1 Introduction

The Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC 2007) identifies greenhouse gas (GHG) emissions as a strong contributing factor to climate change. California accounts for 0.53 % of the world’s population but 1.6 % of the world’s GHG emissions due to the high level of economic activity in the state. California Governor Arnold Schwarzenegger signed the Global Warming Solutions Act of 2006 (AB 32) mandating that California GHG emissions should return to 1990 levels by the year 2020 ((Assembly Bill No. 32 Chapter 488 2006). Opponents of AB 32 argue that the law will hinder California’s economy because the costs of compliance outweigh the benefits of climate change mitigation. This argument does not account for the co-benefits that AB 32 may provide through improved air quality due to a simultaneous reduction in traditional criteria pollutant emissions such as particulate matter (PM), oxides of nitrogen (NOx), oxides of sulfur (SOx), and reactive organic gases (ROGs). The potential for an air quality co-benefit is significant since California currently experiences some of the highest air pollution concentrations in the United States.

Quantifying the potential air quality co-benefits of AB 32 is a complex undertaking because atmospheric chemistry is inherently non-linear. A decrease in NOx emissions reduces the formation of secondary PM at low NOx/ROG ratios but increases the formation of secondary PM at high NOx/ROG ratios. The emissions changes associated with AB 32 may therefore not produce proportional reductions in PM2.5 concentrations. A full analysis with a chemical transport model is necessary to understand the impact of AB 32 on population exposure to PM2.5. Previous studies suggest that air quality benefits associated with climate change mitigation policies may be appreciable (Nemet et al. 2010). Reduced air pollution exposure due to GHG mitigation measures can avoid up to 13,000 premature deaths per million population (Wilkinson et al. 2009; Woodcock et al. 2009; Markandya et al. 2009; Friel et al. 2009; Smith et al. 2009; Haines et al. 2009) leading to substantial economic benefits that can compensate for GHG mitigation costs (Bollen et al. 2009; Netherlands Environmental Assessment Agency 2009). Predicted air quality and health co-benefits vary substantially depending on the mitigation approach and the application regions.

The purpose of the present study is to quantify how AB 32 could impact ground level concentrations of airborne particulate matter with aerodynamic diameter less than 2.5 μm (PM2.5) in California. The measures outlined in the AB 32 Scoping Plan (California Air Resources Board 2008c) are analyzed to determine associated changes in criteria pollutant emissions during the year 2030, 10 years after the 2020 target year for full AB 32 implementation. The effects of AB 32 measures on different criteria pollutant emissions and ambient PM2.5 concentrations are analyzed during an extreme 12-day SJV stagnation episode and over a time period representative of an annual average. These analysis periods were selected based on 7 years of General Circulation Model predictions downscaled using regional climate models (Zhao et al. 2011a, b; Mahmud et al. 2010). The effects of AB 32 measures on PM2.5 mass and PM2.5 nitrate, sulfate, elemental and organic carbon concentrations are quantified over the entire state of California using a regional air quality model that accounts for atmospheric transformation. The potential public health implications are then inferred from population-weighted concentrations combined with mortality estimates derived from epidemiological studies. Part 2 of this study will examine the air pollution co-benefits from the transportation sector of California Governor’s Executive Order S-3-05 which calls for an 80 % reduction in GHG emissions, relative to 1990, by the year 2050.

2 Methods

2.1 Determination of criteria pollutant emissions changes associated with AB 32

AB 32 contains numerous measures to reduce greenhouse gas emissions that span multiple economic sectors (California Air Resources Board 2008b, 2010c). Criteria pollutant emissions scaling factors for each measure were calculated based on Environmental Impact Reports and/or energy consumption reports. Criteria pollutant emission changes not defined by AB 32 were obtained from supplemental information sources such as US EPA and/or these criteria pollutant emission scaling thresholds were set based on practical limits of market adoption as discussed in the following sections. Emissions control levels were taken directly from AB 32 or inferred from energy efficiency improvements. When ranges for emissions reductions or efficiency improvements were given in the reports referenced, the average value within the range was used. This assumption generates ≤0.2 % uncertainty in the statewide emissions for each of the criteria pollutants.

AB 32 measures were organized into five categories based on economic sector or sub-sector (see Table 1). These categories have been described in this study as “implementation levels” because they were applied in a cumulative fashion (each implementation level builds on effects of previous levels; see Figs. 1 and 2) to quantify how controls for each sector contribute to PM2.5 reduction. The cumulative total of all implementation levels is the total emission or air quality impact of AB 32 from all sectors. Note that AB 32 does not define implementation levels and this method of organization is used only to provide greater understanding of the underlying relationships between AB 32 and PM2.5 concentrations in California. A summary of the BAU emissions and the emission changes associated with each implementation level is shown in Table 2. A more detailed summary of the emissions changes associated with individual AB 32 measures is provided in Supplemental Information.

Table 1 Categorization of AB 32 measures into five cumulative levels. Cumulative refers to the inclusion of all measures incorporated in previous economic sector implementation levels (see Fig. 2). Not all AB 32 Scoping Plan measures are shown here; only measures considered by this study
Full size table
Fig. 1
figure 1

Reductions or increases caused by each AB 32 implementation level for total particulate matter mass (PM), particulate elemental carbon (EC), particulate organic carbon (OC), oxides of nitrogen (NOx), oxides of sulfur (SOx), reactive organic gases (ROG), and ammonia (NH3)

Full size image
Fig. 2
figure 2

Relative change in emissions rates caused by each cumulative AB 32 implementation level for a PM and b NOx, where “Scen 0” is the BAU scenario. The difference between implementation levels is the change in emissions from the sources in that category. Levels 1–5 correspond to 1) industrial, 2) electric utility and natural gas, 3) agricultural, 4) on-road mobile, and 5) other mobile emission sources. Each “scenario” or model simulation is the cumulative of each additional level of emission reductions. Levels 2 and 3 are linked since part of the electrical generation capacity is shifted to methane combustion systems on large dairies

Full size image
Table 2 Statewide 2030 daily average PM2.5 emission rates and changes to emissions associated with AB 32. Uncertainty estimates in column three reflect the results of a sensitivity analysis conducted using the high and low estimates for control levels (where available)
Full size table

2.1.1 Emission changes from industrial sources—Implementation Level 1

AB 32 Implementation Level 1 includes emissions control measures that reduce GHG emissions from large industrial facilities. The measures incorporated into Level 1 include energy efficiency audits, refinery flare recovery, natural gas distribution improvements, and increased use of composting to generate biogas for electricity generation. Industrial point sources that emit >0.5 MMTCO2e per year such as oil and gas extraction, hydrogen plants, and mineral plants were affected by the energy efficiency audits measure (California Air Resources Board 2009b). All refineries and cement production plants were included in the energy efficiency audits measure even if they emitted <0.5 MMTCO2e because these industries were identified as extremely energy intensive in the AB 32 Scoping Plan. Energy efficiency improvements stemming from the audits were assumed to reduce energy consumption for cement plants (6.25 %), refineries (15 %), and all other facilities (10 %) with equivalent reductions in criteria pollutant emissions (Worrell and Galitsky 2005, 2008; Coito et al. 2005; Xenergy Inc. 2001). Several of these values represent the average of bounding estimates for possible efficiency improvements; see the discussion of sensitivity analysis associated with Table 2. Other Scoping Plan measures targeted the oil and gas industrial processes that emit methane and CO2 such as oil and gas extraction, transmission, distribution, flare recovery and removal of methane exemption at refineries (California Air Resources Board 2008e). Together CO2 and CH4 make only minor direct contributions to urban and regional particulate air pollution concentrations in California and so these specific measures have little impact on the results in the current study. Implementation Level 1 also includes commercial and industrial waste management measures from the AB 32 Scoping Plan such as methane control at landfills. The landfill gas recoverable for electricity generation has the potential to increase current landfill gas energy production by 36 % (Environmental Protection Agency 2011; R.W. Beck Inc. 2006) hence reducing commercial, institutional and government solid waste landfill TOG emissions, but offsetting less than 1 % of the state’s electricity generation. All emissions associated with electrical generation from biogas were represented using the appropriate chemical profiles for these facilities with emissions rates scaled by total electrical output.

2.1.2 Emission changes from electrical generation and natural gas sources—Implementation Level 2

AB 32 Implementation Level 2 defined in this study applies to traditional utilities (electricity and natural gas). The effects of measures that impact the consumption and generation of electricity from traditional power plants were adjusted to recognize that California imports ~33 % of its electricity demand from out-of-state generation (California Energy Commission 2009a). As a result, the decreases in electricity consumption forecasted for 2030 mitigated electricity generation for both in-state and out-of-state sources. The current study assumed that the relative portfolio of fossil fuel and renewable energy generation for in-state and out-of-state sources would remain at 2008 levels in the absence of AB 32. The expected electricity consumption changes for all AB 32 measures were then first applied to fossil fuel sources, thus eliminating coal and natural gas electricity generation out of state (leaving large hydro, renewable, and nuclear out-of-state imports), followed by elimination of in-state coal and petroleum electricity generation, and reduction of in-state natural gas electricity generation (in rank order of preference).

The measures included in Level 2 broadly cover residential and commercial natural gas consumption associated with appliances (Kavalec and Gorin 2009; Flex Your Power 2007) and building construction and retrofits that have improved energy efficiency standards, solar water heating (California Air Resources Board 2008b), and overall reduced water consumption leading to lower water heating needs (American Gas Association 2011; Klein 2005). Specific measures within the Level 2 category include: (1) “greening” of new and existing schools, state, residential, and commercial buildings, (2) water recycling, runoff reuse, and efficiency measures, and (3) the Renewable Portfolio Standard (RPS) and million solar roofs initiative. Measures to “green” buildings were applied to both existing and new buildings. New green buildings were assumed to adopt building standards that reduced energy consumption by 6.8 % and appliance standards that reduced energy consumption by 8.8 % (California Energy Commission 2009a; Kavalec and Gorin 2009). Energy efficiency retrofits of existing buildings was assumed to reduce average building energy consumption by 7.5 % (California Energy Commission 2005, Kavalec and Gorin 2009b). Green buildings led to the largest reduction in electricity demand from the electricity measures. Water conservation measures reduced the energy associated with the transport, distribution, and treatment of water (Klein 2005). The specific measures included in the AB 32 Scoping Plan address water use and system efficiency (Gleick et al. 2005; State Water Resources Control Board 2008), recycling (Department of Water Resources 2003), and runoff reuse (Garrison et al. 2009) translating to a 5.2 % reduction in electrical consumption associated with water use in California. Energy efficiency goals reduced annual electrical demand in the state of California by 32,000 GWh (8 %) with equivalent reductions in criteria pollutant emissions in the year 2030 (California Air Resources Board 2008c). The installation of 4GW of combined heat and power units to utilize waste heat for electricity generation displaced 30,000 GWh of electricity generated from fossil fuel combustion in the year 2030 (California Air Resources Board 2008a).

The Renewable Portfolio Standard (RPS) aims to increase the use of renewable energy sources for electricity generation from 12 % in 2010 to 33 % by 2030. Renewable electricity generation can be accomplished by technologies with essentially no criteria pollutant emissions such as wind, solar photovoltaic, and small hydropower. Emissions from solar thermal, and geothermal electricity generation were increased in proportion to the change in energy generation from each technology (California Air Resources Board 2010a). Renewable electricity generation also includes combusting fuels that emit criteria pollutants such as solid fuel biomass, landfill gas, and anaerobic digester gas. The combined effect of the RPS measure increased criteria pollutant emissions from renewable sources within California (California Air Resources Board 2010). The net effect of all the measures contained within the Level 2 category defined in the current study reduced in-state electricity generation from fossil fuels (Electric Power Group LLC 2004; California Energy Commission 2009a) and increased electricity generation from renewable fuels, leading to reduced emissions for all criteria pollutants except PM.

2.1.3 Emission changes from agricultural sources—Implementation Level 3

Under Implementation Level 3 in the current study, “large dairies” were retrofitted with anaerobic digesters for methane capture and electricity generation. Large dairies were not explicitly defined by AB 32, but the US EPA defines a large dairy to be ≥500 dairy cows per facility (United States Environmental Protection Agency 2004). Over 1,000 dairy farms in California have ≥500 dairy cows, accounting for over 90 % (1.67 million) of all CA dairy cows in 2007 (USDA 2009). Adoption of biogas digesters on all of these dairy farms was considered to be prohibitively expensive. In the current study, “large dairies” were defined as those with projected emissions of total organic gases (TOG) >5,100 kg/h in the year 2029. A total of 16 dairies meeting this criteria were assumed to install plug flow anaerobic digesters (Western United Resource Development Inc. 2009) yielding a total electrical generation capacity of 54.8 MW. Each digester was assumed to produce an average of 0.166 kW/ft3 of methane combusted with no flaring. All digester units were further assumed to incorporate H2S scrubbing to meet the SCAQMD criteria of 40ppmv H2S in the emissions to prevent corrosion of the exhaust system and to prevent formation of sulfate aerosol in the atmosphere. The CO, NOx, VOC, and PM emissions from each digester were specified at the level described by the SJV Best Available Control Technology (BACT) per MWh (San Joaquin Valley Unified Air Pollution Control District 2008). Ammonia emissions were reduced substantially since ammonia was combusted with the biogas rather than released into the atmosphere directly from the dairy waste. Based on work by Shaw and colleagues (Shaw et al. 2007), most of the VOC emissions from dairy cows are produced by animal respiration and enteric fermentation, and therefore the TOG area emissions were reduced by only 1 % due to dairy waste anaerobic digestion.

2.1.4 Emission changes from on-road mobile sources—Implementation Level 4

All AB 32 measures that reduced on-road vehicle emissions through use of alternative transportation modes, reduced energy intensity, or altered the fuel mix were incorporated into Implementation Level 4 in the current study. For example, High Speed Rail (HSR) was assumed to reduce highway travel (Federal Railroad Administration 2005) and thus criteria pollutant emissions from on-road gasoline engines. Transportation mode shifts associated with HSR were based on the calculated reduction of vehicle miles traveled (VMT) and air travel for city-to-city trips (Federal Railroad Administration 2005). Energy intensity was decreased through the Pavley measures I and II (California Air Resources Board 2007a, 2008b), which call for increased efficiency through higher average fuel economy for the passenger fleet (California Air Resources Board 2008d, e). Energy intensity was also decreased by the tire tread and inflation program for light duty vehicles, reduced fuel consumption through hybridization of commercial vehicles, and assumed aerodynamic efficiency gains for heavy duty vehicles. Fuel mixtures were affected by the Low Carbon Fuel Standard (LCFS) (California Air Resources Board 2009), which increased the production of alternative ethanol and biodiesel fuel feedstock while incorporating biofuels and compressed natural gas into the fuels for the light duty vehicle fleet. Fuel mixtures were also affected by increased market penetration of Zero Emission Vehicles (ZEV) powered by electricity and hydrogen fuel cells. It was assumed that 0.330 million (0.33 M) ZEVs would be driven on California roads by 2030 in accordance with the 2020 LCFS target (California Air Resources Board 2009). The LCFS further assumes the penetration of other advanced technologies or emerging vehicles into the marketplace by the year 202, including battery electric (0.22 M vehicles), plug-in hybrid electric (0.67 M vehicles), and fuel cell (0.11 M vehicles). It is forecasted that there will be 38.95 M vehicles registered in California by 2030 (California Department of Transportation Division of Transportation System Information 2009) and therefore ZEVs would account for ~1 % of the total fleet while other advanced technologies would account for 2.6 % of the total vehicle fleet. Diesel vehicles were influenced by additional measures related to goods movement. Specifically, drayage trucks operating within 80 miles of the Ports of San Diego, Long Beach, Los Angeles, Hueneme, Oakland, and San Francisco were required to install diesel particulate filters (DPFs) or otherwise meet the 2007 tailpipe emissions standards for heavy duty diesel vehicles (California Air Resources Board 2007b, 2010b). The regional goods movement system-wide efficiency goal was assumed to target heavy duty diesel vehicle emissions within ports (California Air Resources Board 2008e). Additional measures that applied to diesel vehicles included improvements to heavy duty aerodynamic efficiency and transport of alternative fuel feedstock through diesel trucks. The mobile fleet emission factors for PM and NOx were based on default MOBILE6 factors which include Tier 2 standards when projected to 2030. SCR controls were not included in this analysis.

2.1.5 Emission changes from other mobile sources—Implementation Level 5

Implementation Level 5 includes measures that affect emissions from non-road modes of transportation such as air, water, rail, or off-road. The Goods Movement Emission Reduction Plan (GMERP) describes measures that were originally designed for reducing criteria pollutants related to international trade and the flow of goods (California Air Resources Board 2006a). Although its original purpose was to reduce adverse health impacts in the communities near ports, some of the measures in the GMERP were also incorporated in the Scoping Plan due to the fuel conservation and efficiency approaches that lead to CO2 emission reductions. The LCFS calls for transport of alternative fuels by rail leading to increased emissions from locomotives (California Air Resources Board 2009). HSR was also assumed to decrease the number of airplane trips within California (Federal Railroad Administration 2005) but slightly increased electrical demand. Cargo handling equipment was assumed to reduce idling and undergo hybridization and electrification (California Air Resources Board 2006b). Transport Refrigeration Units (TRUs) were assumed to be off when not in transit (California Air Resources Board 2008b). Several goods movement measures reduce emissions from ocean-going vessels or harbor craft. Cleaner ships that employed efficient engine designs, regular engine maintenance, fuel efficient steering through a vessel speed reduction near land, and even electrification at ports greatly reduced criteria pollutant emissions from ships (California Air Resources Board 2006a). Construction equipment and generators were not included in the other mobile source category since these sources were not addressed by AB 32.

2.2 Air quality modeling

Raw emissions inventories were based on the statewide inventory produced by the California Air Resources Board (CARB) projected for the year 2029 under compliance of the statewide implementation plan (SIP). The on-road mobile emissions inventory for the South Coast Air Basin (SoCAB) was provided by the South Coast Air Quality Management District (SCAQMD) for San Diego, Imperial, San Bernardino, Riverside, Los Angeles, and Orange counties. All other stationary, mobile, and area-wide emissions in the SoCAB were represented using the CARB statewide emissions inventory.

The CARB statewide raw emissions inventory had a spatial resolution of 4 × 4 km on a Lambert Conformal projection of the earth’s surface forming a grid with 190 × 190 cells. The SCAQMD raw emissions inventory had a spatial resolution of 5 × 5 km on a Universal Transverse Mercator (UTM) coordinate system. The SCAQMD emissions were mapped to the Lambert projection and all calculations were performed on this grid. Both emissions inventories had temporal profiles for individual sources to model seasonal and/or day-of-week variations in emissions rates. The SCAQMD on-road mobile emissions inventory had diurnal scaling to represent average daily traffic patterns. All on-road mobile emissions and biogenic emissions were adjusted for variations in temperature and humidity (Mahmud et al. 2010). Soil dust was treated as a primary area source emission. None of the AB 32 measures limit emissions of soil dust, and so this source was constant for all simulations. The raw emissions were processed with detailed gas composition profiles and PM size and composition profiles to create inputs for modeling. Emissions were processed in eight different categories corresponding to the AB 32 implementation categories: electric utility, residential and commercial natural gas, on-road gasoline mobile, on-road diesel, other mobile, industrial, agricultural, and other miscellaneous.

The future air pollution episodes used to evaluate the air quality impacts of AB 32 were identified by downscaling Global Circulation Model (GCM) results produced by the Parallel Climate Model (PCM) using the Weather Research and Forecasting (WRF) regional climate model for the years 2047–2053 (Mahmud et al. 2010). The stagnation episode predicted for January 1–16, 2050 was selected as the representative meteorological input conditions for a severe future air pollution episode observed in the SJV (Mahmud et al. 2010). Nine episodes lasting 16 days each distributed evenly through the year 2052 were selected as representative meteorological conditions for future annual averages (Mahmud et al. 2010). Previous studies show that the episodes are representative of the 7-year average future meteorology in California (Mahmud et al. 2010).

Both meteorological fields and processed emissions were provided as inputs into the UCD-CIT source-oriented 3D Eulerian photochemical airshed model. The UCD model represented PM using 15 logarithmically scaled size bins from 0.01 to 10.0 μm. Gas phase photochemical reactions were represented using the SAPRC90 chemical mechanism (Ying et al. 2007) and dynamic gas to particle conversion reactions were represented using the treatment described by Jacobson (Jacobson 2005). The vapor pressures of inorganic gases above each particle were calculated using a modified version of ISORROPIA (Nenes et al. 1998). An absorption model based on smog chamber experiments was used to predict secondary organic aerosol formation (Ying et al. 2007). Pollutant transport from sources outside California is generally minor since the prevailing wind condition during air pollution events is west to east with the Pacific Ocean on the upwind side of the simulation. Hemispheric background conditions over the Pacific Ocean were used to represent long range transport. The background concentrations used in the current study were based on conditions measured during the California Regional Particulate Air Quality Study (CRPAQS) as used in previous modeling exercises (Ying et al. 2008).

Air quality simulations were run at a resolution of 8 × 8 km for each 16 day simulation period. The four initial days were omitted from the final data analysis to reduce the influence of initial and boundary conditions. Six air quality simulations were run separately during the extreme stagnation period (Jan 2050) to account for the change in air pollution associated with each AB 32 implementation level. Together, this creates a “staircase” of emission changes, with each step representing a different cumulative level of AB 32 implementation (see Fig. 2). The final level represents the change of emissions from all AB 32 measures. Two air quality simulations were run over the annual average episodes to evaluate air quality under BAU conditions and under full adoption of all AB 32 measures.

3 Results

3.1 Business-as-usual PM2.5 concentration during an extreme pollution event

Business as Usual (BAU) PM2.5 mass concentrations were predicted to be highest near the cities of San Francisco, Los Angeles, Fresno and Stockton during the extreme stagnation episode. The maximum 12-day average PM2.5 mass concentrations reach 25–30 μg m-3 at these locations with sharp spatial gradients around urban centers reflecting the signal from primary combustion sources in addition to the regional dust and ammonium nitrate signals. Predicted PM2.5 mass concentrations reach 22 μg m−3 in Imperial county due to windblown dust. Sea salt aerosol from breaking waves accounted for approximately 10 μg m−3 over the ocean cells but only trace amounts of this material reaches inland locations. The major components of the particulate matter mass include elemental carbon (EC), organic carbon (OC), sulfate, nitrate, and ammonium ion. EC and OC are produced by primary combustion sources that are clustered in cities and along transportation corridors. Sulfate concentrations peak at the locations of maximum shipping activity around the ports of Los Angeles, San Francisco and Oakland and in offshore shipping lanes. The highest concentrations of EC, OC and sulfate occur in the port of San Francisco and Oakland. Ammonium nitrate concentrations are predicted to reach a maximum of approximately 7.3 μg m−3 in the SJV where emissions of ammonia from agricultural sources are highest.

3.2 Impact of AB 32 on PM2.5 concentrations during an extreme pollution event

Figure 4 displays the change in PM2.5 concentrations caused by various levels of AB 32 implementation during the simulated extreme pollution event. AB 32 defined Implementation Level 1 and 2 produced large PM2.5 mass reductions at specific industrial and electric utility point sources with exceptionally high emission rates. Figure 3b and c show reductions of 1.8 μg m−3 PM2.5 mass near Concord (industrial Level 1 controls) and reductions of 4.3 μg m−3 PM2.5 mass near Salinas (electric utility and natural gas Level 2 controls) over the 8 × 8 km2 or 16 × 16 km2 area surrounding each facility. Regional effects from these point source controls are muted with changes <1 μg m−3. The dairy manure digester and electricity generation implemented in the SJV under defined Implementation Level 3 causes an increase in PM2.5 mass of about 1 μg m−3 mostly due to increased concentrations of secondary nitrate and sulfate (see Supp. Info.). These components draw more ammonium into the particle phase in the SJV resulting in reduced transport of ammonium to surrounding regions and therefore reduced ammonium nitrate concentrations in surrounding regions. AB 32 defined Implementation Level 4 reduces concentrations of PM2.5 mass associated with on-road vehicles by a maximum amount of 0.7 μg m−3 in San Francisco and Los Angeles with lesser reductions in the SJV and regions of slight (~0.2 μg m−3) increase in the foothills of the mountains surrounding the SJV. Reduced NOx emissions in the SJV lead to reduced nitrate concentrations in the SJV which allows more ammonia to be transported to the surrounding regions where it can contribute to particulate ammonium nitrate formation. AB 32 defined Implementation Level 5 (other mobile AB 32 measures) reduces PM2.5 concentrations by 5 μg m−3 in the areas immediately around the ports of Los Angeles and San Francisco due to decreases in EC and sulfate concentrations associated with shipping measures. Figure 4a summarizes that the net effect of all the AB 32 measures within Implementation Levels 1–5 will reduce PM2.5 mass concentrations by ~2–3 μg m−3 over the major urban areas of California with peak reductions of 5.5 μg m−3 at the ports and around major point sources during the simulated severe winter stagnation event. However, the addition of 54.8 MW of electrical generation capacity from dairy biogas digesters in the SJV leads to increased PM2.5 mass concentrations of 0.3 μg m−3 at the location where the ammonia emissions are the highest.

Fig. 3
figure 3

Business as Usual (BAU) PM2.5 concentrations (without AB 32 controls) during the January 2050 12-day severe winter stagnation event. PM2.5 concentrations are shown for panel a total mass, and component contributions from b EC, c OC, d sulfate, e nitrate and f ammonium ion. All results are in μg m−3

Full size image
Fig. 4
figure 4

Changes in PM2.5 mass concentration (μg m−3) for all defined AB 32 Implementation Levels for the January 2050 episode. Panel a illustrates cumulative effects of all Levels while panel b–f illustrate individual changes associated with each Level relative to the previous Level. Red indicates increased concentrations while blue indicates decreased concentrations. Please note that each panel has a different color scale and are not necessarily symmetric around zero

Full size image

3.3 Impact of AB 32 on population-weighted concentrations

Population weighted concentrations are a useful metric to evaluate the public health implications of policies that produce mixed outcomes for air pollution concentrations. Population weighted concentrations account for the spatial variability of the pollutant relative to the population so that different emissions control strategies can be compared. The population-weighted average concentration of PM2.5 mass, EC, OC, sulfate, nitrate, and ammonium were calculated for the last 12 days of each 16 day simulation period. Results are summarized for the major air basins in California and for the state as a whole. California’s population is projected to grow from 37.3 million in 2010 (United States Census Bureau 2011) to 49.2 million in 2030 (California Department of Finance 2007). The current study accounted for the projected population growth but the year 2000 spatial pattern of population density was used to maintain consistency with the assumptions inherent in the emissions inventory. All January 2050 episode results are expressed as the percent change relative to BAU for each Implementation Level. The variability of the daily average PM2.5 concentration for the extreme pollution episode was used to calculate the 95 % confidence intervals bracketing the change in population-weighted concentrations using the student’s t-distribution test with 11 degrees of freedom. The variability of the daily average PM2.5 concentration for the 108 days simulated for 2052 is presented in Fig. 6 as a box and whisker plot.

Figure 5 illustrates that full implementation of AB 32 measures reduce population weighted PM2.5 mass and component concentrations in California by ≥6 % during a severe air pollution episode. Measures within defined Implementation Levels 2, 4, and 5 (electricity generation and natural gas, on-road and other mobile sources) each produce approximately 2 % reductions in population weighted PM2.5 concentrations with smaller savings achieved by Implementation levels 1 and 3 (industrial and agricultural sources). Each Implementation Level affects PM2.5 component concentrations differently. Statewide population-weighted EC concentrations are affected most strongly by Implementation Levels 4 and 5 which target transportation sources, while OC concentrations are most strongly affected by Implementation Levels 2 which target electricity generation and Implementation Level 4. Population-weighted sulfate concentrations are most strongly affected by Implementation Level 5 while ammonium nitrate concentrations are most strongly affected by Implementation Levels 2 and 4.

Fig. 5
figure 5

Population weighted average change in PM2.5 mass, EC, OC, sulfate (S(VI)), nitrate (N(V)), and ammonium (NH4) during the 12 day extreme stagnation episode

Full size image

Each air basin in California experiences a different response to AB 32 due to the variation of emission sources and activity levels for each region within the state. Generally speaking, the SoCAB experiences the greatest reduction in PM2.5 mass and component concentrations due to AB 32 emissions control measures (Fig. 5d), while the SJV experiences the smallest reductions (or even slight increases for PM2.5 sulfate) (Fig. 5c). The single strongest factor contributing to this increased PM2.5 concentration trend in the SJV is the shift in electrical generation from combusting traditional fossil fuel outside the SJV to combusting renewable dairy biogas within the SJV. This AB 32 measure produces a clear GHG benefit for the state of California because it decreases bio-methane emissions into the atmosphere, but it worsens air quality for residents of the SJV. The NOx produced by the dairy digester combustion systems contributes to increased ammonium nitrate concentrations in the SJV. Dairy biogas also contains trace amounts of sulfur (unlike natural gas) leading to increased SOx emissions and sulfate formation even with the addition of substantial controls. The urban centers in the SoCAB benefit strongly from vehicular emission reductions and shipping measures at the port of Los Angeles.

The annual-average results (see Fig. 6) are generally consistent with the extreme event results (see Fig. 5). Annual-average population-weighted PM2.5 concentrations decrease by statistically significant amounts (p < 0.05) in each basin due to the adoption of AB 32. The statewide reduction in 24-h average PM2.5 exposure has a median value of ~5–6 % with a maximum as large as 10 % and a minimum as small as 3 % depending on the exact meteorological conditions on a given day during the annual average period. PM2.5 sulfate concentrations for the SJV experienced the least reduction and occasional increases with a median value of −0.7 %. Nitrate and ammonium exhibited the greatest sensitivity to daily variations in meteorology, with changes to statewide population exposure ranging from ~5–38 % and ~3–27 %, respectively. The SoCAB experienced higher annual average population-weighted PM2.5 mass reductions of ~6 %, while SJV and Sacramento experienced reductions of ~3–4 % in response to AB 32.

Fig. 6
figure 6

Box and whisker plot of the minimum, 25th percentile, median, 75th percentile, and maximum change (%) in population weighted PM2.5 concentrations caused by AB 32 over 108 days representing the annual average in 2052

Full size image

3.4 AB 32 health benefits associated with reduced PM2.5 concentrations

The change in mortality (ΔM) due to reduced exposure to PM2.5 was calculated using Eq. (1):

$$ \Delta M = \sum\limits_{{i = 1}}^n {M{P_i}\left( {1 - \frac{1}{{{ \exp }\left( {\beta \,\Delta P{M_{{2.5,i}}}} \right)}}} \right)} $$
(1)

The change in mortality (ΔM) was determined using projected 2030 populations ages ≥30 years (P i ) (State of California Department of Finance 2007) for each 64 km2 grid cell i in California. The change in PM2.5 (ΔPM 2.5) for each cell under the BAU scenario versus AB 32 scenario (all AB 32 measures incorporated) was quantified using the results from the 108 simulated days that are representative of the annual average meteorology. The basecase value of the risk factor β was taken to be approximately 0.01 (Roman et al. 2008). The basecase mortality rate, M, was set to 0.006343 for urban grid cells (≥500 people per square mile), or 0.009911 for rural grid cells (<500 people per square mile) (United States Department of Health and Human Services (US DHHS), Centers for Disease Control and Prevention (CDC), National Center for Health Statistics (NCHS) 2012). The uncertainty of the change in mortality predicted by Eq. 1 was calculated by using β of approximately 0.006 (Pope et al. 2002) and 0.012 (Laden et al. 2006) for the lower and upper uncertainty bounds. Further uncertainty in mortality estimates were quantified using the alternative method described by Jacobson (2010). This alternative approach used the same mortality rate for urban and rural locations (M = 0.008097) and a range of risk factors (βlow = 0.001, βmedium = 0.004, βhigh = 0.008) when PM2.5 ≥ 8 μg m−3. The value of β was divided by 4 when PM2.5 < 8 μg m−3. The change in mortality for all methods was expressed in monetary terms using the Value of a Statistical Life (VSL) approach (Viscusi and Aldy 2003).

Table 3 summarizes the results of health impact calculations. Equation (1) predicted that 880 premature deaths per year would be avoided in 2030 due to reduced PM2.5 concentrations associated with the adoption of AB 32. The uncertainty range for this basecase estimate was 560–1,100 deaths per year. The annual average mortality from business-as-usual PM2.5 exposure was estimated to be ~14,000 (9,100, 18,000) deaths per year in 2030. Hence, a reduction of ~880 (560, 1,100) deaths per year represents a 6.2 % reduction in the annual mortality caused by PM2.5 in California. Assuming a VSL of $6.2 million, the estimated total monetary value of this AB 32 co-benefit was predicted to be $5.4B/yr ($3.5B/yr–$7.0B/yr). The alternative method for mortality estimation predicted an 11.5 % reduction in mortality due to the adoption of AB32 with an uncertainty range for associated costs that spanned the basecase estimate.

Table 3 Mortality incidence and cost valuation associated with AB 32 due to changes in PM2.5. Method 1 incorporates Eq. 1, with urban and rural mortality rates, and β coefficients of 0.009355, 0.006, and 0.012 for the average, and lower and upper uncertainty range shown in parenthesis. Method 2 incorporates no specific urban or rural mortality rate, but uses a different low, medium, and high β coefficient for 0–8 μg/m3 or greater than 8 μg/m3. Intermediate calculations use full precision while final results are rounded to two significant figures for presentation
Full size table

Fann et al. have previously estimated the health benefit of emissions control measures for the SJV in California (Fann et al. 2009). The emissions changes summarized in Table 2 of the current study combined with methodology described by Fann (accounting for the dose–response relationship summarized by Eq. 1) produced an estimated health benefit of approximately $44,500 per short ton (~$49 k per metric ton) of emissions reduction (averaged across all pollutants). The health benefits summarized in Table 3 ($5.44B/yr) combined with the emissions reductions summarized in Table 2 (~150,000 metric tons per year of PM, NOx, SOx, and NH3) yielded a health benefit of $35 k ($23 k–$45 k) per metric ton of emissions reduction (averaged across all pollutants). The methods employed in the current study differ from the methods employed by Fann et al. due to the increased accuracy of the treatment for climate/meteorology, emissions, and air pollutant concentrations.

4 Conclusions

California’s Global Warming Solutions Act of 2006 (AB 32) is predicted to reduce population-weighted statewide PM2.5 concentrations by ~6 %. Each California air basin experiences a unique response to AB 32 control measures due to the heterogeneous pattern of emission sources, previous control measures, and the new strategies adopted to reduce GHG emissions. The SoCAB experiences the greatest reductions in PM2.5 concentrations due to transportation measures while the SJV experiences the smallest reductions or even slight increases in PM2.5 concentrations mainly due to the increased use of dairy biogas for electricity generation in this region. GHG measures targeting transportation sources provide the largest co-benefits for PM2.5 reduction in California. Transportation sources such as ships that previously were not subject to controls provide a strong opportunity to improve regional air quality while at the same time reducing GHG emissions. Energy efficiency measures are also beneficial to air quality, especially for cities with large point sources. Some renewable electricity generation strategies may increase air pollution concentrations even though they reduce statewide GHG emissions. In the present study, dairy methane capture and electricity generation measures in the SJV may increase population weighted concentrations of PM2.5.

The ~6 % reduction in PM2.5 exposure associated with AB 32 in the current study is predicted to reduce air pollution mortality in California by 6.2 %, avoiding 880 (560–1,120) premature deaths per year for the conditions in 2030. The monetary benefit from this avoided mortality is estimated at $5.4B/yr ($1B/yr–$7.5B/yr) with with a weighted average benefit per tonne of $35 k/tonne ($23 k/tonne–$45 k/tonne) of PM, NOx, SOx, and NH3 emissions reduction.

The measures in AB 32 are tailored to the opportunities in California given the high degree of pre-existing emissions controls as well as the use of clean fuels in the state. It is very likely that different states within the US or other nations around the globe that have less stringent controls for criteria pollutant emissions or very different energy portfolios (e.g. primarily coal) may have a much larger potential criteria pollutant co-benefits from the adoption of GHG emissions controls. Furthermore, alternative mitigation policies such as land use change may be more optimal and cost-effective in achieving integrated AQ and GHG goals for other regions in the world. The results of the current study should not be used to project the co-benefits of GHG mitigation outside of California.