Learning Materials·October 15, 2024
This EANET training case study, led by researchers from Hankuk University of Foreign Studies, demonstrates how to assess spatial and seasonal patterns of VOCs in Seoul using a combined monitoring design: wide-area passive air samplers (PAS) deployed at 25 urban-network sites, co-interpreted with active/online PAMS measurements and compared against the official CAPSS emissions inventory. It’s a practical walk-through of methods, QA/QC, analysis, and how observations line up (or don’t) with inventory-based expectations.
Why VOCs and how they were measured.
The deck recaps common VOCs analysis methods—TD-GC/FID, TD-GC/MS, PTR-MS—and sampling via canisters, passive diffusion tubes, and active tubes with pumps. For the Seoul study, PAS (Supelco FLM Carbopack deactivated tubes) captured 56 EPA PAMS target compounds, analyzed by TD-GC/FID; in parallel, Seoul’s PAMS network delivered hourly VOCs with an online TD-GC/FID system (May 2022–May 2023). This dual setup balances broad spatial coverage (PAS) with high temporal resolution (PAMS).
Network design & instruments.
Twenty-five PAS sites span Seoul’s urban monitoring network. The passive–active comparison is framed by a clear methods table (advantages: PAS = low-cost, simple, scalable; active/online = captures short-term fluctuations, continuous). The slides show the Supelco passive tube structure, the TD-GC/FID platform, and analytical conditions used for PAS and online PAMS (dual-column configurations—AL-PLOT + BP-1 / CP Sil 5CB, specified preconcentration temperatures, temperature ramps, and FID settings). Calibration used a certified PAMS standard at 100 ppbv (Rigas, Korea) diluted to 1, 3, 5, 10 ppbv.
Data quality and processing.
A precision table for 23 representative VOCs reports RMS/MAD precisions and biases (e.g., n-pentane ~5–6% precision; o-xylene ~8% precision; some species show small but significant biases), and the workflow highlights peak alignment to stabilize retention-time drift across batches—critical when aggregating multi-site datasets. The message: PAS can deliver decision-grade data when calibration, QA/QC, and alignment are handled systematically.
What the maps show: city-wide patterns.
Seasonal TVOCs maps (June, September, February) reveal consistently high concentrations in Seoul’s southwest, attributed to a mix of heavy traffic, automobile paint shops, proximity to a metropolitan landfill, and airport influence. In summer, central districts also show elevated VOCs—likely from printing businesses and enhanced solvent evaporation at higher temperatures. Species maps unpack sources: toluene is highest in central/southwest (vehicle/solvent signals), while isoprene peaks in forested outskirts, tracking biogenic influence.
Passive vs active: broadly consistent, seasonally nuanced.
Co-located comparisons show comparable concentration levels between PAS and active/online measurements overall. However, active data emphasize September peaks (capturing short events), while passive (integrated) samples show higher June levels—illustrating how integration time and meteorology shape apparent seasonality. This is a recurring theme: choose the method that matches the decision need (episodes vs spatial baselines).
Source apportionment and the CAPSS inventory reality check.
Using PAS data in a PMF framework, the team mapped area and road source contributions. For road traffic, CAPSS spatial patterns matched PAS-derived PMF maps well. For area sources, PAS/PMF also identified high emissions in the southern/southwestern city, but southeastern hotspots present in CAPSS were not confirmed by PAS/PMF—flagging inventory misallocations or under-observed activities. Compositionally, both PAS and CAPSS agree that aromatics dominate (>50% of TVOCs) and that ARO1/ALK2/ALK3 are major groups; yet there are notable mismatches: PAS found ARO1 as the most dominant (vs ARO2 in CAPSS), benzene ~4% in winter in PAS (vs 12% year-round in CAPSS), and ALK1 much higher in PAS than inventory. These gaps matter for ozone precursor control and health risk prioritization.
Modeled vs observed concentrations.
A forward comparison of TVOCs shows good overall consistency: PAS observed ~11–13 ppbv, modeled concentrations from CAPSS data are ~12–15 ppbv. Still, the deck notes a likely summer underestimation in the inventory—consistent with the central-city summer hot spots seen in PAS maps. Such model–measurement comparisons help target inventory improvements by season, area, or source class.
Training takeaways.
Multiple methods are complementary: use PAS for large-scale, long-term mapping; pair with active/online platforms to capture short-term peaks and support mechanism studies. 2) Calibration, QA/QC, and peak alignment are non-negotiable for credible PAS results. 3) Inventory evaluation should use both spatial patterns and species composition (not just totals): agree where you can (e.g., roads) and fix where you don’t (e.g., area-source misfits, seasonal biases). 4) Policy translation: maps that separate traffic, solvent/industry, and biogenic signals (e.g., toluene vs isoprene) help cities choose the right levers—I&M and fuel-vapor controls near corridors, solvent management in central districts, and green-space management for biogenic–anthropogenic interactions.
Bottom line.
Seoul’s case proves that passive samplers can deliver high-value, city-scale intelligence on VOCs when embedded in a hybrid monitoring–inventory framework. The method is cost-effective, scalable, and—paired with robust analytics—good enough to validate inventories, reveal missed hot spots, and guide ozone-precursor strategies. Differences in seasonality and speciation between PAS observations and CAPSS estimates are not bugs; they’re action cues for inventory refinement and targeted controls.
Keywords
Passive air sampling (PAS); Seoul urban air network (25 sites); EPA PAMS target compounds (56 species); TD-GC/FID; Supelco FLM Carbopack X; active/online PAMS (hourly); calibration (PAMS 100 ppbv → 1/3/5/10 ppbv); precision & bias; peak alignment; seasonal TVOCs maps; toluene (central/southwest), isoprene (forested outskirts); PMF source apportionment; CAPSS inventory comparison; aromatics > 50% of TVOCs; ARO1 vs ARO2 mismatch; benzene share (winter 4% vs CAPSS 12%); modeled 12–15 ppbv vs observed 11–13 ppbv; summer underestimation.