Increased contribution of biomass burning to haze events in Shanghai since China's clean air actions

China’s PM2.5 pollution has decreased ~30% from 2013 to 2017 following clean air actions, but the changing source contributions of carbonaceous aerosols, especially black carbon (BC) and water-soluble organic carbon (WSOC), remain unclear. Severe winter haze (daily PM2.5 ≥ 100 µg m−3) persisted during 2018–2020 across North, Central and East China, with Shanghai frequently downwind of a north–central–east pollution belt. Carbonaceous aerosols are key PM2.5 components and exert strong climate forcing. The research question is how BC source contributions and geographic origins in Shanghai have changed post-clean-air actions, particularly during winter haze. The study aims to quantify seasonal BC sources (biomass vs. fossil; further split into liquid fossil vs. coal) and their provincial/sectoral origins by combining dual-carbon isotope constraints (Δ14C and δ13C) with chemical transport modelling. Understanding these contributions is critical for effective air quality management, regional haze mitigation, and climate impact reduction.

Prior work documents the health and climate impacts of BC and carbonaceous aerosols and shows large historical emissions in China. Emission inventory (EI) studies reported substantial reductions in fossil-fuel emissions due to controls targeting power, industry, and transport, but uncertainties remain, especially for residential sources. Earlier isotopic studies in East Asia (Shanghai, Beijing, Tianjin, Zhejiang) during 2013–2014 winter generally found lower biomass fractions in BC (~20–33%) compared to this study’s winter. Studies in Delhi and the Sichuan Basin reported higher biomass contributions in some seasons. Transport modelling (e.g., FLEXPART) has been used to evaluate BC source regions and validate EIs, including in Arctic studies. Observational δ13C/Δ14C approaches provide robust separation of fossil versus contemporary sources and sub-division of fossil (liquid vs. coal) for BC, and have been applied in China in earlier years but with limited seasonal coverage after stringent emission controls.

Study design and sites: Year-round PM2.5 sampling was conducted simultaneously at suburban Chongming Island (Yangtze River Estuary, YRE) and urban Shanghai from December 2018 to January 2020. Eighty samples were collected at YRE and eighty-six in urban Shanghai using high-volume PM2.5 samplers on pre-combusted quartz-fiber filters. Concentration measurements: PM2.5 mass was determined gravimetrically after 48 h equilibration (RH 50%, 25°C). OC and BC (EC) were measured using a DRI Model 2001 thermal/optical carbon analyzer with the IMPROVE_A protocol. WSOC was obtained from water extracts (ultrasonication, centrifugation, 0.2 µm filtration), decarbonated, and analyzed by high-temperature catalytic oxidation (vario TOC select). Water-soluble ions (e.g., K+) were measured using ion chromatography (DIONEX ICS-5000+). Blanks were negligible (<0.5% of signals), and triplicate RSD <5% for BC, TC, WSOC, and K+. Isotope analyses: Selected high-loading samples were isolated for isotopes: BC (16 samples total: 13 YRE across four seasons and 3 urban winter haze), WSOC (17 YRE: 7 winter haze, 10 other seasons), and TC (3 urban winter haze). Filters were acid-fumigated (2 M HCl, 24 h) and dried to remove carbonates. BC was isolated using the DRI instrument with IMPROVE_A TOR; average BC recovery ~87%. WSOC extracts were freeze-dried and encapsulated. δ13C was measured by IRMS (precision ±0.2‰), Δ14C by AMS (MICADAS and 0.5 MV XCAM), with corrections for fractionation and post-1950 decay; OxII standard precision better than 10‰. Δ14C endmember for fossil set to −1000‰; contemporary biomass Δ14Cbio ≈ +70 ± 35‰, reflecting dominant fresh biomass in China. Biomass vs fossil fractions (fbio for WSOC/TC, fbb for BC) were computed using isotopic mass balance. Monte Carlo (MC) source apportionment: Dual-isotope (δ13C, Δ14C) of BC was used with MC simulations (100,000 iterations) to estimate posterior PDFs of source fractions for biomass (fbio), liquid fossil (fliq fossil), and coal (fcoal) subject to mass-balance, using literature endmembers for δ13C and Δ14C. Means and SDs were derived from posterior distributions. Transport modelling (FEG): The FLEXPART v10.4 Lagrangian model ran in backward mode for each sampling period from YRE, extending 30 days, driven by ECMWF 1° meteorology, including dry deposition and wet scavenging. A lognormal particle size distribution centered around 250 nm diameter was used. Anthropogenic BC emissions were from ECLIPSE v6B (GAINS-based) with explicit biofuel vs fossil split; agricultural waste burning and wildfires from GFED v4.1s (0.5°, daily). The coupled FLEXPART–ECLIPSE–GFED (FEG) system simulated BC concentrations, sectoral contributions (residential, transport, industry, energy, waste, open fire, shipping, gas flaring), and provincial/geographical contributions (grid-level). Model performance was evaluated against observed BC and isotope-derived fbb.

- Seasonal BC concentrations and sources: - Observed BC at YRE varied widely, peaking in winter (0.6–7.7 µg m−3); winter mean BC 3.5 ± 2.0 µg m−3 (n=22). WSOC winter mean 7.9 ± 3.9 µg m−3 (n=22). - Monthly means mirrored EI seasonal trends: higher in winter–spring, lower in summer, rising again in autumn–winter. - Strong correlations between carbonaceous components and PM2.5 (R2 = 0.61–0.86), indicating carbonaceous aerosols dominate PM2.5 variability. - Isotope-constrained sources: - Biomass burning fraction in BC (fbb-BC) showed strong seasonality: winter 45 ± 7% (n=5), spring 36 ± 3% (n=2), summer 31 ± 5% (n=3), autumn 37 ± 7% (n=3). - Urban Shanghai winter fbb-BC ~40% (n=3), slightly lower than YRE due to local fossil vehicle influence. - Winter fbb-BC (~45%) is 15–30% higher than in summer and markedly higher than January 2013/2014 reports for Shanghai/Beijing/Tianjin/Zhejiang (~20–33%). - MC deconvolution showed: coal contribution relatively constant (∼17–33% across samples); liquid fossil fraction lowest in winter (~30%) and highest in spring–summer (~45%); biomass increases dominate winter BC elevation. Highest coal contributions (31–33%) aligned with enriched δ13C-BC (≈ −25‰) and North China footprints; lowest coal (17–20%) with most depleted δ13C-BC (≤ −26.5‰). - Concentration apportionment indicated biomass burning drove winter BC increase (BCbb = 2.5 ± 0.5 µg m−3 vs 0.6 ± 0.2 µg m−3 in cleaner seasons), while liquid fossil and coal also contributed. - WSOC at YRE in winter haze had fbio ≈ 60% with WSOC 10.6 ± 4.0 µg m−3 (n=7). Urban TC in winter haze had fbio ≈ 46% with TC 21.2 ± 2.7 µg m−3 (n=3). Potassium correlated strongly with 14C-based biomass content, supporting biomass burning dominance. - Model–observation agreement: - FEG model reproduced BC concentrations and seasonality (R = 0.72, P < 0.05); some winter underestimation likely due to EI uncertainties (residential biofuel underestimation) and wet deposition during sampling. - Modelled fbb-BC agreed with observations (R = 0.77, P < 0.05) but was biased low by 3–15%, consistent with underestimation/misallocation of residential biofuel emissions in EIs. - Seasonal co-variation of fbb-BC with BC concentrations (R = 0.79 monthly; 0.99 seasonal). - Open wildfire BC in Shanghai winter was minor (~1%); residential biomass contributed >30% of BC. - Geographic and sectoral origins: - Trans-provincial sources dominated winter pollution (80–96% of BC), with Jiangsu, Anhui, Henan, and Shandong contributing ~65% combined; notable contributions also from North China (Shandong, Shanxi, Hebei, Nei Mongol, Gansu) for coal-influenced events (e.g., 14 Jan 2019). - Summer cleaner periods were more influenced by local/southeastern provinces (Zhejiang, Fujian, Guangdong) with lower biomass burning. - Sectoral contributions showed strong seasonality: residential ~50% in winter, dropping to ~17% in summer; transport increasing from 22% (winter) to ~41% (summer) with similar absolute BC across seasons; industry relatively constant; wildfires highest in summer (~2%) but ~1% otherwise. - Policy-relevant insights: - Residential biomass burning is increasingly important in winter haze under current controls; EI likely underestimates residential biofuel shares in key provinces (e.g., Hebei ~23%, Henan 28%, Nei Mongol 8%, Shanxi 10%).

The findings show a substantial post-control shift toward biomass burning as a key driver of winter BC pollution in Shanghai, despite reductions in fossil-fuel emissions from power, industry, and transport. The strong wintertime fbb-BC (~45%) and dominant trans-provincial contributions from China’s central–east corridor indicate that Shanghai acts as a receptor of transported residential emissions, with biomass burning elevating BC and WSOC during haze episodes. The consistency between isotope-constrained apportionment and FEG simulations (with modest bias) strengthens confidence in the identified sources and geographic origins, while the bias suggests EI underestimation/misallocation of residential biofuel use. Sectoral analyses emphasize that wintertime residential emissions dominate, whereas transport is relatively more important in summer. These results directly address the research question by quantifying seasonal source fractions and provincial/sectoral origins and by demonstrating the critical role of regional transport, thereby informing targeted mitigation strategies and regional coordination for haze control.

A year-long dual-isotope observational study combined with FLEXPART–ECLIPSE–GFED modelling reveals that biomass burning, especially from residential sources, has increased its contribution to winter haze in Shanghai to ~45%, higher than in 2013/2014 and substantially above summer levels. Trans-provincial transport from Jiangsu, Anhui, Henan, and Shandong accounts for the majority of winter BC, while sectoral analysis identifies residential emissions as the largest winter contributor, followed by transport and industry; wildfires are negligible. Clearing haze and mitigating climate impacts require substantial reductions in regional residential solid-fuel use, alongside continued controls on urban transport and industry, and inter-provincial coordination. The integrated isotope–model framework provides a robust, scalable approach for diagnosing sources and can be applied to other East and South Asian megacities to guide effective policy. Future research should improve residential biofuel emission characterization in inventories, expand isotope sampling across more sites and years, and evaluate mitigation scenarios with updated EIs and coupled models.

- Emission inventory uncertainties, especially underestimation or misallocation of residential biofuel emissions, lead to model underprediction of fbb-BC and some BC concentrations. - Wet deposition during certain sampling periods likely caused lower observed concentrations than simulated (model did not capture all precipitation scavenging events perfectly), particularly in spring–summer monsoon conditions. - Limited number of isotope-resolved samples per season (especially spring) constrains fine-scale temporal resolution of source apportionment. - Δ14Cbio endmember variability (biomass type/age) introduces uncertainty, though assessed via Monte Carlo; coal/liquid fossil δ13C endmembers also carry inherent variability. - Study focuses on one suburban receptor and one urban site; broader spatial coverage would strengthen regional representativeness.