Why Removing Pollution Increases Ozone
Ground-level ozone is not emitted — it is formed by the reaction of nitrogen oxides (NOx) and volatile organic compounds (VOCs) in sunlight. In VOC-limited areas like the LA Basin and Bay Area cores, NOx actually scavenges ozone through titration. Remove the NOx and you remove the scavenger. Ozone goes up. In NOx-limited areas like the Central Valley and rural basins, the opposite happens — removing NOx reduces ozone.
Heavy-duty vehicle electrification (T5) removes a large amount of NOx quickly from highway corridors that cross both regimes. The statewide story is positive: PM2.5 benefit (78 deaths avoided in 2025) plus a net O3 benefit from NOx-limited Central Valley (+57 deaths avoided) outweighs the O3 disbenefit in VOC-limited LA cells (−39 deaths caused). The local story is different.
This is not a modeling artifact. The NOx–VOC–O3 chemistry is well-established (Sillman 1999, NRC 1991). VOC-limited ozone formation is the reason that CARB distinguishes between NOx-limited and VOC-limited basins. The phenomenon is real and has been observed in Los Angeles, Houston, and European cities during lockdown NOx reductions in 2020.
Statewide Benefit, Concentrated Local Harm
T5 is net positive statewide in every year — the PM2.5 benefit plus the Central Valley O3 benefit outweighs the LA Basin O3 disbenefit. But the harm does not disappear; it concentrates. In 2025, 6.4M Californians live in cells where the O3 disbenefit exceeds the PM2.5 benefit. That shrinks to 2.0M by 2040 as the fleet turns over and VOC controls catch up. All other policies (T1–T4) have smaller and more distributed NOx cuts, so their harmed population stays at the statewide baseline of about 1.0M.
| Year | Net Deaths (Statewide) | PM2.5 Benefit | O3 (LA cost − CV benefit) | Harmed Pop. | LA Basin Harmed |
|---|---|---|---|---|---|
| 2025 | +95.4 | +77.6 | −39.3 / +57.1 (net +17.8) | 6.4M (19.9%) | 43.4% |
| 2030 | +253.0 | +212.4 | −88.6 / +129.2 (net +40.6) | 5.0M (15.6%) | 34.1% |
| 2035 | +493.2 | +430.7 | −135.0 / +197.5 (net +62.5) | 3.1M (9.6%) | 21.0% |
| 2040 | +689.1 | +612.9 | −160.1 / +236.2 (net +76.1) | 2.0M (6.3%) | 13.7% |
T5_heavy_duty scenario. Statewide net deaths = PM2.5 deaths avoided + net O3 deaths avoided (O3 benefit in NOx-limited cells minus O3 cost in VOC-limited cells). Harmed population = residents of cells with net-negative health impact. LA Basin is the dominant VOC-limited region; SJV and Bay Area cores see smaller harm. Source: Inv 07 (Ozone Crossover), Inv 02 (Transport MC).
The statewide total hides where the harm falls. A program that avoids 95 deaths statewide can still be the worst option for the 6.4M people in LA Basin VOC-limited cells. Policy sequencing — when to cut LA Basin NOx versus Central Valley NOx — matters as much as total NOx reduction.
Same Deaths Avoided, Very Different Who
All five transport policies deliver statewide net benefit from day one. What separates them is where the harm and benefit fall. T5 concentrates both: the largest NOx cut, the biggest Central Valley O3 benefit, and the biggest LA Basin O3 disbenefit.
The lesson is not “don’t electrify heavy-duty trucks.” T5 avoids 95 deaths statewide but hurts 4.1× more people than T2, which avoids 224. The same dollar budget spread across all vehicle classes (T2) buys more benefit with far less local harm. Regional targeting matters as much as scale.
ISRM source–receptor matrix · NOx–VOC–O3 parameterization (VOC-limited / NOx-limited regime map) · Di et al. CRF for PM2.5 · Turner et al. CRF for O3 · Census tract-level population weighting · 5 transport scenarios × 4 projection years (2025–2040)