Scientific shortcomings in the EPA's endangerment finding from greenhouse gases.

• Author: Michaels, Patrick J.
• Position: Environmental Protection Agency - Report

On April 24, 2009, the U.S. Environmental Protection Agency issued a finding of "proposed endangerment" from climate change caused by six greenhouse gases, with the largest contributions to warming resulting from emissions of carbon dioxide and methane (EPA 2009a).

The EPA also referred to this document as a "proposed finding" in response to a 2007 decision by the U.S. Supreme Court, Massachusetts v. EPA, which empowered the EPA to make such a finding for greenhouse gases under existing law. This was the Court's interpretation of Section 202 (a) of the Clean Air Act Amendments of 1990.

In Proposed Endangerment, the EPA requested comments "on the data on which the proposed findings are based, the methodology used in obtaining and analyzing the data, and major legal interpretations and policy considerations underlying the proposed findings" (EPA 2009a: 18890).

We answered the EPA's request in a filing on June 23, 2009 (Michaels, Knappenberger, and Davis 2009). This article details some of the most relevant findings in our response. In general, we found that Proposed Endangerment suffered from systematic errors that were inevitable, given that the way in which the EPA chose to determine the required background science had to result in both biased and outmoded climate science. In addition, the EPA made grand and sweeping assumptions about human adaptation to climate that are of such illogie as to invalidate the entire study. We believe that these systematic errors call into question any attempt on the EPNs part to subsequently issue regulations on greenhouse gas emissions. Other entities that may use the EPA's document as a basis for emissions reductions will be using similarly incomplete science and be subject to severe and public criticism. This conclusion has obvious implications for upcoming negotiations in Copenhagen in December 2009, on a replacement for the Kyoto Protocol.

The EPA's major systematic errors with regard to science concerned the way in which it provided scientific background for Proposed Endangerment. The essence is given in Section III (A), "Approach in Utilizing the Best Available Scientific Information":

EPA has developed a technical support document which synthesizes major findings from the best available scientific assessments that have gone through rigorous and transparent peer review. The TSD therefore relies most heavily on the major assessment reports of both the Intergovernmental Panel on Climate Change and the U.S. Climate Change Science Program. EPA took this approach rather than conducting a new assessment of the scientific literature [EPA 2009a: 18894].

In doing so, the EPA is ultimately relying on the May 2007, Fourth Assessment Report (AR4) of the United Nations' Intergovernmental Panel on Climate Change (IPCC 2007). Given the lead times for publication of that compendium, (1) some of the cited science very well could have been modified, altered, or invalidated in the subsequent period of dose to three years between the dosing time for AR4 input and Proposed Endangerment. The Climate Change Science Program (CCSP) reports consist of a long series of 16 separate documents, with publication dates between 2006 and 2009 (see EPA 2009b). Many of these were contemporary with AR4. CCSP documents, as demonstrated below, relied heavily on AR4, as did the EPA in its technical support document (TSD).

As we will also show, there were multiple instances where both the United Nations' AR4 and the United States' CCSP missed, omitted, ignored, or unfairly dismissed relevant portions of the refereed scientific literature that are highly germane to Proposed Endangerment. This is an additive to the problem of the timeliness of AR4 and many of the CCSP reports that the EPA's TSD relied so heavily upon. (2)

None of the compendia upon which the EPA relied make any mention of the concept of "publication bias," which is a natural skewing of professional literature toward "cause and effect" publications. In the case of global warming, the literature that is surveyed in compendia is likely to be overwhelmingly biased toward linkages between climate change and certain effects, even though negative results--no effect--might in fact be scientifically noteworthy. This phenomenon was first defined by Rosenthal (1979) where he described the "file drawer problem," in that research demonstrating negative results tends to be put in the back of a researcher's files because they are inherently more difficult to publish.

Publication bias has been well documented in both the economic and biomedical literature (e.g., de Angelis et al. 2004, Chan et al. 2004) but was first noted in the climatic literature by Michaels (2008). In general, he concluded that publication bias would create biased compendia which, if relied upon for policy (as is the case for Proposed Endangerment), would likely overestimate threats from and underestimate immunities to climate change. In this study, we find a disturbing number of omissions of citations in both AR4 and the CCSP foundation documents for Proposed Endangerment, consistent with the notion of publication bias.

Stephen Jay Gould described another form of publication bias, which is similar in nature to the normative behavior of science described classically by Thomas Kuhn (1962) in his book, The Structure of Scientific Revolutions. Gould (2002: 763) argued that publication bias results from "prejudices arising from hope, cultural expectation, or the definitions of a particular theory [that] dictate only certain kinds of data will be viewed as worthy of publication, or even documentation at all." From the examples we shall give, it is quite clear that Proposed Endangerment was not immune to the influence of publication bias.

Another major systematic problem with Proposed Endangerment is that it purposefully neglects adaptation both to secular climate and climate change (whether caused by greenhouse gas changes or not). In fact, it is the nature of our species to adapt to climate with clothing, shelter, and social structures. Because it is also natural for climate to change, we also adapt to change at the same time. As will be demonstrated below, some of those adaptations in fact result in a society that is more resilient to prospective warming caused by greenhouse emissions. The EPA, in a key statement, ignores this, and instead considers adaptation to any climate or climate change as prima facie evidence for negative impacts of greenhouse gases, when in fact such adaptation may immunize society against those negative impacts. The EPA clearly violates this reality when it states:

However, it is the Administrator's position that the purpose of the endangerment analysis is to assess the risks posed to public health and welfare, rather than to estimate how various adaptation and greenhouse gas mitigation policies may ameliorate or exacerbate any endangerment that exists. Indeed, the presumed need for adaptation and greenhouse gas mitigation to occur to avoid, lessen or delay the risks and impacts associated with human-induced climate change presupposes that there is endangerment to public health or welfare [EPA 2009a: 18894].

We will cite examples where adaptation to natural (as well as anthropogenerated) climate and climate change clearly reduces or in fact may reverse endangerment, resulting in improvements in health and welfare despite greenhouse gas induced warming.

Method of Analysis

Our analysis isolates key statements in Proposed Endangerment that are either no longer correct, because of changes in global warming science, or are incorrect because of omission of relevant information from the refereed scientific literature. This article offers only a few examples out of many. (3) However, before that is undertaken, it is important to discuss a major misconception in Proposed Endangerment about the documents upon which it is based.

As noted earlier, Proposed Endangerment is supported by the EPA's technical support document, which is based upon the CCSP and AR4 reports. As a result, the TSD purports to synthesize "major findings from the best available scientific assessments that have gone through rigorous and transparent peer review" (EPA 2009b).

This is hardly the case. In peer reviewed science, an article, or book, or compilation is submitted to an impartial editor. The editor sends it out to reviewers whose identity is not revealed to the authors. The reviewers comment on the manuscript, recommending acceptance, rejection, or modification.

The editor then sends the reviews, or summarizes them for the authors, asking for changes on the manuscript or response to the review comments as a prerequisite for publication. If the authors do not respond satisfactorily or do not change the manuscript to meet the editor's request, the manuscript is not published.

The IPCC's AR4 clearly does not conform to this model. Instead, compilations of various aspects of climate science, called "zeroeth order," taken from individual contributions of a large number of sub-specialists and modified by "coordinating lead authors" (who are chosen by governments rather than by scientists) are sent out for review. The reviewers are not anonymous. The lead authors then choose which comments to respond to, and then produce a second draft. There is no independent editor who can demand, upon condition of publication, that certain reviews be either responded to in the text or refuted outside of the text. The EPA is not relying on peer review in the normal scientific sense when it states that the IPCC reports are "peer reviewed."

A similar situation accrues for the CCSP documents. Again, the writers determine which reviews to respond to rather than an editor who is independent of the production process. Finally, the same process applies to the EPA's TSD itself. So it is clear that for the three "foundation" documents cited in Proposed Endangerment, their description as "peer reviewed" in no way...