Simply Statistics A statistics blog by Rafa Irizarry, Roger Peng, and Jeff Leek

The Supreme Court takes on Pollution Source Apportionment...and Realizes It's Hard

Recently, the U.S. Supreme Court heard arguments in the cases EPA v. EME Homer City Generation and _American Lung Association v EME Homer City Generation. _SCOTUSblog has a nice summary of the legal arguments, for the law buffs out there.

The basic problem is that the way air pollution is regulated, the EPA and state and local agencies monitor the air pollution in each state. When the levels of pollution are above the national ambient air quality standards at the monitors in that state, the state is considered in “non-attainment” (i.e. they have not attained the standard). Otherwise, they are in attainment.

But what if your state doesn’t actually generate any pollution, but there’s all this pollution blowing in from another state? Pollution knows no boundaries and in that case, the monitors in your state will be in non-attainment, and it isn’t even your fault! The Clean Air Act has something called the “good neighbor” policy that was designed to address this issue. From SCOTUSblog:

One of the obligations that states have, in drafting implementation plans [to reduce pollution], is imposed by what is called the “good neighbor” policy.  It dates from 1963, in a more elemental form, but its most fully developed form requires each state to include in its plan the measures necessary to prevent the migration of their polluted air to their neighbors, if that would keep the neighbors from meeting EPA’s quality standards.

The problem is that if you live in a state like Maryland, your air pollution is coming from a bunch of states (Pennsylvania, Ohio, etc.). So who do you blame? Well, the logical thing would be to say that if Pennsylvania contributes to 90% of Maryland’s interstate air pollution and Ohio contributes 10%, then Pennsylvania should get 90% of the blame and Ohio 10%. But it’s not so easy because air pollution doesn’t have any special identifiers on it to indicate what state it came from. This is the source apportionment problem in air pollution and it involves trying to back-calculate where a given amount of pollution came from (or what was its source). It’s not an easy problem.

EPA realized the unfairness here and devised the State Air Pollution Rule, also known as the “Transport Rule”. From SCOTUSblog:

What the Transport Rule sought to do is to set up a regime to limit cross-border movement of emissions of nitrogen oxides and sulfur dioxide.  Those substances, sent out from coal-fired power plants and other sources, get transformed into ozone and “fine particular matter” (basically, soot), and both are harmful to human health, contributing to asthma and heart attacks.  They also damage natural terrain such as forests, destroy farm crops, can kill fish, and create hazes that reduce visibility.

Both of those pollutants are carried by the wind, and they can be transported very large distances — a phenomenon that is mostly noticed in the eastern states.

There are actually a few versions of this problem. One common one involves identifying the source of a particle (i.e. automobile, power plans, road dust) based on its chemical composition. The idea here is that at any given monitor, there are particles blowing in from all different types of sources and so the pollution you measure is a mixture of all these sources. Making some assumptions about chemical mass balance, there are ways to statistically separate out the contributions from individual sources based on a the chemical composition of the total mass measurement. If the particles that we measure, say, have a lot of ammonium ions and we know that particles generated by coal-burning power plants have a lot of ammonium ions, then we might infer that the particles came from a coal-burning power plant.

The key idea here is that different sources of particles have “chemical signatures” that can be used to separate out their various contributions. This is already a difficult problem, but at least here, we have some knowledge of the chemical makeup of various sources and can incorporate that knowledge into the statistical analysis.

In the problem at the Supreme Court, we’re not concerned with particles from various types of sources, but rather from different locations. But, for the most part, different states don’t have “chemical signatures” or tracer elements, so it’s hard to identify whether a given particle (or other pollutant) blowing in the wind came from Pennsylvania versus Ohio.

So what did EPA do? Well, instead of figuring out where the pollution came from, they decided that states would reduce emissions based on how much it would cost to control those emissions. The states objected because the cost of controlling emissions may well have nothing to do with how much pollution is actually being contributed downwind.

The legal question involves whether or not EPA has the authority to devise a regulatory plan based on costs as opposed to actual pollution contribution. I will let people who actually know the law address that question, but given the general difficulty of source apportionment, I’m not sure EPA could have come up with a much better plan.