Black carbon emissions from traffic contribute substantially to air pollution in Nairobi, Kenya

Air pollution is a leading environmental health hazard worldwide, with approximately one million premature deaths per year in Africa linked to ambient air pollution. Rapid, unplanned urbanization in Sub-Saharan Africa (SSA), combined with inadequate infrastructure and policies, is driving increasing pollution levels. Despite large health and climate impacts, SSA air pollution remains understudied, with few observation sites and a scarcity of long-term measurements, which are essential for assessing impacts. Existing data indicate high and rising pollutant levels, and projections suggest drastic increases in anthropogenic emissions. In Nairobi, elevated exposure to PM2.5 and black carbon (BC) poses significant health risks, but mitigation is hindered by limited understanding of source contributions. BC, a toxic and climate-warming component of PM2.5, originates from fossil fuel combustion and biomass burning. Bottom-up emission inventories carry large uncertainties in SSA due to poorly constrained activity data and transferability of emission factors. Radiocarbon (14C) analysis offers an effective, source-specific tracer to apportion fossil versus biomass BC, as fossil BC is 14C-dead and biomass-derived BC carries contemporary 14C signatures. This approach, robust to atmospheric processing, has been applied elsewhere but not yet in SSA. This study applies radiocarbon source apportionment over a full year in Nairobi, quantifying the relative fossil and biomass contributions to BC and assessing seasonal variability, alongside PM2.5 composition measurements.

The paper synthesizes prior evidence that SSA cities experience rapid growth in air pollution amid limited monitoring infrastructure and short-duration campaigns, hampering robust health and environmental impact assessments. Bottom-up emission estimates in SSA are highly uncertain due to poorly constrained fuel-use activity data and emission factors developed for other regions that may not apply locally. Previous studies highlight high PM2.5 and BC exposures in Nairobi and other SSA cities, with indications of increasing trends. Radiocarbon (14C) analysis has been established as a reliable method for differentiating fossil versus biomass sources of BC in remote regions and polluted cities outside SSA, overcoming ambiguities of conventional chemical tracers. However, before this study, 14C-based BC source apportionment had not been conducted in SSA, leaving a significant knowledge gap.

Study area and period: An urban background site was established on a rooftop at the University of Nairobi (1.279°S, 36.817°E; 1690 m a.s.l.; ~17 m above ground) in a park-like environment near the city centre, away from direct local sources (industrial sites, traffic hotspots, dumpsites). The surrounding roads have low vehicle density with restricted public service vehicles and heavy trucks. Sampling covered March 2014 to February 2015. Sampling: A high-volume sampler (Digitel DH-77, PM2.5 inlet) operating at ~30 m3 h−1 collected 24-hour quartz fibre filter samples (Millipore, 150 mm; prebaked 450 °C, 6 h) every 6th day to represent all weekdays. Monthly field blanks were collected. In total, 66 filters were sampled. Gravimetric mass: PM2.5 mass was determined gravimetrically (pre/post-weighing) in a temperature and humidity-controlled room (20 ± 1 °C; 40 ± 5% RH) after 24 h equilibration. Chemical analyses: Water-soluble inorganic ions (WSII) were analysed by ion chromatography (Dionex Aquion; established protocol). Elemental carbon (EC; mass-based BC) and organic carbon (OC) were measured with a thermal-optical transmission (TOT) analyser (Sunset Laboratory) using the NIOSH 5040 protocol. Instrument calibration used sucrose standards; analytical performance verified with NIST SRM 8785. OC values were blank-corrected using field blanks (0.9 ± 0.3 µg cm−2; ~0.02 µg m−3). No BC was detected in blanks (n=13). Triplicate analyses yielded mean RSD of 3% for OC and 2% for EC (instrumental errors 5% and 6%, respectively). WSII average relative s.d. was <5% for all ions. Organic aerosol mass was estimated as OA = 2.1 × OC. Radiocarbon source apportionment: Approximately every second filter (n=28) was processed to isolate BC using a modified Sunset TOT protocol. BC was thermally separated from OC, combusted to CO2, purified with Ag and Mg(ClO4)2 traps, cryo-trapped in liquid N2, and sealed in ampoules with Ag and CuO, then combusted (400 °C, 6 h) to remove impurities. 14C of CO2 was measured by AMS at Uppsala University. Fractional fossil versus biomass contributions were computed by isotopic mass balance: Δ14Csample = fbiomass × Δ14Cbiomass + (1 − fbiomass) × Δ14Cfossil, with Δ14Cfossil = −1000‰ and a regionally parameterized SSA Δ14Cbiomass = +57 ± 52‰ for 2015. Potential carryover of pyrolyzed carbon in the helium phase could shift Δ14C of BC by up to ~30‰, within measurement uncertainty. Meteorology and transport: Local meteorology (wind, temperature, rainfall) was obtained from Jomo Kenyatta International Airport and GDAS. Five-day air mass back trajectories were computed every 6 h using NOAA HYSPLIT v4 with GDAS 1°×1° fields, arrival height 100 m AGL (1890 m a.s.l.). Fire detections were retrieved from NASA FIRMS (MODIS).

• Annual mean PM2.5 at the urban background site was 27 ± 6 µg m−3, exceeding the 2021 WHO annual guideline (5 µg m−3) and the 24-hour limit (15 µg m−3) on all sampling days, with limited seasonality (slightly higher in dry periods). • PM2.5 composition: Carbonaceous aerosols dominated (64 ± 11%). Organic aerosol (OA = 2.1 × OC) contributed 49 ± 7% of PM2.5; black carbon (BC) contributed 15 ± 4%. Water-soluble inorganic ions (WSII) contributed 13 ± 5%, dominated by sulfate at 1.8 ± 0.8 µg m−3 (7 ± 3%). About 25 ± 5% of PM2.5 mass was unaccounted for (likely aerosol water and crustal/elemental components). • Sea-salt contribution to WSII was <2% (Na+ based), with Na+ correlating with Mg2+ (R2 = 0.79, p < 0.01). SO42− and NH4+ were strongly correlated (R2 = 0.75; R2 = 0.84 after removing one outlier), with a molar NH4+/SO42− slope of 2.2, indicating ammonium sulfate formation. • BC concentrations were persistently elevated at 3.9 ± 1.2 µg m−3 with no clear seasonal trend, comparable to previous Nairobi background measurements and higher than typical urban background levels in Europe and North America. The BC/PM2.5 ratio (~15%) was high relative to many cities globally. • Seasonal statistics (mean ± sd): PM2.5 (µg m−3): Spring 28 ± 4; Summer 30 ± 5; Fall 28 ± 8; Winter 22 ± 4. BC (µg m−3): 4.2 ± 1.2; 3.9 ± 0.9; 3.6 ± 1.2; 3.9 ± 1.5. • Radiocarbon source apportionment showed highly 14C-depleted BC (Δ14C ≈ −840 ± 34‰) throughout the year, implying dominant fossil sources. The mean fossil fraction was 85 ± 3%, corresponding to fossil BC 3.4 ± 1.1 µg m−3 and biomass BC 0.6 ± 0.1 µg m−3. No discernible seasonality in fossil/biomass fractions was observed. • Source marker ratios (e.g., K+/EC and NO3−/EC) indicated some influence of regional biomass burning during the boreal summer dry period when air masses overlapped with fire regions, but this influence did not materially affect the 14C-inferred BC source fractions. • The meteorological context showed limited seasonality in local parameters; back trajectories reflected seasonal shifts consistent with the ITCZ, and satellite fire data indicated potential periods of long-range transport during March–November, yet fossil BC remained dominant.

By applying radiocarbon apportionment to BC over a full year, the study directly quantifies that fossil fuel combustion overwhelmingly dominates BC in Nairobi (85 ± 3%), addressing the key question of source contributions in an SSA megacity. The constancy of Δ14C and fossil fraction across seasons, despite periods of regional biomass burning upwind, implies BC sources are predominantly local and persistent. Together with a high BC/PM2.5 ratio (~15%) and elevated absolute BC levels (3.9 ± 1.2 µg m−3), the results suggest a unique urban pollution profile in SSA, with BC being a particularly severe and toxic component. The findings align with Nairobi’s heavy traffic congestion, an older, less efficient vehicle fleet, high fuel consumption in the transport sector, and potential altitude-related combustion inefficiencies, pointing to traffic as the primary source of fossil BC. Although biomass burning leaves signatures in certain ion ratios (e.g., K+/EC), the lack of a corresponding shift in the 14C of BC indicates that regional fires during the study year did not substantially alter background BC source fractions, likely due to differing emission factors and atmospheric processing/fate of various tracers. Comparisons with other regions show similarly high fossil BC fractions in Europe and North America but at lower concentrations; East Asian cities show comparable concentrations and high fossil fractions; South Asian cities can exhibit clear seasonal biomass impacts not observed here. Overall, the results underscore the need to target traffic emissions to reduce BC exposure and associated health and climate impacts.

This year-long study provides the first radiocarbon-based source apportionment of BC in an SSA megacity, showing that fossil fuel combustion, most likely traffic, contributes about 85% of BC in Nairobi with minimal seasonal variability. PM2.5 levels are consistently above WHO guidelines, with carbonaceous aerosols dominating mass and BC making up an unusually large fraction (~15%). The evidence points to persistent local fossil sources as the key driver of BC pollution. Policy-relevant implications include prioritizing BC mitigation via improved public transport, non-motorized transport infrastructure, vehicle emission and fuel standards, and fleet modernization. Given the high toxicity and climate-warming properties of BC, focusing on BC can yield co-benefits for health and climate. Future work should extend 14C-based BC source apportionment to other SSA cities, enhance spatial coverage within cities (including hotspots), improve temporal resolution, and integrate with detailed emissions and receptor modeling to better inform targeted interventions.

• Single urban background site in Nairobi; findings may not fully capture intra-city variability or hotspot exposures. • Sampling every 6th day and 14C analysis on roughly half the samples (n=28) limits temporal resolution for source apportionment. • Potential carryover of pyrolyzed carbon during thermal separation could shift Δ14C of BC by up to ~30‰, though within reported uncertainties. • Biomass endmember uncertainty (Δ14Cbiomass = +57 ± 52‰) introduces uncertainty into fbiomass/ffossil estimates. • While marker ratios suggest some influence from regional fires, differences in emission factors and atmospheric processing may decouple these markers from BC and limit detectability in 14C. • The study period (2014–2015) represents one annual cycle; interannual variability was not assessed.