Shared research project · Literature-to-design guide

When Canadian smoke crosses the border

Use economics to trace each step between a fire burning in Quebec or Manitoba and a death in a U.S. county—then match each step to evidence that public data can actually provide.

Frame the problemSee the designStart a project
Core distinctionSmoky is not Canadian

A smoky day in a U.S. county can come from U.S., Mexican, or Canadian fires. Exposure and attribution are different measurements.

Useful unitCounty × month

Public mortality data are county-month; fire and wind data are daily and gridded, so every source is aggregated to one panel.

Claim ceilingIdentified slope, modeled total

The per-µg/m³ effect comes from the research design; the total death count also relies on an attribution model.

The research problem

The project’s umbrella question is:

How many deaths in the United States does smoke from Canadian wildfires cause, and what are those deaths worth?

The answer is not one jump from a fire to a death. It is a chain of physical and economic links, and each link is measured by a different dataset.

01Fireradiative power
02Transportupper-level winds
03Surface smokePM2.5 at ground level
04Exposurewhere people live
05Deathscounty-month counts
06ValueVSL in dollars

The 2023 Canadian fire season sent heavy smoke into the eastern United States, including the June 7–8 episode over New York. Published estimates of the U.S. death toll exist, but they are health-impact assessments: they multiply modeled smoke exposure by a dose–response function estimated in other populations (Zhang et al. 2025; Pye et al. 2026). No study yet estimates U.S. mortality from Canadian smoke using variation that is specific to Canadian fires. That gap is the project’s opening.

Which number?

Four different quantities are often called “the death toll from smoke.” Name the one you mean before comparing studies.

Health-impact estimateBorrowed slope × modeled exposure

Zhang et al. report about 4,100 acute U.S. deaths from 2023 Canadian smoke (Zhang et al. 2025).

Causal slopeDeaths per µg/m³ of smoke

Identified by a research design, not assumed from another setting.

Valued totalAttributed deaths × value of a statistical life

Converts deaths into an externality measured in dollars.

Comparison What differs What it means Can the project match it?
Acute vs. chronic Short-run deaths vs. deaths from long-term exposure Zhang et al.’s 4,100 is acute; their ~33,000 is chronic, a different estimand Acute, yes; a county-month design identifies short-run mortality
Same-day vs. monthly Lag 0 vs. deaths within a month (plus later months) A monthly slope includes within-month lags Approximately, with the lag difference stated
Thresholded vs. all smoke Only “Canada smoke days” vs. all Canadian smoke A narrower day filter mechanically lowers the count Yes, by rebuilding their day filter on our grid
National vs. sample Whole U.S. vs. counties large enough for stable counts Excluded small counties must be added back Yes, through state-remainder units
Important

Interpretation rule: a number above or below 4,100 is not evidence that the health-impact assessment is wrong until lag window, exposure definition, and geography are matched. The project builds that comparison one change at a time.

The identification problem

A naive regression of deaths on smoke is biased because smoke does not arrive at random.

Deathsct depend on smokect but smoke co-moves with heat, drought, stagnant air, and the weather patterns that also grow fires

Heat alone kills, so a smoke–death correlation can overstate the effect of smoke. Measurement error in smoke products and people avoiding smoke push the other way. The design must supply variation in smoke that is unrelated to local conditions.

What the literature teaches

1. Wildfire smoke kills—and causal estimates come from domestic smoke

Causal studies of U.S. smoke find that smoke raises mortality and emergency visits, often nonlinearly (Miller et al. 2024; Heft-Neal et al. 2023). National panels link smoke PM2.5 to mortality and project large future burdens (Qiu et al. 2025). Smoke also lowers earnings and test scores and changes behavior (Borgschulte et al. 2024; Wen and Burke 2022; Burke et al. 2022). These studies mostly rely on smoke from U.S. fires. None isolates smoke from Canada.

2. Borrowed dose–response functions, and why ours might differ

The 2023 estimates apply a pooled multi-country wildfire relative risk for acute deaths (Chen et al. 2021) and a meta-analytic relative risk for chronic deaths (Chen and Hoek 2020). If the project estimates a different slope for the eastern United States, at least four explanations compete:

Measurement

The smoke product is off.

Satellite and kriged products can miss diffuse, aged smoke, especially in the East where plumes often sit aloft.

Look for: disagreement between smoke products and ground monitors that varies by region.
Chemistry

Long-traveled smoke is different.

Smoke that crosses 1,000+ km ages chemically, so its toxicity per µg/m³ may differ from fresh smoke.

Look for: slopes that change with transport distance, holding the population fixed.
Adaptation

People used to smoke protect themselves.

Air purifiers, alerts, and habits may cut harm where smoke is common; eastern residents have had less practice.

Look for: smaller per-unit effects in counties with more past smoke, within the same region.
Population

The exposed people differ.

Age, baseline health, air conditioning, and housing change how much outdoor smoke reaches lungs.

Look for: slopes that vary with age structure and air-conditioning prevalence (Janssen et al. 2002).

3. Transboundary pollution has economic precedents

China → Korea

Jia & Ku

Chinese pollution carried by dust storms raised mortality in South Korea (Jia and Ku 2019).

Project use: a design that combines transport events with source-country pollution.
China → Korea

Heo, Ito & Kotamarthi

Trajectory-based exposure to Chinese pollution and Korean mortality, valued as an international externality (Heo et al. 2025).

Project use: the template for turning a spillover estimate into a dollar cost.
Indonesia → Singapore

Sheldon & Sankaran

Fire radiative power from Indonesian fires instruments for Singapore’s pollution and health outcomes (Sheldon and Sankaran 2017).

Project use: the closest precedent for a foreign-fire instrument.
Indonesia 1997

Jayachandran

Smoke from Indonesian fires increased early-life mortality (Jayachandran 2009).

Project use: evidence that distant fire smoke has measurable mortality effects.

4. Wind as a natural experiment

Changes in wind direction move pollution across places for reasons unrelated to local health, which gives a credible instrument for pollution (Deryugina et al. 2019; Anderson 2020; Schlenker and Walker 2016). Fires upwind vs. downwind of a place provide similar variation (Rangel and Vogl 2019). This project combines both ideas: Canadian fire activity located upwind of a county, carried by the day’s winds.

Because such an index is built from a formula, some counties are always more exposed simply by geography. Recentering subtracts the exposure a county would expect from other fire seasons under the same winds, so that only the unusual part of each season’s fires remains (Borusyak and Hull 2023).

FiresCanadian radiative power
Winds850–700 hPa transport
Kerneldistance + direction
Exposurecounty-month index
Recentersubtract expected
Instrumentunusual fire exposure

5. Adaptation, defensive behavior, and value

People adapt to environmental hazards. Heat–mortality relationships weakened as air conditioning spread and are smaller where heat is common (Barreca et al. 2016; Heutel et al. 2021). Pollution alerts change behavior (Neidell 2009), and households pay to defend themselves with air purifiers (Ito and Zhang 2020). These results make adaptation a plausible reason why eastern U.S. counties, which rarely faced heavy smoke before 2023, might suffer more per unit.

Deaths

identified slope × smoke attributed to Canada × population, summed over counties and months

Dollars

deaths × value of a statistical life, with a life-years version because smoke deaths skew old (Viscusi and Aldy 2003; Aldy and Viscusi 2008)

Evidence map: design, result, and transfer

The table treats each paper as design → estimand → limitation → project use.

Study Context and design What it credibly contributes Transfer to this project
Zhang et al. (Zhang et al. 2025) 2023 Canadian fires; transport modeling + borrowed dose–response North American acute and chronic burden; 4,100 acute U.S. deaths The comparator; match its lag, day filter, and geography before comparing
Pye et al. (Pye et al. 2026) EPA air-quality modeling of 2023 smoke, PM2.5, ozone, air toxics Official assessment including ozone and toxics Reminds us smoke is a mixture; the project’s slope is per µg/m³ of the Canadian smoke mixture
Miller, Molitor & Zou (Miller et al. 2024) U.S. smoke shocks; instrumental variables; elderly mortality Causal, concave dose–response for smoke A benchmark slope; concavity matters for monthly averages
Qiu et al. (Qiu et al. 2025) U.S. county panel of smoke PM2.5 and mortality National smoke–mortality function and future burden Nearest national estimate; not identified from Canadian variation
Deryugina et al. (Deryugina et al. 2019) Wind-direction instrument; Medicare mortality Credible PM2.5–mortality effect from wind shifts Design template for wind-driven variation
Sheldon & Sankaran (Sheldon and Sankaran 2017) Indonesian fire radiative power → Singapore Foreign fire intensity as an instrument Precedent for using fire radiative power across a border
Jia & Ku (Jia and Ku 2019) Dust-transport days × Chinese pollution → Korea Transboundary mortality effect Shows how transport timing isolates a foreign source
Borusyak & Hull (Borusyak and Hull 2023) Econometric theory for formula instruments Recentering removes bias from non-random exposure Why the project subtracts expected exposure from other seasons
Heutel, Miller & Molitor (Heutel et al. 2021) U.S. temperature–mortality by climate region Harm per hot day is smaller where heat is common Motivates testing whether smoke harm falls with past smoke
Childs et al. (Childs et al. 2022) Machine-learning daily smoke PM2.5 for U.S. counties Standard smoke measure for 2006–2023 Robustness smoke product; the main product covers through 2025

What is already known about 2023

Several studies examine the 2023 episode directly. Health-impact assessments estimate the North American and U.S. burden (Zhang et al. 2025; Pye et al. 2026). In New York City, asthma-related emergency visits rose during the June smoke (Chen et al. 2023). Among about 53,000 hemodialysis patients in 22 states, days under the 2023 smoke plume were associated with higher same-day mortality and hospitalization (Song et al. 2025). Satellite smoke climatologies show that smoke is a regular summer presence over U.S. cities: it is present on about one in five days that exceed the 24-hour PM2.5 standard, and on 10–20% of ozone exceedances in places such as the Northeast corridor (Kaulfus et al. 2017; Brey and Fischer 2016). The record 2023 Canadian season followed early snowmelt, drought, and unusual heat (Jain et al. 2024).

A new study must add something these do not:

Identify

Use Canadian-specific variation

Estimate the slope from Canadian fires upwind of each county, not from a borrowed function.

Compare

Match the scope

Rebuild the published estimate step by step so a difference is interpretable.

Extend

Use twenty seasons

Cover 2006–2025, not only 2023, so identification does not rest on one episode.

Value

Price the externality

Convert attributed deaths into dollars that no market or treaty currently prices.

What each dataset can—and cannot—measure

Source Role Useful measures Appropriate join Cannot establish
NASA FIRMS Fire activity (instrument) fire location, date, radiative power, satellite, collection 2° source cell × local day Smoke amount, smoke height, or where the smoke went
ERA5 daily statistics Transport winds; weather regimes u/v winds at 850 and 700 hPa; 500 hPa height; surface weather grid → county population weights Whether smoke actually arrived at the surface
CSU kriged smoke PM2.5 (Dryad) Main smoke measure (2006–2025) total PM2.5, background, HMS flag; smoke = flag × (total − background) 15 km grid → county-day The source country of the smoke
Childs et al. smoke PM2.5 Robustness smoke measure (2006–2023) daily smoke PM2.5 by county county-day 2024–2025; source attribution
NOAA HMS smoke polygons Smoke extent and episode flags analyst-drawn plume outlines and density polygon → county-day Surface concentration (plumes are often aloft in the East)
EPA EMBER Canadian share of smoke (2023) base, zero-fire, and zero-Canadian-fire model runs 36 km grid → county-day An independent check (it shares fire inputs with the instrument)
EPA AQS Ground truth daily monitor PM2.5 and ozone monitor → county-day Concentrations where there is no monitor
CDC WONDER Deaths (county × year, county × month) resident deaths by cause and age county-month Counts of 1–9 (suppressed); individual records
NCHS public-use deaths (NBER mirror) Pre-sample monthly deaths, 1999–2004 county × month deaths, counties ≥100,000 people fixed county geography Small counties or any year after 2004
SEER populations Denominators and weights county population by age county-year → month Deaths or exposure

Current project data: constraints to respect

  1. HMS smoke is often aloft. A plume overhead is not smoke at breathing height. Use HMS for episodes and extent, not for exposure.
  2. The MODIS record changes processing version in 2023. Collection 6.0 covers 2003–2022 and 6.1 covers 2023–2025. Canadian fire totals show no break, but always carry the collection flag.
  3. WONDER suppresses counts of 1–9. The project keeps counties with at least 20 deaths per month in 1999–2005 (1,640 counties, about 94% of the 2005 population) and treats the rest as state remainders.
  4. The design is outcome-blind. Mortality for 2006–2026 is not extracted until the analysis plan is registered. Student projects use exposure and pre-sample data only.
Tip

Aggregate once, carefully. Fires are daily and gridded, winds are daily and gridded, deaths are county-month. Document every conversion—local solar day vs. UTC day, land vs. water area, population vs. area weights—because each one can shift the answer.

A research design for this project

Hypotheses and discriminating evidence

Hypothesis Observable implication Rival explanation Evidence that would move the claim forward
H1 · Level Unusual upwind Canadian fire raises smoke and then deaths Fire seasons coincide with heat waves Balance tests on weather that smoke cannot change; heat controls; placebo downwind fires
H2 · Scope The matched estimate differs from the borrowed dose–response Lag, day filter, and geography differ A bridge that changes one scope element at a time
H3 · Adaptation Per-unit harm is larger where past smoke was rare Age, air conditioning, income, smoke chemistry Within-region variation in past exposure, holding those rivals fixed
H4 · Persistence Effects appear in later months too Deaths moved forward (harvesting) Cumulative effects over 0–3 months and beyond
H5 · Mechanism Alerts and searches rise with Canadian smoke Attention driven by news, not exposure NWS alerts and search data tied to measured smoke

Claim ladder

Level 1

Describe

When and where did Canadian smoke reach U.S. counties?

FIRMS + HMS + smoke PM2.5 + population
Level 2

Attribute

How much of each county’s smoke came from Canada?

Add EMBER model runs and the transport index
Level 3

Identify

How many deaths did Canadian smoke cause per µg/m³?

Add the recentered instrument and outcome-blind gates
Level 4

Value

What is that externality worth, and to whom?

Add the value of a statistical life and age profiles

Move upward only when the data support the next claim. Levels 1 and 2 are open to student projects now.

Student projects

Student projects stay at Levels 1–2 and never use 2006–2026 mortality.

1

Reproduce one input

Rebuild a county-day series (smoke, fires, or weather) from the raw files and check it against the project panel.

2

Pick one episode

Choose a smoke episode (for example June 2023 or July 2021) and document it: dates, source fires, plume maps, surface PM2.5.

3

Compare two measures

Put HMS plumes, CSU smoke, Childs smoke, and AQS monitors side by side for that episode.

4

Map the exposure

Map population-weighted smoke by county; test how the map changes with area vs. population weights.

5

Write the evidence memo

State the claim, the unit, the comparison, the rivals, and the claim ceiling.

6

Stress-test it

Change one construction choice (threshold, day definition, weights) and report what moves.

Contribution paths

Episodes

Build an episode atlas

Document each Canadian-smoke episode reaching the East, 2006–2025, with dates, source provinces, and affected population.

Measurement

Compare smoke products

Quantify where HMS, CSU, Childs, and monitors agree or disagree, and why the East is harder.

Attribution

Explore EMBER

Compare EMBER’s Canadian share with a simple wind-direction share for June 2023.

Behavior

Track alerts and attention

Link NWS air-quality alerts and search interest in air purifiers to measured smoke.

Literature

Own one mechanism

Trace adaptation, smoke chemistry, or air conditioning from theory to a testable prediction.

Policy

Build the policy chronology

Record exceptional-event filings, alert rules, and U.S.–Canada disputes over smoke, with dates.

Evidence memo template

For every result, record:

  1. Claim: the one sentence the figure or table supports;
  2. Unit and sample: what one row represents and which places and dates are included;
  3. Measure: which smoke, fire, or weather product, and how it was aggregated;
  4. Comparison: what variation the result rests on;
  5. Mechanism: which physical or economic process predicts it;
  6. Rivals: at least two alternative explanations;
  7. Sensitivity: which construction choice changes it; and
  8. Claim ceiling: description, attribution, causal effect, or value.
CautionData governance
  • Never request, open, or share 2006–2026 mortality files; the project is outcome-blind until its analysis plan is registered.
  • Follow CDC WONDER’s data-use terms: never publish or try to recover counts of 1–9.
  • Never place API keys or account credentials (Copernicus, Earthdata) in the course repository.
  • Keep large raw files on the project drive; share derived county-level tables with a provenance note.

Decision rule for the project

Research principle The design identifies the slope; every other step to the headline number must be shown.

Report the identified per-µg/m³ effect first, then each attribution and scope choice that turns it into deaths and dollars—so a reader can see exactly what the number depends on.

References

Aldy, Joseph E., and W. Kip Viscusi. 2008. “Adjusting the Value of a Statistical Life for Age and Cohort Effects.” Review of Economics and Statistics 90 (3): 573–81. https://doi.org/10.1162/rest.90.3.573.
Anderson, Michael L. 2020. “As the Wind Blows: The Effects of Long-Term Exposure to Air Pollution on Mortality.” Journal of the European Economic Association 18 (4): 1886–927. https://doi.org/10.1093/jeea/jvz051.
Barreca, Alan, Karen Clay, Olivier Deschênes, Michael Greenstone, and Joseph S. Shapiro. 2016. “Adapting to Climate Change: The Remarkable Decline in the US Temperature-Mortality Relationship over the Twentieth Century.” Journal of Political Economy 124 (1): 105–59. https://doi.org/10.1086/684582.
Borgschulte, Mark, David Molitor, and Eric Yongchen Zou. 2024. “Air Pollution and the Labor Market: Evidence from Wildfire Smoke.” Review of Economics and Statistics 106 (6): 1558–75. https://doi.org/10.1162/rest_a_01243.
Borusyak, Kirill, and Peter Hull. 2023. “Nonrandom Exposure to Exogenous Shocks.” Econometrica 91 (6): 2155–85. https://doi.org/10.3982/ECTA19367.
Brey, Steven J., and Emily V. Fischer. 2016. “Smoke in the City: How Often and Where Does Smoke Impact Summertime Ozone in the United States?” Environmental Science & Technology 50 (3): 1288–94. https://doi.org/10.1021/acs.est.5b05218.
Burke, Marshall, Sam Heft-Neal, Jessica Li, et al. 2022. “Exposures and Behavioural Responses to Wildfire Smoke.” Nature Human Behaviour 6 (10): 1351–61. https://doi.org/10.1038/s41562-022-01396-6.
Chen, G., Y. Guo, X. Yue, et al. 2021. “Mortality Risk Attributable to Wildfire-Related PM2.5 Pollution: A Global Time Series Study in 749 Locations.” The Lancet Planetary Health 5 (9): e579–87. https://doi.org/10.1016/S2542-5196(21)00200-X.
Chen, Jie, and Gerard Hoek. 2020. “Long-Term Exposure to PM and All-Cause and Cause-Specific Mortality: A Systematic Review and Meta-Analysis.” Environment International 143: 105974. https://doi.org/10.1016/j.envint.2020.105974.
Chen, Kai, Yiqun Ma, Michelle L. Bell, and Wei Yang. 2023. “Canadian Wildfire Smoke and Asthma Syndrome Emergency Department Visits in New York City.” JAMA 330 (14): 1385–87. https://doi.org/10.1001/jama.2023.18768.
Childs, Marissa L., Jessica Li, Jeff Wen, et al. 2022. “Daily Local-Level Estimates of Ambient Wildfire Smoke PM2.5 for the Contiguous US.” Environmental Science & Technology 56 (19): 13607–21. https://doi.org/10.1021/acs.est.2c02934.
Deryugina, Tatyana, Garth Heutel, Nolan H. Miller, David Molitor, and Julian Reif. 2019. “The Mortality and Medical Costs of Air Pollution: Evidence from Changes in Wind Direction.” American Economic Review 109 (12): 4178–219. https://doi.org/10.1257/aer.20180279.
Heft-Neal, Sam, Carlos F. Gould, Marissa L. Childs, et al. 2023. “Emergency Department Visits Respond Nonlinearly to Wildfire Smoke.” Proceedings of the National Academy of Sciences 120 (39): e2302409120. https://doi.org/10.1073/pnas.2302409120.
Heo, Seonmin Will, Koichiro Ito, and Rao Kotamarthi. 2025. “International Spillover Effects of Air Pollution: Evidence from Mortality and Health Data.” Review of Economics and Statistics, ahead of print. https://doi.org/10.1162/rest_a_01581.
Heutel, Garth, Nolan H. Miller, and David Molitor. 2021. “Adaptation and the Mortality Effects of Temperature Across U.S. Climate Regions.” Review of Economics and Statistics 103 (4): 740–53. https://doi.org/10.1162/rest_a_00936.
Ito, Koichiro, and Shuang Zhang. 2020. “Willingness to Pay for Clean Air: Evidence from Air Purifier Markets in China.” Journal of Political Economy 128 (5): 1627–72. https://doi.org/10.1086/705554.
Jain, Piyush, Quinn E. Barber, Stephen W. Taylor, et al. 2024. “Drivers and Impacts of the Record-Breaking 2023 Wildfire Season in Canada.” Nature Communications 15: 6764. https://doi.org/10.1038/s41467-024-51154-7.
Janssen, Nicole A. H., Joel Schwartz, Antonella Zanobetti, and Helen H. Suh. 2002. “Air Conditioning and Source-Specific Particles as Modifiers of the Effect of PM10 on Hospital Admissions for Heart and Lung Disease.” Environmental Health Perspectives 110 (1): 43–49. https://doi.org/10.1289/ehp.0211043.
Jayachandran, Seema. 2009. “Air Quality and Early-Life Mortality: Evidence from Indonesia’s Wildfires.” Journal of Human Resources 44 (4): 916–54. https://doi.org/10.1353/jhr.2009.0001.
Jia, Ruixue, and Hyejin Ku. 2019. “Is China’s Pollution the Culprit for the Choking of South Korea? Evidence from the Asian Dust.” Economic Journal 129 (624): 3154–88. https://doi.org/10.1093/ej/uez021.
Kaulfus, Aaron S., U. S. Nair, Daniel A. Jaffe, Sundar A. Christopher, and Scott L. Goodrick. 2017. “Biomass Burning Smoke Climatology of the United States: Implications for Particulate Matter Air Quality.” Environmental Science & Technology 51 (20): 11731–41. https://doi.org/10.1021/acs.est.7b03292.
Miller, Nolan H., David Molitor, and Eric Zou. 2024. The Nonlinear Effects of Air Pollution on Health: Evidence from Wildfire Smoke. Working Paper No. 32924. National Bureau of Economic Research. https://doi.org/10.3386/w32924.
Neidell, Matthew. 2009. “Information, Avoidance Behavior, and Health: The Effect of Ozone on Asthma Hospitalizations.” Journal of Human Resources 44 (2): 450–78. https://doi.org/10.1353/jhr.2009.0018.
Pye, Havala O. T., William T. Hutzell, Neal L. Fann, et al. 2026. “The Risks to Human Health of Air Toxics, PM2.5, and Ozone from the 2023 Canadian Wildfires.” Environmental Science & Technology Letters 13 (2): 268–74. https://doi.org/10.1021/acs.estlett.5c01181.
Qiu, Minghao, Jessica Li, Carlos F. Gould, et al. 2025. “Wildfire Smoke Exposure and Mortality Burden in the USA Under Climate Change.” Nature 647 (8091): 935–43. https://doi.org/10.1038/s41586-025-09611-w.
Rangel, Marcos A., and Tom S. Vogl. 2019. “Agricultural Fires and Health at Birth.” Review of Economics and Statistics 101 (4): 616–30. https://doi.org/10.1162/rest_a_00806.
Schlenker, Wolfram, and W. Reed Walker. 2016. “Airports, Air Pollution, and Contemporaneous Health.” Review of Economic Studies 83 (2): 768–809. https://doi.org/10.1093/restud/rdv043.
Sheldon, Tamara L., and Chandini Sankaran. 2017. “The Impact of Indonesian Forest Fires on Singaporean Pollution and Health.” American Economic Review 107 (5): 526–29. https://doi.org/10.1257/aer.p20171134.
Song, Hanyi, M. Liang, N. E. Sieck, et al. 2025. “The 2023 Canadian Wildfires and Risk of Hospitalization and Mortality Among Hemodialysis Patients in the United States.” Kidney International Reports 10 (6): 1750–60. https://doi.org/10.1016/j.ekir.2025.04.002.
Viscusi, W. Kip, and Joseph E. Aldy. 2003. “The Value of a Statistical Life: A Critical Review of Market Estimates Throughout the World.” Journal of Risk and Uncertainty 27 (1): 5–76. https://doi.org/10.1023/A:1025598106257.
Wen, Jeff, and Marshall Burke. 2022. “Lower Test Scores from Wildfire Smoke Exposure.” Nature Sustainability 5 (11): 947–55. https://doi.org/10.1038/s41893-022-00956-y.
Zhang, Qiang, Yuexuanzi Wang, Qingyang Xiao, et al. 2025. “Long-Range PM2.5 Pollution and Health Impacts from the 2023 Canadian Wildfires.” Nature 645 (8081): 672–78. https://doi.org/10.1038/s41586-025-09482-1.
Back to top