Skip to main content

California Freight Cleanup → Element 4

Does cutting freight NOx actually reduce ozone?

P(net ozone benefit) = 0.0005 — 99.95% disbenefit in our scenarios

In LA, the answer is no — at least under the standard linear reduced-complexity model that fast cost-benefit work usually relies on. The LA Basin runs in a VOC-limited chemistry regime; in that regime, cutting NOx raises ozone, not lowers it. The atmospheric chemistry has been settled since Sillman (1995), but the linearized models routinely used in cost-benefit work miss it. We climbed a five-level chemistry ladder and ran nearly 25,000 uncertainty draws to pin down the sign and magnitude. Result: in 99.95% of plausible scenarios, the ozone channel is a net disbenefit. The dollar gap between the linear and regime-aware treatments in our cascade is $10.4B.

Two linked investigations close this element. The first (the ozone chemistry investigation) climbs a five-level model chain for a 20% NOx reduction across three California air basins — the LA Basin, San Joaquin Valley, and Sacramento Valley. At the simplest level, the model predicts a 1.84 ppb ozone reduction in LA (a benefit). At the regime-aware level, the same scenario yields +2.10 ppb more ozone (a disbenefit) — a 3.94 ppb reversal, validated against 31 EPA monitoring sites (+1.67 ppb median 2023→2024 trend). The LA Basin sits firmly in a “VOC-limited” chemistry regime, where cutting NOx releases radicals that form more ozone, not less. SJV and Sacramento don’t flip, but LA’s 17.9 million residents represent 72% of the three-basin population and control the aggregate.

The second investigation takes that regime-aware chemistry and runs a 5-dimensional uncertainty analysis across 24,576 quasi-random draws to formally quantify ozone-death uncertainty. Inputs varied: the dose-response slope, the NOx reduction fraction, the LA Basin chemistry ratio, model ozone sensitivity, and meteorological variation. The chemistry threshold classification uses Sillman (1995) VOC/NOx indicator thresholds — stricter than the Jin et al. (2020) threshold of 8.0. That tightening moved Sacramento into the VOC-limited regime and shrank what had been a small ozone benefit there by approximately 67%. Result: the distribution of ozone deaths is centered at −453.5 avoided deaths per year (meaning 453.5 net deaths caused), with only P(net benefit) = 0.0005 of draws showing a positive outcome.

Two-panel comparison showing P(net benefit) dropping from 3.3% under Jin et al. thresholds (left) to 0.05% under Sillman 1995 thresholds (right). The rightmost panel histogram is almost entirely negative.
Figure 1. Pre- vs. post-Sillman regime threshold comparison. Switching from Jin et al. (2020) to Sillman (1995) VOC/NOx thresholds shrinks P(net benefit) 66×, from 3.3% to 0.05%. The reclassification of Sacramento from NOx-limited-side transitional to VOC-limited-side transitional is the primary driver.
Bar chart of total-order and first-order Sobol indices for five inputs. NOx reduction fraction leads at ST=0.50; beta_O3 second at ST=0.425.
Figure 2. Sobol sensitivity indices: what drives ozone-deaths uncertainty. NOx reduction fraction (ST = 0.50) and the Turner CRF slope βO3 (ST = 0.425) carry the largest first-order shares (S1NOx = 0.373, S1βO3 = 0.250; combined first-order attribution ≈ 62% of output variance). Total-order indices include interaction terms and sum to > 1; the ST−S1 gap on βO3 signals strong NOx×βO3 coupling. Meteorological jitter is the lowest driver (ST = 0.084).

99.95% of scenarios: the “ozone co-benefit” is an ozone harm

Across 24,576 Saltelli draws spanning realistic ranges of the Turner CRF, the NOx reduction fraction, the LA Basin VOC/NOx ratio, ISRM ozone sensitivity, and WRF-Chem meteorological jitter, only 0.05% of draws return a positive deaths-avoided figure for the ozone channel. The mean is −453.5 deaths-avoided per year (i.e., 453.5 net deaths caused), with P5 = −943.5 and P95 = −128.2. The finding is robust across the full prior envelope — not a borderline result.

Sillman (1995) thresholds tighten the conclusion 66× relative to the prior analysis

Before applying Sillman (1995) VOC/NOx indicator thresholds, the pre-Sillman run returned P(net benefit) = 0.033 — a 3.3% chance of net ozone benefit. After applying the Sillman thresholds (which reclassified Sacramento as VOC-limited-side transitional), P(net benefit) dropped to 0.0005 — a 66× reduction. The tighter conclusion reflects published peer-reviewed regime chemistry calibrated against California basin VOC/NOx measurements. Citation: Sillman S. (1995), J. Geophys. Res. 100(D7):14,175–14,188.

The size of the NOx cut drives most of the uncertainty; the dose-response function is second

The Sobol total-order index for the NOx reduction fraction is 0.500 (36% of ΣST), reflecting that the magnitude of the harm scales directly with how aggressively NOx is cut in a VOC-limited basin. The Turner CRF comes second at ST = 0.425 (30% of ΣST), with a large S1→ST gap (0.175) indicating substantial multiplicative coupling with the NOx cut. The ISRM ozone sensitivity (LA Basin) ranks third at ST = 0.235. Meteorological jitter is the lowest driver at ST = 0.084 — the conclusion is not primarily meteorology-sensitive.

The LA Basin ozone result flips sign when chemistry regime is accounted for: −1.84 to +2.10 ppb

The first-pass linear treatment predicted a 1.84 ppb ozone reduction in the LA Basin under a 20% NOx cut — a benefit. Regime-aware chemistry (a CMAQ isopleth correction calibrated on Jin et al. 2020 California isopleths) predicts the opposite: +2.10 ppb — more ozone, not less. In a VOC-limited basin, NOx scavenges OH radicals that would otherwise react with VOC and suppress O3; removing NOx releases those radicals into the photochemical cycle and elevates ozone. The linear framework is directionally wrong for the LA Basin, which holds 17.9 million residents.

The fidelity gap between linear and regime-aware ozone accounting is worth ~$10.4B in our cascade

Our first-pass linear treatment of the ozone channel reported a $7.3B ozone co-benefit (Turner CRF, $11.6M VSL). The regime-aware second pass returns −$3.1B (a disbenefit). The $10.4B swing is the dollar exposure of using a linear air-quality model in a VOC-limited basin. The chemistry isn’t novel — what’s useful here is the cascade-scale dollar quantification. Any ratepayer-benefit estimate built on linear co-benefit accounting in the LA Basin is going to carry an error of about this size; the correction is to switch to regime-aware fidelity for the ozone channel.

This element feeds two other elements: