Where the exposure comes from (row) → who breathes it (column)
person·µg/m³ = one microgram per cubic metre breathed by one person for a
year. Click any cell — or any county on the map below.
How to read this matrix
Each cell is the population-weighted PM2.5 exposure that sources in the row
county deliver to people in the column county. The white-outlined diagonal is the share
each county delivers to its own residents. Row and column totals include the diagonal; the
net-export figures in the right panel exclude it.
Counties are ordered north to south, so near-neighbour transport sits near the
diagonal.
The map below the matrix draws one line of this table on geography: pick a county and
it shades every other county by the transfer between them, on the same scale and the same
colours as the cells above. Who pollutes it draws that county's column,
Where its pollution goes draws its row, and the matching row or column is
outlined in the table while you look at it.
Figures are modelled increments that Bay Area sources add on top of a 2005 baseline, from
the InMAP source-to-receptor matrix SR-2005-annual-v02.0 (2005 annual-average weather)
driven by Climate TRACE 2025 total-annual emissions — not measured ambient air quality.
The county map could not be loaded, so this page is
showing the matrix only. Every number above is unaffected — the map draws the same figures
on geography and adds nothing to them. The colour scale below describes the table.
What these numbers are
Marginal increments that Bay Area sources add on top of a 2005 baseline, under our
source-apportionment model — a modelled contribution, not measured ambient air quality. Emissions
are Climate TRACE total annual 2025 figures (full-year source totals, not a single month).
Closed 9-county system
Only the nine Bay Area counties are represented on both the source and the receptor side.
Transport in and out of the region is truncated, so edge counties (Napa, Sonoma, Solano) sit low.
Meteorology and time
InMAP SR-2005-annual-v02.0 uses 2005 annual-average frozen meteorology, a single ground-level
emission height, and an annual steady state. A different weather year would move the numbers.
Ports, airports and road traffic
Climate TRACE pins these mobile and diffuse emissions to a single point, and the model then
injects the whole plume into a single grid cell — and the cells holding those sources measure
3.4 to 27 km across, not a uniform ~4 km. That produces a spurious single-cell hotspot, which
is why they are off by default and labelled as a sensitivity check when switched on. They account
for roughly 71% of the modelled all-sector Bay signal
(shipping alone about 52%).
Lower-confidence sectors
Food, beverage and tobacco processing has weaker underlying emissions data than the industrial
sectors. Switch it off to see how much of the picture depends on it.
The health-scale panel (right, secondary)
Deaths avoided are estimated with the GEMM concentration–response function, which is an
upper-bound relationship, and with a baseline mortality rate measured in the 25-and-over
population applied to the whole population — both push the estimate high. Dollar figures price a
small reduction in mortality risk spread across a large population, at an illustrative value of a
statistical life.
Where the numbers come from
Full source list, per-number derivations and re-derivation commands are in the
Data, methods & provenance drawer (the button at the bottom right of the screen).