The Committee to Review the Draft Climate Science Special Report (“The Committee”) commends the CSSR authors for producing an impressive, timely, and generally well-written report. The Committee was generally impressed with the breadth, accuracy, and rigor of the draft CSSR. The draft CSSR emphasizes the robust evidence that human emissions of greenhouse gases (GHGs) have substantially warmed the planet and are causing myriad changes to the Earth system, some of which are effectively irreversible on human timescales.
The draft CSSR draws on existing climate change assessments while also providing important new research findings and observations. Assessments of climate science are now routinely produced. Authoritative documents include the science volume of America’s Climate Choices (NRC, 2012), the Working Group I contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (AR5WG1, IPCC 2013), and the climate science chapter of the Third U.S. National Climate Assessment (NCA3, Melillo et al., 2014). The draft CSSR is new and significant in several ways. First, it focuses on changes in the climate system as they affect the United States and provides a much more comprehensive evaluation of physical climate changes than was included in the climate science chapter of NCA3. Second, the report provides a synthesis of recent manifestations of continued climate change: a new global temperature record set in 2014, which was broken in 2015 and again in 2016 thanks in part to a strong El Niño event (e.g., Lean and Rind, 2008, who quantified the contribution of El Niños to global temperature); continued decline in Arctic sea;1 and record high globally averaged atmospheric carbon dioxide (CO2) concentration which has now passed 400 ppm.2 Third, the draft CSSR includes several significant advancements that have been made in the science of climate change, including the rapid development of the field of extreme event attribution, which also was the subject of a recent National Academies report (NASEM, 2016a), and new evidence concerning the Antarctic ice sheet that raises and better quantifies the upper bounds of projected sea level rise (SLR).
These recent, observed changes in Earth’s climate system and substantial advancements in the science of climate change underscore the importance of up-to-date assessments. By building on previous work and also by showing recent advances, the draft CSSR provides a valuable update. The CSSR will also serve as a useful resource for evaluating the implications of climate change for the United States and its territories, which will be the subject of NCA4, due for release in 2018.
II.1 OVERALL COMMENTS
The Committee agrees that the draft CSSR is largely accurate and generally represents the breadth of available literature pertaining to the state of the science at the time of writing, with the exception of some specific topic areas detailed in this report. Assessment reports like the draft CSSR are most effective when they convey sufficient detail using relatively simple language. This can be achieved by providing authoritative statements about the current state of the science, which necessarily include some facts that have been well established for decades, and also recent observations and findings. Impactful assessments also use scientific language that is accurate
enough for the specialist to know exactly what is meant, while also being comprehensible to a broad audience. The draft CSSR generally demonstrates these characteristics, although the Committee notes below some ways that the draft report can be improved.
The draft CSSR could be strengthened by more clearly distinguishing, in the chapters and the Executive Summary (ES), what is truly new and significant. Separating this new information from the longstanding foundational science that underpins the report would improve its impact and usability. A list of “what’s new” appears at the end of the ES, but the Committee suggests that each chapter examine its key findings and find ways to delineate what is a new or significantly updated observation, a new or important line of evidence, or is simply an important and significant aspect of climate change that was already part of the foundation of the science. This emphasis could be achieved through specific language more clearly identifying which key findings are new, by reducing the amount of text devoted in key findings to long-accepted truths, by reordering the key findings, or by color-coding the text of the key findings.
The U.S. regions provided in the draft CSSR (that will also be used in NCA4) have been modified since NCA3. One result of this change is that a new Caribbean region has been created. The draft CSSR barely mentions the Caribbean and includes no results for the region that the Committee could find, apart from the maps of projected temperature and precipitation change (e.g., Figure 6.7). Any data and findings that can be provided would probably be useful to the authors of the Caribbean chapter of NCA4. If data and findings cannot be provided, that should be noted.
To strengthen the impact and message of the draft CSSR, the Committee recommends adding quantitative statements to the key findings throughout the report where possible. Values are provided for some key findings (usually those related to temperature) and are effective in making the messages more impactful, but more values could be reported. More specific recommendations in response to the questions in the Statement of Task about data and statistics are provided throughout this report.
Throughout the draft CSSR, it would also be helpful to better link related topic areas across chapters, to provide guidance to the reader. For instance, in Chapter 10 where drought is discussed, it should be indicated that Chapter 8 covers drought in greater detail.
II.2 RESPONSE TO THE STATEMENT OF TASK
Are the Goals, Objectives, and Intended Audience of the Product Clearly Described in the Document? Does the Report Meet Its Stated Goals?
The Front Matter (page 1, lines 2-13 of the draft CSSR, see also Box 1 of this report) adequately describes the goals and objectives and, with the exception of the omission of the Caribbean and other smaller examples provided later in this review, it meets those goals. The intended audience is described as follows: “Much of this report is intended for a scientific and technically savvy audience, though the Executive Summary is designed to be accessible to a broader audience” (page 1, lines 14-15 of the draft CSSR, also provided in Box 1 of this report). The Committee considers this description of the audience to be insufficiently clear. For instance, a technically savvy audience may be interpreted as those with familiarity with technological advancements, which is not necessarily equivalent to a general understanding of the physical sciences contained in the draft CSSR. As such, the Committee suggests rewording this statement as follows:
The material presented in the chapters of this report is intended to be understood by a scientifically literate audience. The Executive Summary is designed to be accessible to a more general audience.
In some places, too many terms are unfamiliar to anyone but a specialist in the field, and in those instances the text fails to meet the goal of communicating effectively to the intended audience. Specific locations in the draft CSSR where this concern arises are noted in Chapter III of this report. Some such terms may be unavoidable, but should be explained and defined in the text or glossary. The table of contents of the draft CSSR includes a putative glossary but that glossary is missing. The Committee provides some specific words that should be considered for inclusion in a glossary and these are listed in the Line Comments (Appendix A).
Does the Report Accurately Reflect the Scientific Literature? Are There Any Critical Content Areas Missing from the Report?
The draft CSSR, in general, accurately reflects the scientific literature, with an emphasis on recent material, with the exception of some specific topic areas detailed in this review. In some instances, the Committee notes minor omissions or significant imbalances where the extent of existing literature on a given topic is not adequately cited or discussed. For instance, the treatment of hydrology in Chapter 8 needs to be more thorough. Some discussion of the concept and quantification of climate sensitivity and transient heat response would be useful to also include, perhaps in Chapter 2, where it is currently mentioned in one line. Recommendations are further detailed in Chapter III for individual draft CSSR chapters, with specific suggestions for improvements and some recommended publications to consider citing.
Are the Findings Documented in a Consistent, Transparent, and Credible Way?
Most of the findings are well documented. However, the Committee provides a number of suggestions where documentation could be improved, with the most significant provided here and additional suggestions detailed in Chapter III.
The traceable accounts that support the key findings often contain an insufficient level of detail and could be better utilized. According to the draft CSSR, traceable accounts support each key finding and “document[s] the supporting evidence, process, and rationale the authors used in reaching … conclusions, and provides additional information on sources of uncertainty through confidence and likelihood statements.” The description of evidence base provided in the traceable accounts for many key findings across many chapters list citations noted to support the finding, but do not summarize the evidence contained within those citations. This results in a low level of detail, making it difficult for readers to understand the evidence base and lessening the impact of the finding. This contrasts with the NCA3, in which many key findings were supported by a full page or more (in the final printed version). This issue needs careful attention throughout the report.
In some places, AR5WGI findings are cited simply as IPCC (2013). For traceability, it would be far better to follow recommended practice and cite the specific chapter, since the entire IPCC report is over 1,500 pages.
Many of the figures (specifically listed in the relevant sections of Chapter III) are presented with insufficient information on how a specific calculation was performed or which data or tools were used. This is a significant weakness, but one that should be straightforward to remedy.
Some chapters are very unevenly represented in the ES. For instance, there are six bullet points for Chapter 12’s five key findings while no key findings from Chapter 10 are listed. This disproportionate representation might be reasonable and justified, but it is not obvious that this is the case. The Committee encourages the authors to consider whether the overall balance of the bullet points is appropriate.
The topic of extreme events should be presented with greater detail and further consideration should be given to the most appropriate metrics to report. The current approach,
especially as used to construct figures, could be better connected to the peer-reviewed literature (by using widely accepted methods and considering multiple metrics). In many cases, an insufficient amount of information is provided for the reader to understand underlying methods. For example, Figure 6.3 (also included as ES.5) contains two time series (bottom panels), but the text in Chapter 6 and associated traceable accounts do not provide any details on how the spatially averaged time series were calculated. Attempts by committee members to reproduce the plot were unsuccessful. In general, because there are several possible metrics for extreme heat in the literature (e.g., Hartmann et al., 2013, page 221), the draft CSSR should assess the consistency of conclusions across metrics and present only those that fairly represent robust conclusions across studies and metrics. For heat, in addition to “Txx” (warmest day of the year), Hartmann et al. (2013) also uses Tx90p (90th percentile day), and various studies have used definitions of heat waves like highest 3-day minimum temperature, heat index, etc. Since conclusions across metrics are inconsistent in some cases, the discussions of changes in extremes should summarize the state of knowledge and describe how/whether the results depend on metrics chosen (e.g., Txx vs. Tx90p).
A related issue of clarity with regard to extremes is spatial consistency. Studies of changes in extreme precipitation at individual weather stations find a wide variety of trends (and results can depend profoundly on which metric is selected); spatially aggregating the trends to a relatively large scale does seem to result in a regionally averaged increase in extreme precipitation (e.g., Min et al., 2011; Zhang et al., 2013) and as shown in Figures 7.3 and ES.4. But, the underlying message of the spatial complexity is not well articulated in the draft CSSR, especially when accompanied by language like “Heavy precipitation events across the United States have increased....” The Committee recommends careful consideration of the appropriate level of detail concerning spatial complexity (e.g., plotting station-level or climate-division trends), robustness across metrics (e.g., plotting multiple time series of different metrics), and traceability. These issues appear in at least Chapters 6, 7, 8, and 9.
Are the Report’s Key Messages and Graphics Clear and Appropriate? Specifically, Do They Reflect Supporting Evidence, Include an Assessment of Likelihood, and Communicate Effectively?
Comments on individual figures are given in Section II.3 (for the ES) and in Chapter III (for individual chapters). Some of this Committee’s recommendations apply to multiple figures. See also the points made previously about clarity and supporting evidence for heat and precipitation extremes.
Some maps presenting climate model outputs use a Mercator projection that leads to a low ratio of data to map area (e.g., Figure ES.3). This results in a majority of the map consisting of information-free gray oceans and more space given to Canada than to the continental United States. Using a different projection, and including Hawai’i and Alaska (but not necessarily devoting space to place them in their correct locations), would allow the reader to learn more about changes projected for the continental United States. Also, the contour intervals used for plotting colors on the maps could be a bit finer to aid the reader. If links could be provided to online plotting tools that NCA4 authors could use, that would further increase the utility of these figures.
The Committee noticed that there are nine graduations of likelihood provided on page 4, but only five are used in the draft report, so they may not all be needed.
As with any report written by a committee, an editing pass will improve consistency and readability. Some chapters achieve excellent readability for the intended audience by minimizing use of jargon, appropriate word choices, and clear language including sentence construction. Chapters that do not read as clearly are noted in this report. The word ‘robust’ is in some respects a term of art with specific connotations, but is used with different meanings in different contexts in
the draft CSSR. The draft also reports carbon (C) in units of both PgC and GtC which are identical, and using both units is needlessly confusing. There may be some advantages to using a CO2 metric such as Gt CO2 throughout, as it is consistent with that used in IPCC AR5 2013. Regardless, the Committee recommends choosing one reporting approach for carbon emissions and using it consistently throughout the CSSR.
Are the Data and Analyses Handled in a Competent Manner? Are Statistical Methods Applied Appropriately?
The previous comments above about extreme events also apply here.
In some places, time periods over which change is discussed are somewhat different. While these constraints sometimes result from citation of published literature and data records, in other cases (which the Committee tried to identify and note) they seem to be more amenable to standardization. The draft CSSR uses a metric of 20th century change defined as the 1986-2015 average minus the 1901-1960 average. The Committee recommends that the CSSR authors recompute the values where possible, using a different method detailed next.
The Committee recommends using the following guidelines that would improve the statistical treatment of data throughout the draft CSSR, and encourages all individual-chapter authors to consistently apply this approach:
- Be clear enough about how each calculation is done that a reader could reproduce or find the reported value or plot.
- Be consistent. As much as possible, minimize differences in baseline time periods and methods (cf pages 13-14).
- Include significance statements and/or ranges as appropriate.
- When consistency is not possible, use methods or baseline time periods established in literature (e.g., IPCC 2013 uses 1850-1900 as a baseline for global mean temperature).
- When discussing rates of change, use slope-based methods (e.g., regression or Theil-Sen, that minimize end effects), rather than comparing time periods, if appropriate for the metric being discussed. Since slope-based methods incorporate all available data, they can better represent rates of change.
- Wherever possible, figures depicting observed trends should indicate the statistical significance of those trends, or confidence intervals.
If these recommendations are incorporated, the “Guide to the Report” section could then be updated to describe the statistical approaches. If the current approach is retained, the descriptors of 1901-1960 should be carefully checked, as there were examples referring to it as “early 20th century” and the like.
Are the Document’s Presentation, Level of Technicality, and Organization Effective? What Other Significant Improvements, If Any, Might Be Made in the Document?
Generally, yes, the level of technicality and organization are effective. Chapter III discusses where specific chapter edits could improve the presentation, level of technicality, or organization, and where other improvements could be made.
II.3 COMMENTS ON THE EXECUTIVE SUMMARY
The ES is strong, well-written, and in most cases accurately represents the consensus and breadth of viewpoints. In this section the Committee focuses comments primarily on the figures and the “New Understanding” and “Better Tools and Approaches” sections of the ES. It is the expectation that authors will address chapter-specific comments provided in Chapter III and then edit the ES further to integrate those recommendations, along with the explicit recommendations for the ES given here.
Figure ES.1: It appears there are missing data in the Arctic and Antarctic, but the color is indistinguishable from ‘no warming’ which is certainly not the case. The Committee suggests introducing a different color, perhaps gray, to indicate missing data more clearly. The figure should also show statistical significance of the trends and add the data source.
Figure ES.2: Since the Paris Agreement aims to implement GHG emissions reductions that would achieve a concentration pathway similar to Representative Concentration Pathway (RCP) 2.6, it would be useful to illustrate the RCP2.6 scenario in this figure. One possible approach to including this could be to have the figure include the boxes to the right indicating the ranges for all four RCPs, as in the IPCC 2013 equivalent (SPM.7).
Figure ES.4: The Committee suggests indicating which, if any, of the trends shown are statistically significant, in addition to considering the previous comments about observed trends, baseline periods, and spatial aggregation of data. Moreover, this figure visually resembles Figure 2.18 presented in NCA3, but the numbers are quite different, perhaps because of the use of a different metric of extreme precipitation. It is fine to show a different figure, but this underscores the previous point about consistency and robustness of measures of extremes, and would benefit from some explanation. It would also be appropriate to explain any other figures that resemble NCA3 graphics but convey a different impression.
Figure ES.5: This figure is problematic for a number of reasons outlined in the previous comments on extremes. Also see Section III.6.
Figure ES.6: This figure does not convey new or important science and could be removed.
Figure ES.8: This figure does not appear in Chapter 12 as foundational material. Additionally, it is busy, difficult to read due to small font, and too complicated. A single panel could be chosen for the ES, and an improved version could appear in Chapter 12. If retained, the maximum value on the y-axis should be set to 365 and the caption should explain that this is an upper limit and results in some curves displaying an inflection point (and in some cases small differences between scenarios, which is counterintuitive at first).
Figure ES.9: This figure provides a compelling illustration of observed sea ice change in the Arctic, but would benefit from a comparison of 2016 (or an average of recent years) with a multi-year average from early in the satellite era, for more robust statistical representation. See also Section III.11.
None of the material from Chapter 10, and too little of the material from Chapter 2, appears in the ES. This may be deliberate, but the Committee considers some of the findings from Chapters 2 and 10 to be worthy of representation in the ES. In particular, a simplified version of Figure 2.6 would improve the ES (see Section III.2 for more details).
The bullet regarding limiting the global mean temperature increase to 2oC (page 27, lines 17-24) that states, “cumulative emissions would likely have to stay below 1,000 gigatons carbon (GtC)” is given without a citation and is inconsistent with the 790 Gt C cited in IPCC AR5 2013. See also Section III.14.
The ES would have more impact if it more clearly emphasized what is new in the draft CSSR relative to previous climate change assessments. The “Summary of What’s New Since
NCA3” at the end of the ES is not prominent, lacks quantitative values, and is weakened by the inclusion of methodological changes. The full list of “Better Tools and Approaches” is more appropriate for Chapter 1. The “New Understanding” sections on extremes (page 29) could particularly benefit from re-ordering the bullets (e.g., moving lines 24-27 later, adding material to lines 21-23 and/or 28-29 to emphasize the large number of types of extremes for which a human contribution has been identified with confidence), and including quantitative statements. The Committee is skeptical about the value of extensive discussion on the ‘hiatus’ given that any time series with a trend and nonzero variance has short periods when the trend is opposite the underlying trend. Rather than continuing to focus on the hiatus, the Committee recommends shortening the discussion for this topic and rephrasing page 29 from line31 through page 30 line 2 with a statement to the effect that short-term variability (resulting in either strongly positive or flat trends) is not the best indicator of whether climate is changing in response to GHGs. The text could also note that conversely a recent string of 3 record warm years (2014-2016), occurring in part as a consequence of a strong El Niño, also does not prove an acceleration of warming—both are artificial statements that result from focusing too much on short periods of record. See Section II.1 of this report for additional recommendations on this topic.
This page intentionally left blank.