compared to the concentrations measured in rural African environments, where concentrations range from 1 to 9 ppb (Adon et al., 2010;Carmichael et al., 2003;Fattore et al., 2014;Martins et al., 2007), and are similar to concentrations found near agriculture sources.

Average concentrations of gases compared with other urban sites
In order to contextualize the gaseous pollution measured in Abidjan in a subregional and international context, in Table 5 we have listed some experimental studies using passive samplers to obtain a comparable measurement technique and also comparable temporal integration.Table 5 presents a comparison between gas concentrations measured in Abidjan with literature data, with a focus on primary pollutants (NO 2 , SO 2 and NH 3 )

Traffic and urban sites
The average concentration of NO 2 in Abidjan (17.8 ppb) is more than 3 times higher than that measured in St John's in India (4.8 ppb) (Kumar et al., 2004).This concentration in Abidjan is comparable to those measured in Bamako (16.2 ppb) and Shenzhen (20.7 ppb) reported by Adon et al. (2016) and Xia et al. (2017).However, it remains lower than the concentrations obtained in Dakar (31.7 ppb) and Al Ain in the Middle East (31.5 ppb) and reported respectively by Adon et al. (2016) and Salem et al. (2009).
The average concentration of NH 3 measured at the Abidjan traffic site (27.2 ppb) in this study is higher than those obtained in Dakar (21.1 ppb), Al Ain (17.1 ppb) and St John's (9.5 ppb), respectively reported by Adon et al. (2016), Salem et al. (2009) and Kumar et al. (2004).Adon et al. (2016) also reported a concentration of NH 3 in Bamako (46.7 ppb) 2 times higher than that obtained in Abidjan.
The average SO 2 concentrations at the traffic and urban sites range from 1.56 to 15.9 ppb.In comparison with the Abidjan traffic site (4.2 ppb), the average SO 2 concentrations in Bamako (3.6 ppb) and St John's (1.56 ppb) are much lower.The concentration of SO 2 in Abidjan is comparable to those measured at the polluted urban sites of Shenzhen (4.9 ppb) in China and Al Ain (5.8 ppb) in the United Arab Emirates.The SO 2 level in Dakar is the highest, with a concentration of 15.9 ppb (Adon et al., 2016) It highlights significant differences in the sources of gaseous pollutants and their level of concentrations in western African and other African cities and those in developing countries.

Suburban sites
The average concentration of NO 2 at the Abidjan suburban site (7 ppb) is more than 2 times higher in comparison with that measured at the Amersfoort suburban site (2.5 and 2.9 ppb), respectively reported by Martins et al. (2007) and Lourens et al. (2011).The concentrations of SO 2 in Abidjan (2.2 ppb) are lower than those measured at Amersfoort suburban site in South Africa (2.8-5.5 ppb).The concentration of NH 3 in Abidjan is the highest, at 22.6 ppb.

Overview of urban NO 2 and SO 2 monitoring studies in African cities
Air quality monitoring networks are very important in assessing the health risk associated with exposure to gaseous pollutants.They also help governments to take measures of mitigation of the levels of pollutant concentrations and to define air quality indices adapted to each country.If some synthesis of air quality measurements for particulate matter concentrations exists in the literature at the scale of Africa (Petkova et al., 2013), there is no overview for gaseous pollutants.Nowadays, it is quite difficult to provide an overview of air quality results for gaseous pollution over Africa because of the availability of only some non-perennial studies carried out within the framework of scientific research programs.The available results are also not always communicated publicly or found in the literature.As this study shows, one of the main gaseous sources of air pollution in African cities is traffic roads emitting NO 2 and SO 2 .This section aims to report the main studies on NO 2 and SO 2 conducted at different timescales (hourly, daily, weekly, monthly and annually) in African urban and industrial environments.This is the result of literature searches based on studies conducted in Africa between 2005 and 2017.Publications in which methodology or data reports were clearly described were selected and listed in Tables S3 and S4 respectively for NO 2 and SO 2 .Recent and unpublished studies have also been presented for comparison.Figure 13 presents the African countries, coloured by UN subregion, for which NO 2 and SO 2 monitoring studies were identified and included in this work.We found NO 2 and SO 2 measurements conducted in at least 20 African countries.Most of those measurements were located in seven western African countries (Senegal, Mali, Burkina Faso, Côte d'Ivoire, Nigeria, Ghana and Benin) and in five eastern African countries (Sudan, Kenya, Ethiopia, Uganda and Tanzania).Studies were also identified for four countries (Algeria, Tunisia, Morocco and Egypt) in northern Africa and three countries (South Africa, Angola and Mozambique) in southern Africa.Only one study was identified in central Africa (Cameroon).To compare the levels of gaseous concentration in Africa with WHO air quality guidelines, we present the concentration level of each pollutant for each country using the mean, the minimum and the maximum values when available.Figures 14 and 15 represent NO 2 and SO 2 levels in African cities as reported in the various studies.Daily or hourly routine NO 2 and SO 2 do not exist in most of the African cities.In most of the studies we found, monitoring campaigns were carried out with different methods (passive sampling, real time analysers or satellites) and over short periods.Thus, only few studies allow the comparison with WHO standards.
Nevertheless, this section represents, to the best of our knowledge, the state of the art of urban NO 2 and SO 2 concentrations measured in African cities and leads to a contextualization of our results in Abidjan.

Nitrogen dioxide (NO 2 )
Short time exposure (1 h) at a concentration of 200 µg m −3 of NO 2 is toxic and has significant health effects.The three studies at this timescale performed by Jackson (2005) in Tanzania and Adon et al. (2016) in Senegal and Mali show that hourly NO 2 concentration range from 18 to 53 µg m −3 , 9.4 to 150 µg m −3 and 15.8 to 135 µg m −3 in Dar es Salam, Dakar and Bamako respectively (Fig. 14).These levels of concentration are below the WHO threshold.
The WHO air quality guideline value for annual exposure to NO 2 concentration is 40 µg m −3 .Seven studies were carried out during at least 1 year in Dakar (Senegal), Ouagadougou (Burkina Faso), Bamako (Mali), Abidjan (Côte d'Ivoire), South Africa and Ben Arous (Tunisia) (Julien Bahino, personal communication, 2017;Chaaban, 2008;Demay et al., 2011;Josipovic et al., 2010;Nanaa et al., 2012).Almost 19 studies were conducted at daily, weekly and monthly scales.Eight studies were conducted on a daily basis.Average 24 h concentrations ranged from 0.1 µg m −3 along Juja Road in Nairobi (Kaboro, 2006) to 752 µg m −3 on the traffic site of Dantokpa market in Cotonou (Mama et al., 2013).Only three studies measuring NO 2 on a weekly scale were identified.The maximum concentration (1015.34µg m −3 ) was measured in the city of Oweri, the capital of the oil-rich region of Imo State in Nigeria (Ibe et al., 2016).Except this extreme value, concentrations range from 9.32 µg m −3 in urban sites of Kampala (Uganda) to 60 µg m −3 in low-income areas of Accra (Ghana).
Average monthly concentration ranges from 2.6 to 87.2 µg m −3 for eight studied sites.The lowest values were measured in Maputo (Mozambique) in a residential site and in a green area.The monthly average concentration in Maputo is 9.13 µg m −3 and varies between 2.6 and 21 µg m −3 (Cumbane et al., 2008).The highest monthly average concentration of NO 2 was measured in Bamako.
In our study, concentrations measured at traffic and industrial sites in Abidjan (33.5 and 39.3 µg m −3 respectively) are of the same order of magnitude as the values obtained by Liousse and Galy-Lacaux (2010) at urban sites in Abidjan (29.8 µg m −3 ) and Yaoundé (33.3 µg m −3 ).Similarly, annual NO 2 concentrations measured in Abidjan (Bahino, 2017) in the DACCIWA mid-term experiment (2015-2017) present values about 25 µg m −3 for the traffic site (A 1 ) that are in the same range of other African capitals studied.However, the table emphasizes the lack of NO 2 measurements in Abidjan at hourly and annual scales allowing the comparison with WHO thresholds.It would therefore be wise for public authorities in Abidjan as well as in other African cities to carry out regular studies at comparable timescales and with recommended methods to obtain reliable results.

Sulfur dioxide (SO 2 )
WHO indicates that changes in pulmonary function and respiratory symptoms appear after a 10 min exposure period to SO 2 at a concentration of 500 µg m −3 .No study carried out with a minimum time interval of 10 min was reported in the literature.
Twenty-four hour SO 2 levels are significantly associated with daily mortality rates (Burnett et al., 2004).The WHO 24 h guideline for SO 2 is 20 µg m −3 .Nine daily (24 h) studies are reported in Fig. 14.Measurements in Cotonou (Benin) have been performed at the crossroads of Dantokpa market, characterized by a large circulation of two-wheeled vehicles.Daily average concentrations range between 784.8 and 3662.4 µg m −3 (Mama et al., 2013).These concentrations are 39 to 183 times higher than the threshold value.In Dakar (68.54 µg m −3 ), Cairo (34 µg m −3 ) and Bamako (29.03 µg m −3 ), SO 2 mean daily concentrations are greater than 20 µg m −3 .It is found that only two studies conducted in Marrakesh (Morocco) and Tunis (Tunisia) have concentration ranges lower than the threshold.In Nairobi and Khartoum the maximum concentration can reach and exceed the threshold.
Approximately six studies report weekly and monthly concentrations of SO 2 (Fig. 14).Generally, weekly and monthly SO 2 concentrations in Africa range from 0.77 to 18.9 µg m −3 .The weekly concentrations measured in the oil-rich region of Imo State (Nigeria) are not in this range (1203 to 1465 µg m −3 ) (Ibe et al., 2016).
The annual average concentrations of SO 2 are reported for 11 studies.The concentrations measured in northern Africa are the highest.Average levels of SO 2 range between 8 µg m −3 in Rabat (Morocco) and 325 µg m −3 in Alger (Algeria).In western Africa, concentrations are lower and range from 0.5 µg m −3 in Ouagadougou (Burkina Faso) to 80 µg m −3 in Dakar (Senegal).The annual average SO 2 concentration in Abidjan (3.66 µg m −3 ) is one of the lowest of western African cities reported in this study (Julien Bahino, personal communication, 2017).Measurements of SO 2 at smaller timescales (i.e. 10 min, 24 h) do not exist in Abidjan or are not published.As for NO 2 , studies should be carried out regularly and according to WHO standards.
Air pollution from human activities has been a longstanding environmental issue.Human activities have led to an unprecedented increase in atmospheric abundance of key air pollutants.High concentrations of ozone and PM adversely impact human health by increasing the risk of a variety of diseases.The high concentrations of gaseous pollutants measured in Abidjan could be responsible for respiratory diseases, which could result in significant costs to human health and the economic capacity of the local workforce.Environmental changes, including air pollution, have already significantly increased the burden of cancer in western Africa in recent years (Val et al. 2013).Premature death in the world in 2010, resultant of air pollution, leads, according to the World Health Organisation, to a combined estimate of more than 7 million premature deaths.
In addition to health impacts, anthropogenic pollutants that react with biogenic emissions can also lead to increased ozone and acid production outside urban areas (Marais et al., 2014), with adverse effects on precipitation and cloudiness that can affect animals, land area, ecosystems and crops and affect many other important socio-economic factors.The worldwide economic loss associated with these crop losses is estimated to represent an economic value of USD 14-26 billion for the year 2000 (van Dingenen et al., 2009).
Data provided by this kind of study can help as follows: to study the expected changes in future air quality in response of various forcings such as changes in the emissions due to urbanization and industrialization, and climate change; to estimate the immediate impacts (e.g.human health effects, crop damage) and long-term effects (e.g.chronic health impacts, deposition on the biosphere, climate impacts) of air pollution (Knippertrz et al., 2015).

Conclusion and recommendations
Abidjan is the economic capital of the western African subregion.The new urban master plan of the city foresees its extension and the construction of many types of infrastructure, which will certainly increase environmental and sanitary problems.The present study reports the gaseous concentrations' measurements using passive samplers performed during an intensive experimental field campaign in the framework of the Work Package 2 of the European DACCIWA project.This work presents an original database of bi-monthly gaseous concentrations (NO 2 , SO 2 , NH 3 , HNO 3 , O 3 ) in the district of Abidjan measured at 21 representative sites.This database allows the levels of gaseous concentration to be characterized and a spatial distribution of gaseous pollution at the scale of the city during the dry season (15 December 2015 to 16 February 2016) to be presented for the first time.
Our results show that there is a great spatial variability of gaseous pollutants in Abidjan.The average concentrations of the main gaseous pollutants range from 2.7 to 25 ppb for NO 2 , 9.1 to 102.1 ppb for NH 3 , 0.2 to 1.4 ppb for HNO 3 , 0.4 to 7.21 ppb for SO 2 and 5.1 to 19.1 ppb for O 3 .
The spatial distribution of gaseous pollutants has been studied according to the main potential sources of pollution nearby the measurement sites.Results show that the concentration level of gaseous pollutants such as NH 3 and NO 2 is very significant at some sites.On the one hand, it highlights the predominance of the domestic fires source with the use of wood and charcoal that emits a large amount of NH 3 .On the other hand, it shows the significant impact of traffic roads on NO 2 and SO 2 emissions.
The average gas concentrations per source of pollution, showed that industrial (20.9 ppb), domestic fires (17.9 ppb) and traffic (17.8 ppb) sites mainly contribute to NO 2 pollution.Concerning NH 3 , domestic fires and waste burning sites constitute by far the main sources of pollution, with an average concentration of 84.9 and 39.1 ppb respectively.In addition to the high concentration of NO 2 , the traffic sites also have the highest SO 2 average concentration (4.2 ppb).
In this study, for the first time we propose an overview of gaseous SO 2 and NO 2 pollutants, which allows a global picture of gas concentrations to be given at the scale of several African cities.It includes 22 publications on NO 2 and SO 2 measurements conducted in 20 African countries during the period 2005-2017.This study indicates that gaseous pollution in Abidjan is comparable to other western and central African cities and also emphasizes the lack of air quality monitoring networks.Only a few studies carried out by research teams within the framework of scientific programs with various objectives give an idea of the state of air quality.These studies used different methods and timescales that do not allow comparison with WHO standards and threshold values.The comparable studies show that the hourly concen-tration of NO 2 in African cities (9.4-135 µg m −3 ) is lower than the WHO standards.Abidjan annual NO 2 concentration is 23.25 µg m −3 , with a range on the scale of African cities from 2 to 175 µg m −3 .Annual NO 2 WHO standards are exceeded in Dakar and Tunis.The daily concentration of SO 2 in African cities ranges from 0.2 to 3662 µg m −3 and exceeds the daily SO 2 WHO standard (40 µg m −3 ) in most of the African cities, except in Marrakesh and Tunis.
These scarce datasets should draw the attention of African public authorities to the urgent need of measuring NO 2 and SO 2 emissions.The results indicate the need of other studies including VOCs and NO x measurements to better understand the production processes of secondary pollutants such as O 3 .In the framework of the DACCIWA WP2 (Air Pollution and Health) program, bi-monthly measurements have been ongoing since December 2014 over three supersites.These measurements will make it possible to estimate the impact of pollutants' concentrations on the population's health through the calculation of real dose-response functions.To guarantee the continuity of the work done in the DACCIWA Work Package 2 and to produce health risk maps for the different combustion sources specific to western Africa based on modelling, the university Félix Houphouët-Boigny of Abidjan will manage a new project called PASMU (Pollution de l'Air et Sante en Milieu Urbain), starting in 2018.All the results presented in this work will help in the development of new projects and will serve to develop the experimental strategy of such a network.In addition, meteorological parameters will be included to finally lead to modelling studies to predict air quality for Africa on regional to urban scales.
Data availability.The pollution data used in this study are original and are publicly available in Table S2 in the Supplement.
Competing interests.The authors declare that they have no conflict of interest.Special issue statement.This article is part of the special issue "Results of the project "Dynamics-aerosol-chemistry-cloud interactions in West Africa" (DACCIWA) (ACP/AMT inter-journal SI)".It is not associated with a conference.

Figure 1 .
Figure 1.Spatial distribution of the 21 sampling sites (A 1 -A 21 ) and major sources of air pollution nearby in the district of Abidjan.

Figure 7 .
Figure 7. Spatial distribution of NO 2 in the district of Abidjan during the dry season (15 December 2015-16 February 2016).Measured concentrations in ppb are in brackets.

Figure 8 .
Figure 8. Spatial distribution of SO 2 in the district of Abidjan during the dry season (15 December 2015-16 February 2016).Measured concentrations in ppb are in brackets.

Figure 9 .
Figure 9. Spatial distribution of NH 3 in the district of Abidjan during the dry season (15 December 2015-16 February 2016).Measured concentrations in ppb are in brackets.

Figure 10 .
Figure 10.Spatial distribution of HNO 3 in the district of Abidjan during the dry season (15 December 2015-16 February 2016).Measured concentrations in ppb are in brackets.

Figure 11 .
Figure 11.Spatial distribution of O 3 in the district of Abidjan during the dry season (15 December 2015-16 February 2016).Measured concentrations in ppb are in brackets.

Figure 12 .
Figure 12.Concentration of gaseous pollutants by anthropogenic sectors of activity in the district of Abidjan.

Figure 13 .
Figure 13.African countries coloured by UN subregion for which NO 2 and SO 2 studies were identified and included in this publication.

Figure 14 .
Figure 14.NO 2 concentration levels in µg m −3 in African capitals at different timescales (hourly, daily, weekly, monthly and annually) as reported in literature.Green bars represent lower and upper range of means if reported.Black points represent average concentration of NO 2 .Red vertical lines illustrate current WHO guidelines at different timescales.

Figure 15 .
Figure 15.SO 2 concentration levels in µg m −3 in African capitals at different timescales (hourly, daily, weekly, monthly and annually) as reported in literature.Green bars represent lower and upper range of means if reported.Black points represent average concentration of SO 2 .Red vertical lines illustrate current WHO guidelines at different timescales.

Table 1 .
Geographical coordinates of sampling sites and majors source of air pollution nearby.

Table 2 .
Parameters of empirical semivariogram for each pollutant.
The results for NO 2 , SO 2 , HNO 3 and O 3 show that the Gaussian semivariogram model is fitted to empirical values.The parameters of the empirical semivariogram are presented in Table2.The empirical semivariogram of NH 3 shows a nugget model.It represents the discontinuity at the origin due to small-scale variation and the lack of spatial autocorrelation of NH 3 concentration.

Table 3 .
Statistical parameters calculated for each interpolation method.

Table 4 .
Gaseous pollutants' concentrations and standard deviation (SD) at 21 sampling sites in Abidjan during the dry season (15 December 2015-16 February 2016).Min Max Mean ± SD Min Max Mean ± SD Min Max Mean ± SD Min Max Mean ± SD Min Max Mean ± SD

Table 5 .
Gaseous pollutants' concentrations (ppb)at different sites of Abidjan and in other cities of developing countries.