U.S. West Coast droughts and heat waves exacerbate pollution inequality and can evade emission control policies

Air pollution in the United States remains associated with 100,000–200,000 premature deaths per year, with over half from fossil fuel combustion and about 10% from electric power plants. Hydrometeorology, including droughts and heat waves, significantly influences power sector emissions, particularly in California where hydropower typically supplies a notable share of electricity and heat waves raise cooling demand. These conditions force greater reliance on fossil generators, increasing emissions of SO2, NOx, and PM2.5 with associated health risks. The study investigates how droughts and heat waves affect the spatial (county-level) distribution of health damages in California and whether these extremes exacerbate existing pollution inequities affecting people of color and communities with high pollution burdens. It also examines whether emissions control policies, such as taxes on emissions, are effective under extreme conditions, noting that severe heat waves can create supply scarcity that may limit policy effectiveness. The purpose is to quantify links between hydrometeorological variability, grid operations, emissions, health damages, and inequality, and to stress-test emissions control policies under extreme weather.

Prior work shows hydrometeorology drives variability in power plant emissions and market outcomes in California, and that air pollution burdens disproportionately impact marginalized communities and people of color. Despite evidence that taxes or financial penalties can incentivize cleaner dispatch under normal conditions, there has been no rigorous weather stress-testing of such policies for the power sector. The study addresses this gap by linking synthetic extreme weather variability to county-level health damages and equity metrics, and by evaluating policy effectiveness during droughts and heat waves.

The study uses the CAPOW (California and West Coast Power System) open-source, Python-based framework to simulate hourly operations of the 2018 West Coast bulk power system with focus on CAISO. CAPOW comprises: (1) a zonal unit-commitment/economic dispatch (UC/ED) model with five aggregated zones (four in California, one in the Pacific Northwest) linked via transmission pathways, optimizing least-cost dispatch over a 48-hour horizon subject to demand, reserves, interchanges, transmission capacities among zones, and generator operating limits; and (2) a stochastic engine generating 500 synthetic years of daily hydrometeorological data (streamflow, air temperature, wind speed, solar irradiance) across multiple observation sites on the West Coast. Synthetic weather generation: Daily temperature and wind speed residuals and irradiance losses are derived relative to 365-day mean profiles, transformed to Gaussian space, and modeled with vector autoregressive (VAR) processes to preserve temporal and cross-site correlations. Irradiance data derive from NSRDB; temperature and wind from NOAA; streamflows from BPA and CDEC. Annual streamflow and temperatures (via HDDs and CDDs) are jointly sampled using Gaussian copulas based on 1954–2008 streamflow and 1958–2008 temperatures, then disaggregated to daily using a method that preserves multi-scale dependencies and temperature–snowmelt relationships (allowing earlier snowmelt in warmer years via nearest historical analogs for daily flow fractions). Synthetic hydrometeorological series are mapped to power system inputs: daily hydropower availability (via mass-balance hydrologic models for major reservoirs and machine learning for high-altitude hydro, plus scaled outputs), hourly wind and solar generation, and hourly zonal electricity demand (via multivariate regressions linking weather to power outputs and load; residuals represented by VAR). Hourly values are resampled from historical BPA and CAISO series conditioned on daily states and day-of-year. Emissions and damages: For each generator, emissions of CO2 are estimated via EPA heat rates and fuel carbon intensity; local pollutants (PM2.5, SO2, NOx) via EPA-reported emission rates (tons/MWh). Health damages from local pollutants are monetized using plant-specific $/MWh rates from the AP3 integrated assessment model, which simulates dispersion and chemistry (for SO2 and NOx) to county-level PM2.5 concentrations, applies concentration–response functions, and monetizes mortality risk (using 2017 populations; static rates across days/years). Market prices are computed as duals of zonal balance constraints and aggregated to a hub price via weights fit to historical CAISO prices. Policy scenarios: Four policy cases are simulated over the same 500-year ensemble: (1) base case (no emissions penalties); (2) local tax (penalties on PM2.5, SO2, NOx per AP3 $/MWh); (3) CO2-only penalty (EPA Social Cost of Carbon, $47.38/ton in 2018); (4) combined local + CO2 penalties. Penalties are applied to CAISO generators only (not PNW) by adding to marginal costs, potentially altering dispatch merit order. Equity metrics use CalEnviroScreen 4.0: racial inequality is the correlation between county damages and percent people of color; pollution burden inequality is the correlation between county damages and county CalEnviroScreen score.

- Damages peak in hot, dry years: Annual local air pollution damages are positively correlated with CAISO demand (R=0.62) and negatively correlated with hydropower availability in CAISO (R=-0.86) and the PNW (R=-0.56). Dry years coincide with high electricity prices and higher damages. - Inequality increases in hot/dry years: Counties with majority people of color and higher pollution burden see disproportionate increases in damages when demand is high and hydropower is low. Average annual per-capita damages (base case): people of color $22.46 vs White $19.56. - Policy effectiveness on average: The local air tax reduces average annual damages from $840,818,811 ($21.27 per capita) to $260,675,737 ($6.59 per capita), a 69% reduction. Per-capita damages fall to $6.60 for people of color (71% decrease) and $6.58 for White residents (66% decrease). - Extremes evade policy: During severe late-summer heat waves (very high demand when hydropower is seasonally low), system operators must use nearly all fossil fuel capacity, including high emitters, to avoid blackouts. On these rare days, the local tax does not reduce damages; dispatch with and without the tax is nearly identical, and market prices can spike to $1000/MWh. The worst day uses 98.7% of fossil fuel capacity and experiences reliability shortfalls. - Temporal patterns: Drought drives chronic, multi-month increases in damages (especially Dec–Jul via reduced hydropower), whereas heat waves cause acute, single-day spikes with the highest damages (e.g., July 30 heat wave in the worst-day year). - Day-level taxonomy of tax performance (Fig. 5): • Zone a (~3.2% of days): Spring snowmelt, high hydro, mild temps, low demand—strong linear prevention of damages by local tax. • Zone b (~75.9%): Mixed conditions—tax remains effective; positive relation between base damages and prevented damages. • Zone c (~20.4%): Late summer/early fall, low hydro, high demand—tax still shifts away from most damaging plants. • Zone d (~0.6%): Heat-wave days—tax efficacy collapses; nearly all fossil capacity needed, preventing reordering away from high emitters. - Supply-side shifts under tax (Fig. 6): On most days (zones a–c), generation shifts from higher-damage ($/MWh) units toward lower-damage units. Under extreme demand (zone d and worst cases), reliance on higher-damage units (6–10 $/MWh) increases, and in the most extreme case, dispatch is unchanged by the tax.

The study links hydrometeorology to operational shifts in the West Coast grid, demonstrating that drought (reduced hydro) and heat waves (increased demand) raise fossil generation and health damages, with hot/dry years producing the worst outcomes. It shows these effects exacerbate existing inequalities, disproportionately impacting counties with majority people of color and higher pollution burden. Emissions penalties on local pollutants are generally effective in reducing health damages by shifting dispatch toward less harmful plants. However, during extreme late-summer heat waves, scarcity forces near-full utilization of fossil capacity, rendering taxes temporarily ineffective and leaving populations exposed to acutely high pollution damages. These insights address the core questions by quantifying how weather extremes drive damages and inequities and by stress-testing policy efficacy. The results are relevant for grid planning and public health protections: scarcity management, demand response compensation reflecting air quality damages, and blackout planning should consider spatial inequities in exposure and timing to protect highly affected communities.

The paper demonstrates that hydrologic variability (especially drought) and heat waves markedly increase power-sector air pollution damages in California and worsen pollution inequality. While emissions taxes on local pollutants substantially reduce average damages and exposures across most days, they do not mitigate damages during rare, extreme heat-wave-driven scarcity events when nearly all fossil capacity is required. Planning and policy should incorporate air quality damage costs into scarcity management, demand response, and outage strategies, with attention to protecting communities bearing the highest burdens. Future research should: (1) incorporate dynamic, weather-sensitive health damage functions reflecting air chemistry and deposition; (2) employ higher-resolution grid models capturing transmission congestion and outages; (3) assess impacts under decarbonization trajectories and climate change; and (4) evaluate targeted interventions for extreme events (e.g., enhanced demand response, storage, and equity-focused curtailment policies).

- Simplified grid representation: Aggregated five-zone topology with transmission constraints among, not within, zones; does not capture detailed intra-zonal congestion. - No forced outages: Generator/transmission outages are not modeled, likely underestimating frequency of extreme scarcity, blackouts, and instances where taxes fail to reduce damages. - Static damage rates: AP3-based $/MWh health damage rates are fixed over time and do not vary with meteorology; this may underestimate high-damage days (e.g., ozone formation in high temperatures) and overestimate low-damage periods. - Stationary weather and 2018 system: Analyses use stationary hydrometeorological uncertainty and the 2018 grid; do not include climate change or future decarbonization. - Policy scope: Emission penalties applied only to CAISO generators (not PNW), potentially omitting cross-regional feedbacks. - No explicit emissions caps: Policies modeled as marginal cost adders rather than binding limits; on extreme days, reliability constraints dominate cost minimization and emissions damages.