Geographic pollution mapping of power plant emissions to inform ex-ante environmental justice analyses

Abstract

Federal and many state agencies are required to perform environmental justice analyses of their policies prior to implementing them to prevent undue impacts on low income and minority populations. However, little academic attention has been paid to the quality of these ex-ante environmental justice analyses. This investigation evaluates the methods used to perform environmental justice analyses during siting and permitting processes. The study uses both the California Energy Commission guidelines for environmental justice analyses and a method that geographically maps air pollution to perform ex-ante environmental justice analyses of three power plants. The objective is to see if results from using these two analysis methods differ substantially. Findings indicate that the mapping technique employed represents a substantial improvement over defining the impacted population using proximity methods because it accounts for the geographical distribution of the hazard. Furthermore, using multiple comparison benchmarks to determine whether the impacted population constitutes an environmental justice population improves upon existing methods by accounting for the spatial distribution of minority and low income populations and for the possibility that there is a relatively high or relatively low percentage of low income and minority persons in both the impacted and comparison regions.

Acknowledgements

The author would like to thank the following people and institutions for their support: The Energy Foundation, the National Science Foundation, Daniel Kammen, William Nazaroff, Alex Farrell, Garvin Heath, Abby Hoats and Shannon Coulter-Burke. The author would also like to thank four anonymous reviewers for their feedback and advice.

Notes

1 This study, like many environmental justice analyses, uses census data to determine the impacted population. Census data indicate population based on their place of residence. As people are often not at home throughout the day because they commute to work, school and other locations, the use of census data to determine the population within a given area may not represent the true impacted population.

2 Instead of using maximum concentrations, it might make sense to determine the population that experiences a certain percentage of pollution concentration that is permissible by regulatory standards. Using the regulatory standard as a benchmark would be more instructive, in that it would allow for a comparison across plants of the demographics of the population that experiences the same concentration of pollution. However, as the concentration of pollution as a result of any one facility is usually much less than the ambient air quality regulatory standard, the percentage of the standard used would have to be quite low in order for any segment of the population to be included. Furthermore, as the purpose of this study is to determine if the impact of a specific facility disproportionately impacts one segment of the population, it makes sense to determine the cut-off relative to the magnitude of impact of that specific facility itself rather than relative to a pre-determined benchmark.

3 According to both the CEC and the President's Council on Environmental Quality, a minority is defined as an individual who is a member of the following population groups: American Indian or Alaskan Native; Asian or Pacific Islander; Black not of Hispanic origin; or Hispanic. Low-income populations are identified based on annual statistical poverty thresholds from the Census Bureau's Current Population Reports, Series P-60 on Income and Poverty.

4 It should be noted that this paper calculates the low-income population, whereas the CEC Environmental Performance Report calculates the population ‘in poverty’, i.e. that portion of the population falling below the poverty line.

5 There are inconsistencies within and among the CEC documents that report the results of environmental justice analyses for each plant. Although the CEC 2003 Environmental Performance Report indicates that the population within 6 miles of the El Segundo plant is 70% minority, the final staff assessment of the facility, which was published in September 2002, cites a different figure. At various points in the final staff assessment, the population surrounding the plant is reported as 60.6% minority, 57.6% minority and between 44.9% and 57.6% minority. The figures cited for the percentage of the population that is low income are less divergent, ranging from 10.11% to 10.85%. There are also discrepancies between the 2003 Environmental Performance report and the CEC staff assessment of the Potrero power plant. The final staff assessment of the Potrero power plant reports the population within a 6-mile radius of the plant as 57.6% minority and 12.3% low income.

6 NO emissions quickly undergo complex transformations as NO pollution interacts with other elements in the atmosphere. Thus in a simplified model such as this, it is more realistic to model NO_x species than NO.

7 The form of the Gaussian plume model used to calculate the dispersion and concentration of emissions from power plants included modifications to account for the effects of 20 reflections from the ground and from the mixing height boundary. The formula used is

for which C is the ground-level concentration (g m⁻³), E is the steady state rate of emission from the source (g s⁻¹), U is the wind speed (m s⁻¹), σ _y and σ _z are the dispersion coefficients in the downwind and crosswind directions (m) as modeled based on the Pasquill – Gifford parameters (Davidson, 1990), H _E is the effective stack height of the emission source, M is the mixing height and n is an index for the number of reflections (Heath et al., 2003). The form of the model used for this analysis is applicable to continuous point sources emitting primary non-reactive pollutants (Seinfeld & Pandis, 1998). The Gaussian plume model is intended to provide a close approximation to the expected pollution concentrations in the vicinity of the power plants. However, these approximations are not perfect. This is the case for all models, as simplifying assumptions must be made because of the constrained availability of data, the inability of a model to capture all natural variations and limitations on the scientific understanding of complex transport and transformation processes. This model was evaluated at 14 641 locations uniformly distributed throughout a square centered on each plant. The points were spaced every 0.5 km extending to 18.6 miles in each of the four cardinal directions.

8 The TMY2 data set includes hourly meteorological conditions for a 1-year period for each of 239 National Weather Service Stations located throughout the USA. Data from the Los Angeles station were used for analysis of the El Segundo power plant, data from the San Francisco station were used for analysis of the Potrero power plant and data from the Sacramento station were used for analysis of the Pittsburg power plant.

9 Adjustments had to be made to the Gaussian plume model to account for various meteorological conditions such as calm hours and mixing heights lower than the effective stack height. The omission of calm hours will cause the model to underestimate the concentration of pollution in close vicinity to the stack. This is especially important for the Pittsburg plant, as approximately 14% of the hours are calm.

10 These are only available for the Oakland, California meteorological measuring station. Thus the Oakland data were used for all locations.

11 Because a given plant is not in operation at full capacity every day of the year, the daily average emissions value was used. Although this is a practical estimate for baseload plants, for peak load generating facilities, this assumption will create some degree of error, as those plants are frequently ramping up and ramping down. Emission rates are not constant while power plant output is varying. In California, NO_x emissions are monitored on a constant basis, thus the total emissions reported for NO_x are all inclusive. However, for other pollutants the reported total annual emissions rate does not necessarily include emissions during ramp up and down. Thus, if this method were used for other pollutants, total annual emissions would be higher than reported. Furthermore, as power plants all ramp up to generate more electricity, the temperature of gaseous emissions will increase, raising the effective stack height. During those periods, the effective stack height will be lower than the effective stack height used in this analysis, leading to higher ground-level concentrations of pollution closer to the plant. Additionally, the Gaussian plume model used in this analysis does not account for the possibility of down washing, which, if it occurs, will increase the concentration of pollution closer to the plant.