State-to-state PM2.5 transport — receptor & source
PROTOTYPE
Scope: real 2025 Climate TRACE emissions run through the real InMAP v02
source–receptor matrix. Loading…
Click any state — see whose emissions reach its air, as
attributable deaths, exposure, or illustrative VSL.
Every other state is shaded by its modeled contribution — cyan marks
the state you picked, never a data value.
Metric
Mode
Scope
Scale
National attributable deaths/yr
—
steady-state, attribution share — current scope
A floor — not a national inventory.
Out-of-state share
—
—
Each metric has its own basis and they differ — deaths 38.5%, exposure
32.0%, ΔPM2.5 48.5% for the current scope. The figure above follows the metric selected.
National exposure
—
person·µg/m³, population-weighted — current scope
Illustrative VSL
—
$11.0M/life (2023 USD) — not an official damage estimate
Hub-attribution impact
—
deaths added by including hub-pinned shipping & aviation
Stays in-state (national)
—
of all modeled point-source PM2.5, summed over every state pair
Crosses a state line
—
a lower bound — see the #1 caveat
Population covered
—
area-weighted receptor population, all 51
Emissions modelled
—
Climate TRACE's sector universe — excludes wildfire, fugitive dust, unpaved roads and construction. A floor, not a national inventory.
t/yr, 5 sectors — see the #2 caveat for what's excluded
51 states, one coral rampRows = receptor, columns = source in the underlying matrix. In RECEPTOR mode every
other state is coloured by dPM[selected][state] — its contribution to the selected
state's air. In SOURCE mode every other state is coloured by dPM[state][selected] —
how much of the selected state's pollution lands there. With the Scale toggle set to PER UNIT
EMITTED (RECEPTOR mode only), the colour instead comes from SRagg[selected][state] =
dPM[selected][state] ÷ E_S[state] — each source's contribution divided by its own
emission rate, which can reorder the map.Scroll or pinch on the map to zoom, drag to pan
Alaska
Hawaii
Puerto Rico
Alaska and Hawaii are real, clickable, fully-selectable states — drawn as
separate small insets, not to scale with the main map or each other, because fitting them
into the same projection as the lower 48 would either shrink CONUS to a sliver (Alaska's Aleutians
wrap past −180°) or waste most of the frame on Pacific ocean (Hawaii). Puerto Rico is drawn too,
but it is the one place in this payload with no SR-matrix row/column — it always renders
as the no-data hatch, never a ramp colour.
Modeled ΔPM2.5, µg/m³ (population-weighted mean)
selected state — the own-state term, excluded from the rampno state selected yetno data (Puerto Rico only)method-masked (state-mean returns 0,
per-cell finds real deaths) — see caveat #1below GEMM threshold everywhere (real
exposure, zero attributable mortality) — NOT clean air
Both hatches above are a
RECEPTOR-role property: they appear when the hatched state is receiving the
deaths/VSL being shown, which is SOURCE mode (any other state's contribution
lands in it) or RECEPTOR mode with one of those 5 states itself selected — not
when you're simply viewing a normal RECEPTOR selection's other contributors. Seeing no
hatch right now does not mean the effect went away; switch to SOURCE mode to see all
five at once.
Bin edges are quantiles of the whole matrix's nonzero
pairs (25th, 50th, 75th, 90th, 97th) — not equal steps. Values span five orders of
magnitude, so equal steps would paint nearly every state one colour, and the distribution
includes each state's own large diagonal (own-state) term, which is why most cross-state pairs
sit in the lower bins and only the very largest transport pairs reach the open-ended top bin.
Why the selected state isn't on the ramp
A state's own-state term is a large share of its own total on every metric
here (on the TECHNICAL ΔPM2.5 view specifically, 64.2% of the whole matrix nationally,
reaching 99.97% for Hawaii) — so if the selected state were coloured on the same scale as
everyone else, it would render as one very dark state next to 50 pale ones, every single time,
regardless of which state was picked. It is marked instead with a white inset ring and a cyan
outline — cyan is reserved for selection, never for a data value anywhere in this
gallery. Its own numbers are printed at right, not folded into the map.
No selection
How to read this
Which species travels — and which stays home?
Share of national exposure and own-state share, per SR species — ALL 5 SPECIES scope.
Primary PM2.5 is emitted directly and settles out close to its source.
The four secondary species instead FORM in the atmosphere from gas-phase precursors (SO2, NOx,
NH3, VOC) over hours to days, so they can travel much farther before becoming fine particles —
this is the mechanism behind the difference. Primary PM2.5 is the most local of the
five species (highest own-state share); nitrate is the most-traveled. Sorted below by
own-state share, most-local first — read from this page's own payload/summary, nothing here is
hand-typed.
Species
Share of national exposure
Own-state share (how local it stays)
Precursor gas emitted (not the aerosol species itself)
Whose tonne does the most harm?
Population exposure delivered per unit emitted.
A tonne emitted where many people live downwind does far more harm
than the same tonne emitted somewhere sparse. Ranked below: each SOURCE state's total
national population exposure delivered, per unit of its own emission rate — a
cross-state-vs-state comparison the map itself cannot make, because the map is always
relative to whichever one state you've selected. Like PER UNIT EMITTED above, this is
magnitude-independent, but NOT inventory-independent — it removes the effect
of how much a state emits, not where within that state its facilities sit; the within-state
layout (hub-pinning included) still comes from Climate TRACE.
ⓘ Health metrics (exposure / deaths / VSL) — methodology & the caveats that matter mostⓘ Methodology & limits — TECHNICAL view (ΔPM2.5 / SRagg, fixed 5-sector baseline): what's included, what's excluded, and how to read a state's own-state share