Mortality vs Exposure — does the ranking flip?

Two rankings, same sources

Ranked by exposure avoided, and separately by deaths avoided — scored with a single statewide baseline death rate, so the ranking cannot reflect local differences in health.

PHYSICAL — land-based industry, fixed smokestacks. ALL ⚠ — adds ports and airports as a sensitivity.

A line that stays flat = it keeps its place; coral = it changed rank.

How to read this chart

Each source is one industry's emissions of one kind of particle, cut by 30%. Exposure avoided is population × µg/m³.
How to read the deaths column: it is a counterfactual estimate — if that cut were made and the cleaner air sustained, we estimate that many fewer steady-state annual deaths in the Bay Area.

Particle types: sulfate (forms in air from SO₂) · nitrate (from NOₓ) · ammonium (from ammonia) · organic (forms in air from vapours) · soot & dust (emitted directly as particles).

Why the order holds — the Bay sits on a narrow slice of the risk curve

Converting exposure into deaths means weighting each place by how steep the health-risk curve is where that place already sits. The curve rises fast at low pollution and flattens at high pollution, so the weight differs from place to place. But the whole Bay Area sits inside the narrow coral band below — across it the weight varies by . That is far too little to reshuffle a list already set by population × concentration, which is why the two rankings come out the same.

Health-risk curve (share of deaths attributable) Where the Bay Area sits (3.5–5.2 µg/m³) 2.4 µg/m³ — no added risk modeled below this

Ranked sources

Sort by · ⇄ = changed rank.
Two rows can print the same value and still be ranked: the order uses the unrounded numbers. Hover a value for full precision. The table scrolls sideways on a narrow screen.

#	Source · particle	Exposure avoided person·µg/m³/yr	Deaths avoided per year	Rank exposure → deaths

ⓘ Methodology & limits

Honesty ledger — assumptions & limits

• These rows should not be added up. Each row is estimated on its own against the same baseline, so rows overlap slightly and are not meant to be added together. The top-10 comparison above adds them anyway, as a close approximation — the overlap is far too small to change which list wins.

• The cuts are assumed, not measured — a flat 30% reduction applied to every industry-and-particle pair, reducing each co-pollutant proportionally. Real abatement strategies target CO₂ and would not look like this.

• Deaths are a counterfactual estimate: what a sustained cut would avoid at steady state, each year, under the source-apportionment model — not a count of events that happened.

• Baseline mortality rate 0.00975 per person per year (IHME GBD 2023: California non-communicable disease + lower respiratory infection, ages 25+, crude), applied uniformly across Bay cells and to the all-ages population — the state is the finest resolution available, and no US county rates exist in that source. It cancels out of the ranking comparison and the top-N ratio; it only scales the absolute death count.

• The GEMM health-risk function (Burnett 2018) is the upper-bound sensitivity; the recommended MR-BRT function runs 30–70% lower and has the same shape here — same verdict, smaller numbers.

• Ports and airports pin mobile and diffuse emissions to a single cell (71 percent of the modeled Bay signal) — excluded from the default ranking, shown under ALL ⚠ as a sensitivity.

• Meteorology is frozen at 2005; emissions inventory is Climate TRACE V5.9.0; the food-and-beverage entry is flagged low-confidence upstream.

• Full ranked caveats, per-number sources and re-derivation commands live in 📄 Data, methods & provenance.

Does the abatement ranking flip?

Two rankings, same sources

Why the order holds — the Bay sits on a narrow slice of the risk curve

Ranked sources