AIR QUALITY AND ENVIRONMENTAL JUSTICE: AN ANALYSIS OF AIR POLLUTION AND WEALTH DISTRUBUTION Gordon Young George Strebel 2.28.05 The greater Houston metropolitan area has come under intense scrutiny in the past decade due to its heavy amount of dangerous air pollution. During the efforts to clean Houston’s air, the concern of environmental justice has been raised. According to the Environmental Protection Agency (EPA), environmental justice is defined as “the fair treatment and meaningful involvement of all people regardless of race, color, national origin, or income with respect to the development, implementation, and enforcement of environmental laws, regulations, and policies” (http://www.epa.gov/oswer/ej/index.html /). The purpose of this study is to determine how particulate air pollutants distributed in correlation with median household income and industrial areas across the greater Houston metropolitan area. The Houston area is a prime candidate for this study due to its high levels of air pollution, its great economic diversity and it’s proximity to major industrial areas. 1 In terms of relevance as a useful study, the EPA has consistently ranked Houston among the cities with the worst air quality records across the United States and because of this has designated the eight-county area containing and surrounding Houston a non – attainment area. This means that the eight-county area does not conform with the current environmental laws pertaining to air pollution standards. This rating has alarmed residents and local environmental authorities, leading to several federal and local measures aimed at cleaning Houston’s air. The concern is that Houston’s underprivileged population located near large industrial areas may be at a higher risk for breathing unsafe air, contaminated with higher levels of ozone and the substances that produce ozone, known as volatile organic compounds (VOC). The term environmental justice seems to be called upon as a catch all phrase to rationalize tough issues that face low income urban areas today. In some instances what appears to be a clear issue of environmental justice might be clear at all. The argument that low income areas have a disproportionately elevated exposure rate to more dangerous levels of air pollution would seem to be true, but on closer examination of the causes of air pollution we might find the argument to be false. To approach a project of this magnitude requires an understanding that the issue not only affects the residents of our study’s counties that make up the Texas Commission on Environmental Quality –(TCEQ) region 12 non-attainment area, but is also a concern impacting the lives of every American. We recognized that the use of GIS has great potential to help us understand the spatial relationship between pollution and wealth distribution, and therefore expected to find existing GIS studies on the topic from which 2 to draw information in order to set the criteria for our own study. What was found was that even though GIS is a powerful tool for spatial analysis and for producing maps that are extremely effective in visually displaying the disproportionate spatial distribution of income and pollution causing facilities, many of these studies have turned out to be misleading and inaccurate, and their findings had often been contradicted by other spatial analysis (Maantay, 161). This understanding led to an analysis requiring two different GIS techniques to approach the answer to our question. Exploring the research question through two different GIS techniques has allowed further insight into not only the relationship between point source pollution, like industrial areas, and the correlation with low income areas, but also how air pollution moves independently of cultural barriers like county lines and census blocks. The geostatistical function of kriging was decided upon to interpolate monthly average ozone data gathered from monitoring stations located within the region 12 non-attainment area, which will be further explained. This allowed us to see where the highest and most dangerous concentrations levels of ozone occurred on any given day, allowing us to overlay monthly kriging averages over the census blocks that meet the criteria for lowest income areas to see if the research question is answered. The second approach used the following GIS geoprocessing functions: buffer, merge, intersect, select by attribute, and select by location, to find the census blocks with the lowest median household income that are located .25 miles from major roads and .50 miles from industrial areas. With time constraints in consideration, the most effective way to approach the question was to split the work up, each person responsible for one of the GIS analysis’ 3 that we would be able to overlay for final analysis. George used the geoprocessing functions to locate the lowest median household income census blocks that were affected by point source VOC’s located .25 miles from major roads and .50 miles from industrial areas, while Gordon used the geostatistics function of kriging to interpolate the highest ozone levels readings from the TCEQ’s monitoring stations located within the region 12 non-attainment area. The first step to building the GIS was to comprehend the many factors involved in air pollution. We needed to not only understand the health effects of air pollution on the population, but how it is formed and distributed over a large area. We took this into consideration when compiling our literature review. Several diversified publications were used in obtaining the necessary information, and many of them were used in determining criteria for the actual analysis. Our approach to the question was to examine the locations of industrial area’s total VOC emissions in relation to median household income of each census block. We could then take an average of the highest ozone levels in an eight-hour time frame during a four month period from June to September 2004 from the air quality monitoring stations located around the TCEQ Region 12 Non-Attainment area. By using the geostatistics feature kriging to interpolate the ozone averages over the eight counties in the region 12 area we would then be able to detect where the highest concentrations levels of ozone were located. Using the geoprocessing functions: buffer, merge, intersect, select by attribute, select by location to find the census blocks with the lowest median household income, that are located .25 miles from major roads and .50 miles 4 from industrial areas. To answer our research question will be able to overlay the monthly kriging averages on the census blocks that meet the criteria for lowest income areas to see if there is a correlation between pollution and the distribution of wealth. Data Needed and Data Sources: Census Block Data The US Census 2000 Tiger Files http://arcdata.esri.com/data/tiger2000/tiger_download.cfm Road shapefile and County outline shapefile and Hydrography shapefile ESRI Website http://arcdata.esri.com/data/tiger2000/tiger_download.cfm Median Household Income US Census SF 3 Files American Fact Finder http://factfinder.census.gov/servlet/DatasetMainPageServlet?_program=DEC& lang=en &_ts= Air Monitoring Stations Locations TCEQ Website http://www.tceq.state.tx.us/ Daily Ozone readings TCEQ Website http://www.tceq.state.tx.us/ Industrial Facilities Locations EPA Website http://www.epa.gov/ We expected the data we used to be accurate because it was acquired from established reliable sources. All GIS data was acquired from Environmental Systems Research Institue (ESRI) and the TCEQ. The non-spatial data was also collected from reliable sources such as the United States Census Bureau and the EPA. The scope of the 5 study ranges from analysis on a census block level up to a regional level, so precision was also an important consideration. The results of all calculations were double checked for accuracy and precision, and all data was reviewed before submission. At this point the data had been collected defined and projected in NAD 83, UTM Zone 14 and inputted into ArcMap. We now needed to set criteria for the following inputs: median household income, industrial locations, major roads and highways, ozone levels and ozone monitoring stations so that we could refine that data. We decided to use census block data so that we could narrow the focus of our study area. Average household income per census block was derived by using the US Census SF 3 average household income census2000 files to create a dbf file in excel that was then joined to the census blocks in ArcMap. We attained “persons per household” by adding the up the average person per household for a family of all eight counties and dividing by the total number of counties to get household population of three. We were then able to manually aggregate income into six levels in ArcMap. Two low income values were derived using the 2004 Health and Human Services Poverty Guidelines of $15,670 for a family of three and U.S. Department of Labor Employment & Training Administration lower living standard income level of $28,050 for a family of three living in southern metro areas. The six levels of census block income were then symbolized by color ramp levels and displayed in ArcMap. 6 Ozone, or O3, is a gas normally formed in the upper atmosphere of the Earth when normal oxygen molecules are broken up and reform as O3, or a molecule containing 3 atoms of oxygen. Ozone is normally a helpful gas that forms part of the upper atmosphere and blocks harmful solar radiation, but ozone can form at the ground level as well. Ozone forms at ground level when oxygen molecules and sunlight interact with volatile organic compounds or VOC’s. Most VOC’s originate from man-made sources and are the result of industrial processes. The Houston area is heavily industrial and according to the Galveston Houston Association for Smog Prevention (GHASP) it is “clear that industrial activity is the ultimate cause of most hazardous air pollution in the Houston region” (3). In fact the association also states that “industrial sources account for 75 percent of VOC’s” (3). The problem of environmental justice is raised when these 7 industrial areas are examined in context of income. The hypothesis of this project is concerned with lower income areas and their population’s exposure to ozone, therefore the health risks of ozone must also be examined. According to the TCEQ ozone can cause various breathing problems among sensitive populations. These problems include minor issues such as coughing or wheezing all the way up to serious problems such as headaches and lung damage. Therefore, sensitive populations near industrial areas are at a risk for these conditions. It is the intention of this project to establish whether ozone levels are more highly concentrated around low income areas instead of medium to high income areas. In order to accomplish this task the geostatisitic analyst extension of ArcMap will be utilized. The Houston area has a very diverse social makeup, ranging from extremely high income to very low. Geostatistics, specifically kriging, will enable us to predict the levels of ozone 8 for a given period across the entire study area, displaying the levels in comparison with different income strata, and with large concentrations of industry, allowing in-depth analysis. Ozone levels of above eighty parts per billion are considered harmful to sensitive populations; therefore this has been established as the threshold of the levels to be monitored in the study. In the aforementioned task, as with any GIS analysis, acquisition and preparation of input data was of the utmost importance. Base map data, including county boundaries and hydrography shapefiles, was acquired from ESRI. The most important dataset of this portion of the project was the air monitoring stations shapefile, which was acquired from the TCEQ. This was a point –feature shapefile for each monitoring station statewide, therefore the stations in the desired counties had to be selected-by-attribute from the entire state ultimately resulting in a shapefile representing 37 active monitoring stations in the TCEQ region 12 area. This shapefile included an attribute table with non-spatial data such as the identification number of each monitoring station, and which types of pollutants it monitors, however it did not include specific measurements taken at each station. To acquire this information it was necessary to access the data on the TNRSS website detailing the daily values of ozone monitored at each station in text format. TNRSS also records days on which ozone reached unsafe levels in the Houston area. It is these days during June, July, August, and September of 2004 that were selected for those to be studied in geostatistical analysis. The peak level of ozone recorded at each monitoring station for each unsafe day was recorded in a Microsoft Excel workbook and used to compile a monthly average reading for each station. The daily values and each 9 monthly average were manually added to the monitoring stations’ attribute table using the ArcMap editor extension. The data was now prepared for geostatistical analysis. The goal of geostatistical analysis is roughly explained by ESRI to be to “take a set of data points and create a model that tells you interesting and useful things about how the data is distributed across an entire area” (ESRI). In order to gain a rudimentary knowledge of geostatistics, the study team enrolled in the ESRI Virtual Campus Basics of Geostatistical Analysis tutorial, available free of charge on the ESRI website. The tutorial discusses a five-step approach to using geostatistics, including data preparation, statistically exploring the data, choosing an analysis method, and testing and comparing the data. The tutorial introduced surface interpolation, which is a very important component in geostatistical analysis. Interpolation is mathematically estimating the occurrence of an event between known events, or in the case of our project, estimating unknown ozone levels across the study area by analyzing the known levels at the monitoring stations. The type of surface interpolation chosen for the task of predicting the ozone values was the potentially very accurate kriging method. Kriging allows spatial data to be interpolated to a high degree of reliability because it mathematically accounts for statistical trends within a dataset, and attempts to account for any possible spatial variability within that very same trend. This spatial variability is known as spatially autocorrelated random error. Autocorrelation takes into account that if there are values above the average in a given area, there also tend to be values below the average in close proximity to that area, therefore allowing the trend to be more accurately mapped (ESRI 10 virtual campus). Each monthly average ozone reading was modeled individually using the kriging function of the geostastical analysis toolbar with striking results. The data was interpolated across the entire study area for each monthly average, resulting four very different surface interpolations. This display indicates that ozone, despite having very specific industrial sources, can be concentrated in various locations throughout the study area. After completing a kriging layer for each of the four monthly averages, and setting the extent of the layer to the entire study area, these layers were individually overlayed over the previously prepared low level income layer. These four individual analysis show very different results, and have been displayed in four unique map outputs. At this point it was also decided to complete kriging maps for each day of the study period, in order to complete an analysis of interpolated daily ozone levels across the entire study area. This analysis has resulted in 19 unique interpolations of data, all of which are very striking, and are displayed as a chronological animation. On some days ozone appears to be clustered around its major industrial sources, many of which are low income areas, but on other days the high ozone levels are dispersed across the study area with seemingly no relationship to the sources. Research has indicated that several environmental factors, including wind, rain, and temperature can effect the dispersion of ozone, therefore we have attributed this seemingly random set of interpolations to be at least partially the result of these outside factors. 11 12 13 The two pronged analysis of this project helps to shed further light upon the relationship between a locations proximity to industrial locations and the actual amount of ozone pollution that location receives. It is not necessarily accurate to assume that since families with lower income live in close proximity to industrial areas that they will automatically experience a higher level of ozone; however what must be taken into account are the influence of weather and other outside factors not studied here. A very high correlation near perfect does exist between median household income and the amount of industrial facilities within individual census blocks, but this appears to have little effect upon ozone levels for that particular area. One very important factor to consider about the outcome of this study is that for all four months, the center of highest pollution is actually located over middle class areas, a result which may alarm many families in these income strata. For industrial locations we chose to use Volatile Organic Compounds (VOC) that contributes to troposphere ozone formation for our criteria. Select by location was used to identify the industrial sites that were located in each census block polygon. By summing the total VOC output for industrial sites located within a census block polygon we were able to weight each polygon with a value of tons per year (TPY) of VOC’s emitted. The EPA article Approval and Promulgation of Implementation Plans; Texas; Permitting of Nonattainment Areas was used to set criteria of 25 TPY for aggregating the data in ArcMap. A .5 mile buffer was then created from each identified industrial location to identify proximity to low income census blocks. 14 Major roads and highways were found by using Select by Attribute to query out all major roads and highways from our roads shapefile and created a new layer in ArcMap. We used the criteria of a of .25 mile buffer from major roads and highways that was obtained from a previous GIS study done by the California Environmental Agency. We then merged this shapefile with the .50 mile highest VOC polygon buffer to create a new shapefile that was intersected with the lowest income census blocks to create a new shapefile of called meets criteria. In order to display the areas with the highest concentrations of ozone, kriging contour lines that displayed areas with over 80 PPB of ozone were selected by attribute. ArcEditor and ArcCatalog were then used to a new polygon shapefile displaying the 80 PPB contour lines. This new shapefile was then clipped to the aggregated median 15 household income census blocks, displaying the census blocks that fall within the contour line. The average income of these census blocks was then used to determine the average incomes of the areas with the highest concentration of ozone. These census blocks actually showed an income range denoting medium to upper middle class. Statistical analysis shows no valid relationship between median household income and VOC emissions per census block (correlation of -0.13644). Further statistical analysis shows that the ratio of industrial locations per census block increases as the median household income decreases. Analysis has concluded that there is a direct correlation between household income and proximity to industrial facilities. There appears to be no direct correlation between actual pollutants (VOC’s and ozone) and household income. GIS has been very valuable in conducting this analysis. Our analysis has taken steps to show that the term environmental justice may be used misleadingly when economic factors dictates a people’s interaction with the environment. Further analysis upon this subject matter should take into account data concerning health issues, school attendance areas, and weather patterns in order to increase accuracy and precision. The process of working together as a group has been an opportunity to learn and develop real world skills and has been very useful in completing such an extensive project. 16 Statistical Charts: Relationship Between Income and VOC Emissions R 2 = 0.0186 3,500 3 ,000 VOC 2,500 2 ,000 1,500 1,000 500 0 0 20000 40000 60000 80000 100000 120000 Household Income Correlation -0.13644 Relationship Between Income and Industrial Sites Ratio of industial Sites Per Census Block R2 = 0.9049 1 0.8 0.6 0.4 0.2 0 0 20000 40000 Income Levels Correlation -0.95124 17 60000 80000 Literature Review: Ahrens, Donald C. Meteorology Today: An Introduction to Weather, Climate, and The Environment. MN: West Publishing Company. “Approval and Promulgation of Implementation Plans; Texas; Permitting of NonAttainment Areas.” Environmental Protection Agency.
. “Clearing the Skies of Texas.” Mechanical Engineering, Sep. 2004, Vol. 126, Issue 9 Darby S. Lisa. “Impact of Surface Wind Patterns on Surface Ozone in Houston, Texas as Revealed by Cluster Analysis.” NOAA/Environmental Technology Laboratory. TexAQS 2000
. DeMers, Michael N. Fundamentals of Geographic Information Systems. John Wiley & Sons, 2004. “Final Approval of the Houston/Galveston Ozone Attainment Demonstration.” U.S. EPA Region-6. 15 Oct., 2001
. “Introduction to ArcGIS 8 Geostatistic Analyst” ESRI Virtual Campus 10 November 2004
. J. Raloff. “U.S. Smog Limit Permits Subtle Lung Damage.” Science News Vol. 13 May 2000, Vol. 157, No 20. Maantay, Juliana. “Mapping Environmental Injustices: Pitfalls and potential of Geographic Information Systems in Assessing Environmental Health and Equity.” Environmental Health Perspectives. 110(suppl 2):161-171 (2002) “OEHHA Traffic-Related Air Pollutant Studies Overview.” ASTHO Health and Environment Electronic Seminar. 24 Oct. 2002 http://www.astho.org/pubs/October%202002%20HEES.pdf Price, Maribeth. Mastering ArcGIS. McGraw-Hill, 2001. 18 “Sensitivity of Regional Ozone Concentrations to Temporal Distribution of Emissions.”Zhining Tao, Susan M Larson, Allen Williams, Michael Caughey, DonaldJ. Wuebbles. Journal of Geophysical Research. Vol. 108, No. D14, 4404. 17 July 2003. “Where Does Houston’s Smog Come From? The Sources of Houston’s Air Pollution and What’s Needed to Clean the Air?” Galveston-Houston Association for SmogPrevention. Oct. 2003
. “Definition of Environmental Justice” 21 Oct 2004, http://www.epa.gov/compliance/environmentaljustice/). 19
Link or Click Back
Here will be a configuration form