Category: Air Quality Issues

In the last several days I have been drafting a review of  the PEAK Coalition report entitled: “Dirty Energy, Big Money” and today I was working on the air quality health impacts section.  I also noticed today that the usual suspects are claiming links between air pollution and Covid-19 susceptibility. In this post I will explain how I could be convinced that the reports underlying presumption that inhalable particulates have dire health impacts is correct.

I am a retired electric utility meteorologist with nearly 40 years-experience analyzing the effects of air quality and meteorology on electric operations.  I have been reviewed health impact claims throughout my career.  This background served me well preparing this post.  The opinions expressed in this post do not reflect the position of any of my previous employers or any other company I have been associated with, these comments are mine alone.

Background

Health impacts associated with inhalable particulates, also known as PM2.5 because it refers to airborne particles with a diameter of 2.5 micrometers or less, turn out to be the primary rationale for all the recent EPA air quality emission reductions cost-benefit analyses.  For example, EPA’s air toxics emission limits were cost effective not because of direct impacts of mercury and other heavy metals but because the control systems for those pollutants would have decreased PM2.5 concentrations and led to alleged health improvements.

Steve Milloy’s Scare Pollution: Why and How to Fix the EPA explains the problems with those health impact claims.  Milloy points out that no one has proven a biological explanation why the inhaled particles will cause fatal inflammation.  The alleged relationship is based on epidemiological statistical evaluation of air quality and health impact data.  The basic problem is that there are many confounding factors known to cause the observed health impacts and trying to tease air quality impacts out of the mix is difficult to prove.

It gets worse.  The studies that are the basis for the alleged air quality health impacts were at relatively high ambient concentrations.  Make no mistake that air pollution can be a very bad thing but the levels of pollution in the United States that clearly caused health impacts occurred many years ago and included a mix of pollutants not found anywhere in this country today.  It gets worse because the dose-health impact relationship is being extrapolated using the linear no-threshold model which has been used to estimate the dose response for radiation health impacts.  The concept is that there is no threshold below which there is no effect.  However, in my opinion and others, extrapolating measurements and responses at high levels down to levels near the level of detection is an unwarranted presumption.  Nonetheless, advocates for ever lower air quality improvements routinely claim health impacts behave the same way.

Public Health Impacts

The primary public health reference in the PEAK Coalition report I am reviewing was the New York City Department of Health and Mental Hygiene’s (DOHMH) Air Pollution and the Health of New Yorkers report.  The PEAK coalition description of air quality public health impacts quotes the conclusion from the DOHMOH report: “Each year, PM2.5 pollution in [New York City] causes more than 3,000 deaths, 2,000 hospital admissions for lung and heart conditions, and approximately 6,000 emergency department visits for asthma in children and adults.”  These conclusions are for average air pollution levels in New York City as a whole over the period 2005-2007.

The DOHMOH report specified four scenarios for comparisons (DOHMOH Figure 4) and calculated health events that it attributed to citywide PM2.5 (DOHMOH Table 5).  Based on their results the report notes that:

Even a feasible, modest reduction (10%) in PM2.5 concentrations could prevent more than 300 premature deaths, 200 hospital admissions and 600 emergency department visits. Achieving the PlaNYC goal of “cleanest air of any big city” would result in even more substantial public health benefits.

It is important to note how air quality has improved since the time of this analysis.  The NYS DEC air quality monitoring system has operated a PM2.5 monitor at the Botanical Garden in New York city since 1999 so I compared the data from that site for the same period as this analysis relative to the most recent data available (Data from Figure 4. Baseline annual average PM2.5 levels in New York City). The Botanical Garden site had an annual average PM2.5 level of 13 µg/m³ for the same period as the report’s 13.9 µg/m³ “current conditions” city-wide average (my estimate based on their graph).  The important thing to note is that the latest available average (2016-2018) for a comparable three-year average at the Botanical Garden is 8.1 µg/m³ which represents a 38% decrease.  That is substantially lower than the PlaNYC goal of “cleanest air of any big city” scenario at an estimated city-wide average of 10.9 µg/m³.

Note that in DOHMOH Table 5 the annual health events for the 10% reduction and “cleanest” city scenarios are shown as changes not as the total number of events listed for the current levels scenario.  My modified table (Modified Table 5. Annual health events attributable to citywide PM2 5 level) converts those estimates to totals so that the numbers are directly comparable.  I excluded the confidence interval information because I don’t know how to convert them in this instance.

I confirmed that the DOHMOH analysis used a linear no-threshold health impact analysis and used their relationship to estimate the effect of the observed air quality reduction. I tested the linear hypothesis by scaling the “current level” scenario number of events to the proportion of the PM 2.5 concentrations (the last row in the table) for the “current level” and the other two scenarios.  My estimated health impacts were all within 1% which proves that the DOHMOH analysis relied on a linear no-threshold approach.  As a result, that means that I could estimate the health impact improvements due to the observed reductions in PM2.5 as shown in the last three columns in the modified table.  I estimate that the observed reduction in PM2.5 concentrations prevented nearly 1,300 premature deaths, 800 hospital admissions and 2400 emergency department visits.

Conclusion

In order to convince me that the PM2.5 health impacts claimed by MOHDOH and many others are correct I need to see confirmation with observed data.  The DOHMOH report claims that in 2005-2007 that PM2.5 concentrations led to, for example, 3,200 premature mortality events.  I have no idea how that number compares to observed values for this parameter or the others included.  I estimate that for the observed reductions in measured PM2.5 the number of premature mortality events would be reduced 1,296 events down to 1,904 events.

The first question for the health experts is whether the change from 2005-2007 to 2016-2018 of 1,296 events could be observed against natural variations or is that number within the normally expected variation.  If not then my hope for verification is not possible but more importantly it also means that the gloom and doom stories of significant health impacts are base on nothing more than insignificant statistical noise that is not really observable.  If those data are greater than expected natural variation, then it would be possible to document improvements in these alleged health impacts due to the 38% decrease in PM2.5.  If that is the case, then I stand corrected.

Here is the thing though.  The percentage of people with asthma in the United States from 2001 to 2018 is not showing a decrease at the same time ambient levels of all air pollutants are going down substantially.  While correlation does not necessarily mean causation, no correlation with a purported cause indicates a bet on a dead horse.  Therefore, I am not holding my breath that the data will show the purported benefits.

Doctors have warned that air pollution increases the risks of dying from COVID-19.  This post compares air pollution levels in Italy with the highest European COV-19 mortality with New York State where the largest number of COVID-19 cases have occurred in the US.  I am sure the general impression is that New York City air quality is so bad that, if this relationship is true, that similar mortality rates are inevitable but I will show that is not the case.

Background

In an interview Dr. Sucharit Bhakdi, a German microbiology specialist, explains that the reason for the apparent global different mortality rates for COVID-19 may be because of different local situations.  He points out Northern Italy and China both have air pollution problems as well as high mortality rates.  Consistent with others he suggests that the lungs of individuals in those areas have been chronically injured over decades and this influences the mortality rates.

I compared Italian data with New York data that I had on hand.  Italian air quality data are available at the European Environment Agency Italy air pollution country fact sheet website.  Air quality data from New York’s monitoring network are available at the New York State Department of Environmental Conservation air quality monitoring website, in their annual reports, and there is an Environmental Protection Agency website that also has the data.  Particulate Matter 2.5 (PM2.5 measures particles that are generally 2.5 micrometers or smaller), Particulate 10 (PM10 measures particles that are generally 10 micrometers or smaller), nitrogen dioxide (NO2) and ozone are measured in both jurisdictions.  Unfortunately. the readily available summary data for PM10 and ozone data are not directly comparable because the air quality standards are different.  New York has only a few monitors for NO2.   The pollutant of most concern for health impacts is PM 2.5 because these particles are small enough to get inhaled into the lung.  There are seven monitoring sites within New York State with directly comparable PM2.5 data and that have reported data since 1999.  The only issue is that those data are not readily available so I have manually extracted these data from annual reports over the years.

Comparison

The Comparison of Italian Average and Selected New York State Air Monitors Annual Average Air Quality Data table includes Italian and New York PM2.5 annual average measurements that are encouraging.  Italian data are downloadable in three categories: Traffic which represents the highest expected levels, suburban/urban background which I assume represents ambient conditions for most people, and rural background which should represent atmospheric concentrations without Italian impacts.  The New York data are listed for the Botanical Garden station in New York City, three Upstate cities, a monitoring location on Long Island that is downwind of New York City and two rural background stations.

The Italian traffic impacts site had a PM2.5 annual average of 18.3 µg/m³, the suburban/urban background was 17.2 µg/m³and the rural background was 16.1 µg/m³.  The good news is that the monitoring location with the highest observed annual PM2.5 concentrations was at the Botanical Garden monitor in New York City was 8.0 µg/m³ which was less than half the rural Italian average background.  Note that in the most recent year there were a total of 23 PM2.5 air monitors operating in New York City.  In 2018 the Botanical Garden monitor had an annual average greater than or equal to all but four of the monitors.  The highest annual average was 10.4 µg/m³ still well under the Italian rural background.  The important thing to note is that all of New York PM2.5 annual averages are smaller than the lowest Italian traffic and suburban/urban background sites since 1999 and the rural background site averages since 2002.  There is no question that New York State air quality is substantively better than Italy for PM2.5.

I also list the nitrogen dioxide data.  New York NO2 air quality levels are only marginally better than Italy.   It is instructive to compare the two pollutants.  NO2 primarily gets in the air from the burning of fuel. NO2 forms from emissions from cars, trucks and buses, power plants, and off-road equipment.  Most small particles form in the atmosphere as a result of complex reactions of chemicals such as sulfur dioxide and nitrogen oxides, which are pollutants emitted from power plants, industries and automobiles.  I suspect the difference between New York and Italian air quality reflects the success of the EPA and New York State air pollution control efforts since 1990.  Since 1999 New York power plant sulfur dioxide emissions are down 99% and nitrogen oxides emissions are down 92%.  Coupled with concurrent reductions from industrial sources this has been a primary factor for the PM2.5 reductions.  I guess that Italian sources have not reduced their emissions as much.  On the other hand, New York continues to struggle with the ozone ambient air quality standards in part because nitrogen oxides emissions from the transportation sector have not come down nearly as much.  This would explain why the Italian and New York NO2 data don’t differ as much.

Conclusion

I think there is a general perception that New York City air quality is poor.  The fact is that while there are still some overall New York issues, the pollution levels have improved significantly.  The good news is that if the hypothesis that the COVID-19 mortality rate is related to chronic air pollution levels and that PM2.5 is a good surrogate for that pollution, then these data suggest that factor will not have as much of an effect in New York State in general and New York City either.  The PM2.5 concentrations are significantly lower than either Italy and, I would presume, China too.

When I first saw a graph from a New York Times article entitled “America’s air quality worsens, ending years of gains, study says” my first thought was that the reporter must have mis-read the analysis report. However, National Bureau of Economic Research working paper 26381 “Recent Increases in Air Pollution: Evidence and Implications for Mortality” by Karen Clay and Nicholas Z. Muller (hereinafter “Clay and Muller”) from Carnegie Mellon University apparently does claim that air quality is getting worse. I am spurred to check this claim out because the last time I checked all the air quality trends were down.

Although the emphasis of my work before retirement was environmental regulatory analysis coupled with emissions reporting I have always been primarily an air quality meteorologist. My experience includes managing an ambient air quality monitoring network, modeling air quality and interpreting the monitoring and modeling results for regulatory applications. Frankly, this claim struck a nerve with me and I felt I had to respond. The opinions expressed in this post do not reflect the position of any of my previous employers or any other company I have been associated with, these comments are mine alone.

Background
The abstract for the Clay and Muller paper states:

After declining by 24.2% from 2009 to 2016, annual average fine particulate matter (PM2.5) in the United States in counties with monitors increased by 5.5% between 2016 and 2018. Increases occurred in multiple census regions and in counties that were in and out of attainment with National Ambient Air Quality Standards (NAAQS). We explore channels through which the increase may have occurred including increases in economic activity, increases in wildfires, and decreases in Clean Air Act enforcement actions. The health implications of this increase in PM2.5 between 2016 and 2018 are significant. The increase was associated with 9,700 additional premature deaths in 2018. At conventional valuations, these deaths represent damages of $89 billion.

The National Trend of PM2.5 graph attributed to the paper in the New York Times article shows an “alarming” reversal of the air quality trend. I am only going to respond to this trend claim and the discussion of possible causes. The discussion of health implications deserves a response too but that will have to wait.

PM2.5

The EPA PM basics page describes this pollutant.  PM2.5 is particulate matter with a diameter 2.5 micrometers and smaller.  It is also called fine inhalable particles and the key point is that these particles are so small that they can be inhaled deeply into the lungs.  Although this can clearly cause health problems there are controversies about threshold effects.  The EPA reference describes the sources of PM:

These particles come in many sizes and shapes and can be made up of hundreds of different chemicals.  Some are emitted directly from a source, such as construction sites, unpaved roads, fields, smokestacks or fires.  Most particles form in the atmosphere as a result of complex reactions of chemicals such as sulfur dioxide and nitrogen oxides, which are pollutants emitted from power plants, industries and automobiles.

Measuring particles this small is difficult and a representative national network for monitoring PM2.5 has only been available since 2000.  However, because most of these fine particles are created from sulfur dioxide and nitrogen oxides, we can use trends from those pollutants as a surrogate for expected levels of PM2.5.

EPA has trend data readily available for all the national ambient air quality standard pollutants.  Also included at the website is an interactive trend report.  Importantly for this post, the EPA website provides links for easy data downloads.

Clay and Muller Analysis

Clay and Muller used daily data from the EPA Air Quality Data System for all observations in the contiguous United States from 2009 to 2018.  They considered not only the total PM2.5 concentration but also the three major species: ammonium nitrate, sulfate and elemental carbon.  The dataset of 1.8 million daily readings was statistically processed to calculate trends not only nation-wide but also in different regions of the country.

Once they found their worrisome trend they examined three possible channels through which the recent increase may have occurred: economic activity, wild fires, and enforcement.  Economic activity was studied by looking at the speciated PM2.5 data to attribute the source of pollution.  Wildfire effects were determined by looking at the West, Midwest, and California regions by omitting June through September data.  To estimate the effect of enforcement they used an EPA database of “actions resulting in a penalty for violations of section 113d of the Act”.

The New York Times article claims that this work shows a reversal of a decades-long trend toward cleaner air and quotes study co-author Nick Muller, a professor of economics, engineering and public policy at Carnegie Mellon as saying “After a decade or so of reductions this increase is a real about-face.”

My Analysis

In order to do this kind of analysis correctly is a big deal.  Ambient air quality refers to the make-up of the air we breathe.  It is affected by air pollution emissions and the meteorological conditions that affect the transport and diffusion of the material between the time it is emitted and the time we breathe it.  For regulatory assessments whether the ambient concentrations comply with the National Ambient Air Quality Standards you need to consider both of those components.  As you can imagine setting up an inventory of all the pollutants that affect PM2.5, developing a meteorological database to model transport and diffusion and finally running a model that not only incorporates those factors but also includes the chemistry that changes the emissions to the chemicals we breathe is a massive undertaking.  Nonetheless, I think we can do something simpler to test the conclusions of the Clay and Muller paper.

Contrary to popular opinion calculating environmental trends is not as simple as you might first imagine.  For example, if the number and location of monitoring stations changes over time then the trends may change because of those changes and not because of some underlying difference in emissions, economic activity or enforcement actions.  As noted previously EPA analyzes and reports on air quality trends.  Importantly their primary consideration is to make sure the data they report represents what is actually happening and does not include data artifacts.  Another point is that because meteorological conditions affect pollution concentrations, we should expect variations in monitored values solely due to weather.  In order to minimize that effect the longer the period of record for the data the better because it averages the weather impacts out.

EPA PM2.5 trend data from a representative network of stations is available from the EPA website starting in 2000 so the obvious first thing to do is to look at all the data available.  The EPA Air Quality Data Summary table lists the annual nation-wide average, the year-to-year differences and the difference over the entire period of record and the difference between 2016 and 2018.  These numbers corroborate the claim that PM2.5 air quality did get worse the last two years.  However, the period of record air quality data in my table show that there has been a marked improvement since 2000:  PM2.5 is down nearly 40%, SO2 is down 82% and NOx is down 33%.  Even though there was a concentration increase the last two years, note that there were three other years when the PM2.5 concentrations failed to go down and two of them had higher increases than the last two years. Similar results are shown in the SO2 and NOx data.

Remember that ambient concentrations are a function of weather and emissions.  Rather than trying to estimate the effect of emissions on concentrations by looking at the observed content it is simpler, and much more likely to be accurate, to simply look at the emissions.  EPA also provides emissions trend data.  The EPA Emissions Data Summary table lists annual PM2.5, SO2 and NOx emissions data in the same format as the air quality data summary table.  These data include emissions from electric utilities, industry, storage, transportation, and a miscellaneous category that includes wildfires.  Estimating emissions from the wide variety of sources means that there is a wide range of data quality but I assume that these values are appropriate for the purpose at hand.  Note that according to these data PM2.5, SO2 and NOx emissions all went down not only over the period of record but also between 2016 and 2018.

Clay and Muller examined three possible channels through which the recent increase may have occurred: economic activity, wild fires, and enforcement.  In their analysis of economic activity, they conclude that “The chemical composition of particulates point to increased use of natural gas and to vehicle miles traveled as likely contributors to the increase in PM2.5”.  Because their analysis did not consider the potential effect of weather on transport and diffusion and because the emissions trend was down even while the ambient concentrations went up, I disagree with that conclusion.  With regards to wildfires, Clay and Muller conclude that wild fires “may account for some of the observed increase in PM2.5 from 2016 to 2018, but not for the general pattern of decline and then reversal.  I think that their methodology is too coarse to pick up a wildfire signal.  Even though the paper doesn’t find a link between enforcement actions and the PM2.5 trend they conclude “The decline in enforcement actions, however, is concerning in light of the increases in air pollution in both attainment and nonattainment counties after 2016.”  In the first place in my experience the majority of enforcement actions have little to do with emissions levels and mostly to do with reporting inconsistencies but I believe it is much more likely that enforcement actions are going down because the regulatory agencies are doing a good job.  That is to be applauded not to be a matter of concern.

Conclusion

I am not impressed with the methodology used in this paper.  Number crunching over a million records to determine a trend has risks that professors of economics apparently did not recognize.  Ambient levels of pollution are affected not only be emissions and the factors that they examined but also by meteorology and monitoring system issues.  The inter-annual changes noted were more likely simply due to meteorology as any change in emissions and precursor emissions that the Clay and Muller paper claimed.

I am trying to give the authors the benefit of the doubt that they did not know any better but I am frustrated that they apparently did not bother to seek the advice of any air quality meteorologist or air pollution monitoring scientist.  I am confident that anyone of those experts would have said the longer the trend the better and don’t expect a perfectly decreasing ambient air quality trend even when the emissions are decreasing over time.  Trying to tease out a rationale for an air quality trend change likely less than the variability of the measurements due to weather is an abuse of air quality trends.  Now that these results have shown up in the New York Times many people have been mislead.

I do not think that the general public understands how much improvement there has been to New York State’s air quality and how big the emissions reductions have been.   This is a summary of the trend of observed levels of SO2, NO2 and ozone since 1980 in New York State and it shows significant improvements. This is a companion to an earlier post showing emissions trends.

I have to apologize for my inability to incorporate tables and graphs in the body of a WordPress blog post. If I had that ability then this post would be a heck of a lot easier to read. Instead I offer three alternatives. Each of the figures and tables is available by links in the following post. I also have prepared a pdf version of this post and you can view that entire document NYS Air Pollution Concentration Trends. Finally that document and a spreadsheet with the data, tables and graphs are available at the NY Pragmatic Environmentalist dropbox.

Although the Environmental Protection Agency has a good air quality trend website, it does not have New York only data available and the NYS Department of Environmental Conservation does not provide a summary of air quality trends. In order to assess New York trends I accessed ambient monitoring data from an EPA website. I made no attempt to limit the monitoring sites used. I just calculated values for all reporting stations. As a result this is not an accurate picture of trends because changing stations can skew the results.

According to the EPA air quality Status and Trends document national air quality levels have decreased significantly from 1990 to 2017:

• Carbon monoxide -77%
• Lead – 80%
• Nitrogen Dioxide (annual) – 56%
• Nitrogen Dioxide (1-hour) – 50%
• Ozone – 22%
• Particulate Matter (2.5µ) (annual) – 41%
• Particulate Matter (2.5µ) (24-hour) – 40%
• Sulfur Dioxide – 88%

For the New York only data I found the following reduction trends:

• Nitrogen Dioxide (annual) – 52%
• Nitrogen Dioxide (1-hour) – 63%
• Ozone – 23%
• Sulfur Dioxide – 93%

New York State annual average ambient (SO2, NO2 and Ozone) trends are shown in Figure 1 NYS Ambient Concentration Trends NYS Trend Summary for SO2 NO2 and Ozone.

In order to compare the air quality to the National Ambient Air Quality Standards I need to show the data in the appropriate reporting format. The standards use complicated averages so the following graphs use the appropriate parameters.

The most problematic pollutant is ozone. The Ozone standard is the 0.070 ppm measured as the annual fourth-highest daily maximum 8-hour concentration, averaged over 3 years. Figure 2 NYS Maximum Annual Ambient 8-hr 4th High Ozone shows the trend of the highest observed value of the fourth-highest daily maximum 8-hour concentration. While averaging over 3 years reduces the values somewhat clearly New York is close to the standard. The highest value in 1988 was 0.148 and in 2017 the observed value was 0.079.

EPA recently instituted a one-hour NO2 standard of 100 ppb measured as the 98^th percentile of 1-hour daily maximum concentrations averaged over 3 years. I did not include the 3-year averaging component but Figure 3 NYS Maximum Annual Ambient NO2 Max 98th Percentile shows the trend of the maximum observed annual value in the state. There are no observed values greater than the standard.

EPA also has a 1-hour NAAQS for SO2. That limit is 75 ppb measured as the 99^th percentile of 1-hour daily maximum concentrations, averaged over 3 years. I did not average these values either so Figure 4 NYS Maximum Annual Ambient NO2 Max 98th Percentile shows the trend of the maximum observed value. There has been a sharp decline in observed values until 2017 when a higher value was observed. That is the result of adding data from a new private monitoring network specifically designed to determine whether there is an issue with this limit for their facility. In 2017 that monitor recorded a 90.5 ppb value for the 99^th percentile.

The spreadsheet, NYS EPA Data for SO2 NO2 and Ozone, lists all the monitoring data for those parameters since 1980. Because of the size of that spreadsheet I did not include it. The spreadsheet included at the dropbox has a summary tab with the data and graphs for the information shown in this post.

I am an air quality meteorologist and a pragmatic environmentalist. My blog usually addresses topics where I appeared opposed to mainstream environmentalist dogma so it has been asked why I even consider myself an environmentalist. I support evidence based environmental controls. Since I started work in my field in 1976 there has been tremendous air quality improvement that addressed serious health and welfare problems. I want to document some of the improvements I have been a party to as an environmentalist in the electric generating industry on Earth Day 2018.

The two primary pollutants associated with acid rain are sulfur dioxide and nitrogen oxides. They are also associated with small particulate matter. United States sulfur dioxide emissions in 1970 31.2 million tons but were only 2.7 million tons in 2016 (91% reduction). United States nitrogen oxide emissions in 1970 26.9 million tons and in 2014 12.4 million tons (54% reduction).

I have been working in New York State most of my career. According to the EPA Clean Air Markets Division, over the twenty year period 1997 to 2016, the sulfur dioxide emission rate dropped 98% from 0.83 to 0.017 lbs per mmBtu. In the same time period, nitrogen oxides emissions dropped 75% from 0.24 to 0.061 lbs per mmBtu.

I am proud of the pollution control improvements at the facilities I worked with before I retired. In particular, I supported the Huntley and Dunkirk coal-fired power plants in Western New York from 1981 to 2010. My job was to report the emissions. The earliest sulfur dioxide and nitrogen oxides data I have for those two plants is from 1984 when the sulfur dioxide emission rate was 2.04 lbs of SO2 per mmBtu and the nitrogen oxide emission rate was 0.56 lbs of NOx per mmBtu. When I retired in 2010, the sulfur dioxide emission rate was 0.527 lbs of SO2 per mmBtu (81% reduction) and the nitrogen oxide emission rate was 0.159 lbs of NOx per mmBtu (73% reduction).

We worked with the New York State Department of Environmental Conservation to implement the control equipment necessary to reduce the emissions. Sulfur dioxide emissions were reduced by changing the sulfur content of the fuel, ultimately using Powder River Basin coal from Wyoming that had a much lower sulfur content that what was used in 1984. It is a testament to the operating staff at those plants that they figured out how to use a much different coal than what the plants were designed to burn when the plants were built before 1960. Nitrogen oxides were controlled by changing the burners a couple of times to more advanced technology and ultimately by adding selective non-catalytic reduction control systems. The addition of a baghouse with activated carbon injection also markedly reduced particulate, opacity and Hg emissions. Sadly despite all these improvements the cost of coal relative to natural gas made both plants uneconomic and they have since shut down.

As a result of these emission reductions, there has been a similar reduction in air pollution concentrations. EPA provides pollutant concentration trend data that documents those reductions. At EPA’s 42 nation-wide SO2 trend monitoring sites the annual average concentration has gone from 154 micrograms of SO2 per cubic meter in 1980 to only 20.2 in 2016 (87% reduction). At EPA’s 23 nitrogen dioxide trend monitoring sites the annual average concentration has gone from 111 micrograms of SO2 per cubic meter in 1980 to only 43.7 in 2016 (61% reduction).

Unfortunately, there has not been a similarly large relative concentration decrease for ozone. At EPA’s 206 nation-wide ozone trend monitoring sites the annual fourth maximum of daily maximum 8-hour average has gone from 0.101 ppm in 1980 to 0.070 in 2016 (31% reduction). Ozone is much more complicated pollutant because it is not directly emitted. Instead it is created in a photo-chemical reaction between nitrogen oxides and volatile organic compounds. As a result there are many more categories of sources to control which complicates improvements.

EPA and others tout the importance on human health of reductions in particulate matter, especially with small particulate matter known as PM-2.5 (the size of the particles is 2.5 microns). EPA only provides trends of PM-2.5 since 2000 because the monitoring equipment was not deployed until then. At EPA’s 455 nation-wide PM-2.5 trend monitoring sites the annual average concentration has gone from 13.4 micrograms per cubic meter in 2000 to only 7.7 in 2016. However, there is a strong correlation between ambient concentrations of PM-2.5 with SO2 and NO2. I did a multiple regression with the 2000-2016 PM-2.5 observations with SO2 and NO2 to guess at the ambient level in 1980. I predict that PM-2.5 concentrations have dropped 68% between 1980 and 2016.

The progress the United States has made in air quality improvement gets overlooked too often today when we seem to hear mostly about problems like ozone that still need to be addressed. However, before 1970 New York City was very polluted and that, for the most part, has been cleaned up. One should also keep in mind that there were some spectacularly wrong predictions made around the first earth day in 1970. Those predictions include the following air quality predictions:

• In January 1970, Life reported, “Scientists have solid experimental and theoretical evidence to support…the following predictions: In a decade, urban dwellers will have to wear gas masks to survive air pollution…by 1985 air pollution will have reduced the amount of sunlight reaching earth by one half….”
• Paul Ehrlich predicted in 1970 that “air pollution…is certainly going to take hundreds of thousands of lives in the next few years alone.” Ehrlich sketched a scenario in which 200,000 Americans would die in 1973 during “smog disasters” in New York and Los Angeles.

Given the demonstrated improvement in air quality as opposed to apocalyptic projections of the past I hope readers keep that in mind when you hear current environmental doom and gloom stories.

This blog is my pragmatic view of environmental issues and, to be honest, goes way down in the weeds because most environmental issues are not simple. Spoiler alert this one is the worst yet.

The reason for this blog post is to document the possibility of a “bad thing” in New York that has a reasonable chance of occurring in late August and September of 2018 and possibly as soon as late this summer. I am worried about compliance and a potential threat to electric system operations with the Cross State Air Pollution Rule (CSAPR) NOx Ozone Season trading program. If the feces get entangled in the impeller remember you heard it here before it happened so you will know that the agencies were told that their plans were risky. Unfortunately in order to describe the “bad thing” you likely need some background information that may put you to sleep.

Before proceeding a disclaimer. Before retirement from the electric generating industry, I was actively analyzing air quality regulations that could affect company operations. The opinions expressed in this post do not reflect the position of any of my previous employers or any other company I have been associated with, these comments are mine alone.

Background

First off, you off to know about trading programs. EPA does a good job describing the fundamentals of cap and trade. What you need to know about this pollution control approach is that there are two components: the cap and tradable allowances for the pollutant covered. The cap sets a limit on the total regional emissions that must be met over a trading season such as a year or during the ozone season from May through September. The cap is set at a level such that the pollutant of interest will be reduced to levels that are supposed to improve air quality to the appropriate standard. Setting the cap level correctly is critically important: too high and the environmental objectives won’t get met and too low and the market mechanism won’t work. It is necessary to measure the emissions accurately and transparently because for every ton of pollution emitted affected sources have to surrender an allowance. EPA’s Acid Rain program is the poster child for a successful cap and trade program because greater than required reductions occurred, earlier than expected and with much lower costs than projected.

The key to cap and trade success is that sources that can implement the most cost-effective controls can install those controls, limit their emissions to less than their allocations and trade their excess allowances to sources with more expensive options. The result is that the cap is met in the most cost-effective manner. My particular concern with cap and trade programs in general, and this one in particular, is that in order for it to be successful somebody has to be able to over-control. The problem is if the cap is set so low that there are no options for sources to over control then there are no chances for generating excess allowances so no one has anything to trade. In the worst case affected sources will only run until they have no more allowances and then they will have to shut down.

Environmental NGOs have argued that cap and trade programs do not guarantee that all the sources reduce their emissions so they claim that it is not fair because some sources will not lower emissions and local air quality will not improve everywhere. However, there are national ambient air quality standards that cannot be exceeded for any pollutant that has a pronounced local impact so no sources should be over an emissions limit that causes those problems. Also the pollutants covered in cap and trade programs are related to regional problems such as acid rain and ozone where the local effects are small. Nonetheless, due to a court settlement, the CSAPR rules include a limitation on state emissions to limit interstate trading to prevent this, in my opinion, non-existent problem.

Updates to the CSAPR were proposed in 2015 and finalized in 2016 that included changes to address this problem. The feature that the CSAPR update rule added for this concern is called the compliance assurance mechanism. In addition to the cap a second level was set to limit interstate transfers. For each state in the trading program, the state’s allowance cap budget plus the newly defined variability limit constitutes that state’s assurance level. Each state’s assurance levels takes into account the inherent variability in the state’s baseline emissions from year to year. The intent is that emissions in states can exceed the assurance level due to natural variability (e.g., hot weather making units run more) but including this means that sources in states cannot rely on out-of-state allowances for routine compliance. In 2017, or any later year, if a state’s total emissions are greater than the sum of the state’s budget and variability limit the assurance provisions are triggered. In this case EPA’s rationale is that the state is using more allowances than necessary for inherent variability and is therefore relying on interstate transfers for compliance. When this provision is triggered, EPA determines which facilities exceeded their individual assurance level and requires them to surrender additional allowances equal to three times the excess over the assurance level.

Because the ozone limit has been racheted down over the years there still are many areas that do not attain the current national ambient air quality standard for ozone. The CSAPR NOX Ozone Season trading program is specifically designed to reduce interstate ozone transport that contributes to that problem. Note, that this is the fifth round of NOx reduction programs for New York. As a result, the easy, cheap and quick NOx control options have already been implemented. It is recognized that pollution control costs increase exponentially as the efficiency increases so any further reductions will be expensive and probably cannot be implemented quickly.

Because the cap level is so important I need to explain how EPA determines the cap size. I could easily double the length of this post and surely put to sleep anyone who has read this far if I were to explain in detail how EPA set this cap. Instead and briefly, they use the production cost model Integrated Planning Model (IPM) to analyze the impacts of air quality policies. This is a massive model that purports to estimate how the entire United States utility sector will react to changes in air quality regulations. In order to do that they have to model not only generator operations, fuel costs and control equipment strategies, but also the transmission system. However, the transmission component has been critically flawed when it comes to New York. In particular, the largest load center in the state, New York City, is mostly on islands, there are limits to the transmission available and consequently there are limits to how much electricity can be transmitted to the City. In order to model the entire United States IPM over-simplifies the New York transmission grid. As a result, EPA IPM modeling projects that the least cost solution is to simply generate power elsewhere and significantly under-estimates the amount of power that has to be generated in the City and Long Island, the resulting emissions necessary to keep the lights on, and sets a cap too low to accommodate the New York City constraint.

The New York allocations from EPA in the draft CSAPR update rule had the same flaw as previous programs because of this short-coming. I was responsible for some comments on the draft and we had some success. The final rule changed the New York allocation for a different reason and raised the final allocation.  In 2016 the New York NOx Ozone Season budget was 10,157 allowances. Even with additional allowances, the final CSAPR 2017 NOx Ozone Season Budget is only 5,135 allowances which is close to a 50% reduction. The 2016 NOx actual ozone season emissions in New York were 6,521 tons which is a 64% reduction from the start of the last New York NOx Ozone Season program in 2008. On the face of it then if Ozone Season emissions in 2017 are the same as 2016, then there will be a 20% shortfall of 1,386 tons.

There is another complication. EPA allows banking, i.e. unused allowances are carried forward and can be used in later years. However, the final regulation for the CSAPR update rule included a reduction in the allowance banks. New York affected sources argued, in vain, that because we had already made significant reductions due to other state initiatives that it would be unfair to discount the banked allowances that were earned as a result of those control investments. EPA calculated that there was a bank of 350,000 allowances in the affected states at the end of 2016. EPA argued that the size of the bank would have precluded additional reduction investments until the bank was reduced considerably so they promulgated a reduction to the total of aggregated variability limits times 1.5. The resulting across the board three to one reduction with no consideration of individual interim state actions was a major hit to NY compliance strategies. If historical emissions remain constant, the affected New York sources only have a bank of 3,060 allowances to cover the shortfall of 1,386 tons.

The EPA allocations are to the state and each state has the right to determine how those allowances are allocated to the affected sources. In order to account for new sources the New York Department of Environmental Conservation sets aside 5% of the total allocation for any new sources that come on line during the year. Previously, any unused allowances eventually were returned to the affected sources. Unfortunately, the Cuomo Administration also had plans for the New York allocations. After the “success” of a new and outside the legislature branch revenue stream from the auction of CO2 allowances for the Regional Greenhouse Gas Initiative, the Administration got wind of these allowances and immediately thought they could do the same thing. However, auctioning this kind of allowance is a whole different ball game and they did not try to auction all the allowances. Instead they siphoned off 10% of the allowances to the Energy Efficiency and Renewable Energy Technology (EERET) account and required that any unused new source set-aside allowances would also go to EERET. So instead of the affected sources getting the full allocation of 5,135 allowances they were only allocated 4,362 allowances. Affected sources in New York begged the State to give them the right of first refusal to buy the allowances that were skimmed off but the language in the rule specified sale on the “open market”. Consequently the State refused to incorporate that request into the sale and, to add insult to injury, specified that all the allowances had to be purchased in one batch. The NY 2017 allowances went to Louisiana and the 2018 allowances went to Texas where because of the size of those state budgets they are a fraction of the variability limit so they will most likely be used there. As best as I can tell the sale of those allowances must have netted over $280,000 for the 2017 EERET allowances.

The final consideration in this tale of an obscure air quality compliance issue is the size of the allowance bank. Academics and environmental NGOs cannot abide large margins between allowances and emissions and, in the case of the RGGI allowance margin, are arguing that the margin should be very small. Their rationale is that if allowances are scarce for those sources that need them to run then they will have to buy them at higher costs which in the case of the RGGI will increase the cost of carbon and eventually influence behavior. On the pragmatic side of affected source compliance however, there are advantages to a comfortable allowance margin. Without delving even deeper into the mire of allowance compliance, there can be regulatory and financial implications in the event that there is an allowance monitoring error that increases emissions discovered after the compliance reconciliation deadline and the affected source does not have enough allowances in its account to cover the difference. Environmental staff associated with emissions monitoring generally recommend keeping at least a 5% buffer in the allowance bank for that contingency. Furthermore EPA acknowledges that there is inherent variability in year to year emissions as specified in their CSAPR variability limit of 21% so companies that provide power to the public like to have banks available to cover operational variations. In my opinion an allowance bank of under 5% is very risky and I would recommend a minimum of 25% to cover operational and monitoring contingencies. The key point is that except in rare instances this issue has only been a theoretical problem at most companies for almost all cap and trade programs.

After I completed the draft of this post I found a recent report on the CSAPR Ozone Season allowance market that may be of interest.

The “Bad Thing”

My congratulations if you have made it this far.

My particular concern is New York compliance with the CSAPR NOx Ozone Season limit. To date no New York cap and trade program has had to deal with a constrained market and I vaguely recall only one instance of a constrained market in any cap and trade program.

Because there is only one update of emissions during the ozone season (at the end of July when the May and June data are submitted), facilities will not necessarily know whether the state has triggered the state’s assurance level with its requirement to surrender additional allowances at the end of the Ozone Season. The result is that facilities will be reluctant to exceed their assurance levels because they will not know whether they only need allowances to cover just the excess or three times the excess because the state exceeded the assurance level cap. There is another aspect to this issue that should not be ignored. Electric generating companies have very strong compliance policies and are very reluctant to even give the perception that they have exceeded their emission limits. It is possible company policies will limit emissions to the assurance level and no higher.

My scenario for a bad thing is that New York will be unable to meaningfully further reduce NOx emissions in the near term. If the next couple of summers are warm that will exhaust the current allowance bank to de minimus levels. The ultimate problem with a cap and trade program is that if allowances are not available then the only compliance option is to not run. There is little question in my mind that CSAPR allowances will be available somewhere but that may not be enough to prevent localized operational disruption due to allowance compliance uncertainties. The cumulative effect of the EPA constraints on interstate trading, the uncertainty of the status of emissions relative to the compliance assurance mechanism, and the lower than appropriate cap on NY emissions exacerbated by the Cuomo administration’s unwillingness to give NY affected sources the opportunity to purchase the allowances taken by the State means that New York State affected sources could easily be in uncharted territory. It is not clear how they will react but risking a compliance penalty is not in their best interests.

So my perfect storm worst case scenario is two warm summers that pushes the state close to the compliance assurance limit and reduces the NY allowance bank for one or more affected-source companies to low levels after the June emissions data are known in early August (it takes a month for the data to get reported). Companies with the small number of allowances available find they cannot purchase enough allowances on the market to cover their emissions and possible CAM penalties or find that the costs are so high they don’t think they can recover the cost of purchasing allowances so they get to the point where they simply have to tell the system operator that their units cannot run. This will precipitate a controversy at best and, in an order of magnitude less likely worst case, could even threaten grid reliability. I don’t think the last possibility is very likely but I do think that bringing system reliability into danger because of the regulatory decisions by EPA and NYS that ignored industry recommendations in this instance is possible.