National Academies Press: OpenBook
« Previous: 2 Key Issues
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 14
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 15
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 16
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 17
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 18
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 19
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 20
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 21
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 22
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 23
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 24
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 25
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 26
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 27
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 28
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 29
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 30
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 31
Suggested Citation:"3 Review of Individual Chapters." National Research Council. 2008. Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change. Washington, DC: The National Academies Press. doi: 10.17226/12135.
×
Page 32

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

3 Review of Individual Chapters This chapter provides specific comments on the four individual chapters of draft Synthesis and Assessment Product (SAP) 1.3. In some cases, these specific comments relate to the overarching comments provided in the previous two chapters of this review. In the other cases, these specific comments are generally minor in nature. The review of each chapter includes a statement that summarizes the committee’s overall thoughts. For some chapters, there are enumerated comments that follow this statement to provide suggested editorial changes or other details for the authors to consider during the revision process. ABSTRACT General remarks: The abstract has two paragraphs summarizing the attribution section, and no paragraphs summarizing the reanalysis section. The abstract would benefit from a better balance. The use of “variations in global sea surface temperatures” can be misconstrued as referring to temporal variations in the global-mean sea surface temperature. The committee suggests deleting “global” wherever it appears in this context. Specific remarks: L61-63: Wording implies that sea surface temperature variations are independent of anthropogenic forcing. PREFACE General remarks: The committee finds the tables corresponding to treatment of uncertainty on Page 11 unhelpful without more context or specific examples. While it makes sense to use terms consistent with IPCC, the quantitative probabilities can only be interpreted in the context of the models that are used to estimate them. A note should be added here that the 14

Review of Individual Chapters 15 specific context will always be made clear throughout the document. One way to improve Table P1 would be to add column headings and include a category for 0.33<p<0.50. Specific remarks: L 131, 138: “climate” should be replaced with “weather and climate” L225: “for” should be replaced with “with” L226: Delete “up through” L229: “supercede” should be replaced with “supersede” EXECUTIVE SUMMARY General remarks: Most of the key findings on attribution listed on P18-21 regarding surface temperature and rainfall trends are based on surface observations and climate model simulations, and thus can be assessed without reanalysis directly. The committee finds that most of the conclusions appear to be independent of reanalysis and that the authors need to strengthen the case that reanalysis is in fact critical in reaching these conclusions, for example by stressing the indirect use of reanalysis for the attribution of climate variability and in testing the Global Circulation Models (GCMs). Page 20: need some statements regarding the fact that the SST changes may be due to anthropogenic forcings. There is some confusion about the usage of “change”. One suggestion is to replace “a change has occurred” (or similar wording) with “an anthropogenic change has occurred” on L418 & throughout: Changes caused by solar variability would be called changes by this document, yet they are natural. Later (lines 3044-3046), for example, changes are partly attributed to natural causes. This language needs to be much more precise. Specific remarks: L278-279: “conditions and, more generally, conditions of other” should be replaced with “conditions, including various”; L278-279: “the oceans” should be replaced with “the atmosphere, oceans” L312: “consistent” should be replaced with “internally consistent” L329: “synoptic (weather)” should be replaced with “regional” L378-383: This evidence is among the weakest on this point in the relevant chapter.

16 Review of CCSP SAP 1.3 L423-426: Quotation of “some published evidence” does not rise to the level of scientific confidence meriting inclusion in the Executive Summary. CHAPTER 1 INTRODUCTION General remarks: Chapter 1 needs to be revised so that it fulfills the educational component of the document. The committee is concerned that this chapter is not written so that it can easily be understood by the non-specialist and that it does not adequately explain reanalysis and attribution and how these techniques are related. These concerns are especially relevant to this chapter, as it sets the stage for (and provides a summary of) the other chapters. The document should be revised by including explanations and use plain language that make the results more easily interpretable to a non-technical audience. Finally, the authors should clearly explain the methodology and its limitations at the outset and the authors should also explain why new reanalysis is needed. Some specific examples are the following: The committee feels that the medical analogy does not work. It describes a process analogous to that done by climate scientists in creating a surface temperature reconstruction such as that shown in Fig. 1.3, but this is not a reanalysis as it is correctly defined in Chapter 2. The same applies to the accident reconstruction analogy. Both are missing the defining characteristic of a reanalysis, which is integrating data into a self- consistent, multivariate representation spanning a long period of time. Simply collecting or retrieving the data and examining it does not capture this essence. L584-585: The interaction effect is an important consideration, but through the rest of the document this effect is ignored and causes are presumed to be linearly additive. For example, more than half of the change is likely to be due anthropogenic effect. A paragraph should be added to discuss the importance of combined effects and should address caveats about the fact that we cannot separate linear trends from natural variability. It is becoming increasingly apparent that reanalysis should also include reanalysis of the chemical state of the atmosphere. There is great need for a skillful reanalysis using a global air quality model, for various reasons: understanding of aerosol-climate interactions, understanding of global transport of air pollution, provision of boundary conditions for regional photochemical simulations, etc. This issue should be addressed in the introduction and throughout the report as applicable.

Review of Individual Chapters 17 Specific remarks: There are inconsistencies in the use of italics/bolds throughout the text. L484: “variable for a specific time and” should be replaced with “variable or set of variables, for a specific time, level, and” L489 and(Ghil and Robertson, 2002; Yeh and Kirtman, 2007) throughout: Improve legibility of reproduced figures. L493-494: “and surface station locations” should be replaced with “and a subset of surface station locations with observations” L500-501: “there are fewer upper-air observations than surface observations, and that there is also” should be replaced with “there is” (since only a subset of surface observations are plotted, such that the number of visible surface and upper air observations are similar, this is not a good place to mention this) L507: Add at end of sentence: “…using data that for the most part had already been analyzed earlier for weather forecasting purposes.” L528: “attribution” should be replaced with “attribute” L529: “Webster’s II” is not the “author”. L593: Delete “becoming increasingly” L595-600: This is hardly a broad list of various major areas of meteorological research. All fall into the single major category of climate change and variability. If it’s difficult to find major areas that don’t use reanalysis data, as the draft states, it can’t be too hard to find more than one major area that does. L609-610: “one measure of uncertainty” should be replaced with “a measure of part of the uncertainty” L610: “phenomena” should be replaced with “identifying phenomena” CHAPTER 2 REANALYSIS OF HISTORICAL CLIMATE DATA FOR KEY ATMOSPHERIC FEATURES General remarks: The committee feels that the chapter contains much useful material that serves to fulfill the mandates of the prospectus. It also feels that the chapter can be improved in several respects. First, the chapter must be revised to make it easier to read. It also

18 Review of CCSP SAP 1.3 assumes the reader to be a technical expert, and should either have a summary for non- technical reader, or clearly state at the beginning that the chapter is intended for a technical audience. The discussion of temperature trends and reanalysis needs to be improved. This discussion should include a description of the usual climate data sets, surface temperature and precipitation and an evaluation of present capabilities of reanalysis. For example, the Observing System is mentioned in the captions of Fig. 1.2 and Fig. 2.4 and on pages 37, 40, 46. Observations play a crucial role in the reanalysis process and the text would benefit from the addition of a synthesis table where all different types of observations used in reanalysis would be listed, along with notes on their spatial and temporal coverage and also the year that they started being included in the model. This would be a very useful for the general public and the data user. The committee believes that the authors should emphasize that long-term climate data sets derived directly from surface and/or satellite observations (such as those for surface air temperature, precipitation, atmospheric water vapor, etc.) will continue, at least for the near-term (5-10yr), to be the main tool for quantifying decadal and long-term climate changes. The authors should also emphasize that reanalysis data will continue to be used largely for studying atmospheric processes and synoptic to interannual variations. Thus, the climate community should continue to invest in producing, updating and maintaining these long-term climate data sets, which should be assimilated into the reanalysis data (e.g., on a daily or monthly basis). The section on Key Findings in Chapter 2 contains several contradictions. The value of reanalysis is promoted, but then several paragraphs outline the uncertainties in everything from the models themselves to the quantity and quality of the underlying observations. A more reasonable finding might have been to acknowledge that reanalysis is a work in progress and then extol the potential that reanalysis offers to describe the current state of the atmosphere in 3 dimensions and to improve the predictability of climate change. This chapter should also emphasize that atmospheric reanalysis should try to make better use of historical records of surface observations (of temperature, precipitation, humidity, pressure and winds) from land stations and marine platforms. This will enable the reanalysis to be truly useful for climate change analyses. The reanalysis can make use of the existing climate analysis data (such as those for daily or monthly air temperature, pressure, humidity, precipitation, and cloudiness), instead of going back to the raw observations and trying to repeat the data quality control processes already done by the climate analysis people. The chapter would benefit from some discussions or reference to other parts of the report on how to improve the quality of the reanalysis data for long-term climate change studies. For example, it would be helpful to make suggestions on how to improve the reanalysis temperature and precipitation in future versions of reanalysis. Some expert opinions on the technical aspects of reanalysis are needed in addition to the mostly user aspects presented in the report. For example, the ERA-40 and JRA-25 have already

Review of Individual Chapters 19 applied many techniques to correct biases and homogenize surface and satellite data. The report should address these technical aspects and what the U.S. efforts should do in future reanalysis projects. The authors should note the need for systematic treatment of representation error, i.e., the variability in the observations due to those physical causes for which the model cannot account. It is possible to establish confidence limits on the hypothesis that the final product be consistent with its underlying assumptions about model and observation errors, and it is essential that these statistical tests be performed. Any statements of confidence derived from such non-parametric tests should be consistent with the statements of confidence that would be obtained from conventional significance tests on time series with Gaussian statistics. One finds, for example, that when a trend is estimated with ordinary least squares regression and tested for significance, the p-value of a zero trend is 0.0 rather than the 0.5 that the non-parametric test is inferred to yield. In general, the confidence levels defined for the non-parametric test seem much too high. Differences between observed and modeled values should, within confidence limits, be consistent with the error models used in the reanalysis. In filtering schemes such as the Kalman filter, the sequence of analysis increments (also called "innovations") should be white. In variational methods, the ending value of the cost function should be a random variable with chi-square distribution. In filtering schemes such as variants of the Kalman filter, a well-known quantity derived as a quadratic function of the innovations should be subjected to the chi-square test. The model biases are very important and need to be considered when interpreting trends, evaluating trend significance, and attempting attribution. For these reasons, model biases should perhaps be given a paragraph in this chapter, where they would be defined and briefly discussed, or this could be included in a text box. The committee finds that the report is relatively silent on the developing coupled data assimilation CFS reanalysis reforecast project. The report should acknowledge that this project is in the process of being launched and it should also mention the development of the Ensemble Kalman filter technique used by Geophysical Fluid Dynamics Laboratory (GFDL). The committee is concerned that the document suggests that reanalysis was used to help understand the surface temperature/precipitation trends over North America (specifically using the 500 mb heights) because the reanalysis data does not characterize all regions correctly. This should either be explained in the document or another example used. Specific remarks: The committee notes inconsistencies in the use of italics/bolds throughout the text. There are also many editorial errors that need to be corrected. For example, many of the papers cited in the text are not in the Reference list (e.g., Folland et al. 1986, cited on page 60, Table 2; Straus and Shukla 2004, cited on p. 67, line 1341; Mo et al.

20 Review of CCSP SAP 1.3 1998, line 1350; Feldstein et al. 2002, 2003, p. 67, line 13341; etc.). Also, some of the figure captions need to be corrected (e.g., the citation in the caption for Fig. 2.9 is incorrect). Page 34: when listing all the key findings, it would be useful to also include reference to the main section(s) that these findings refer to. Page 37, section 2.1.2: ‘analysis’ is introduced. The second sentence refers to “accomplishing this purpose”. This paragraph should be rephrased. L711: There is a double (period) “surface..” at the end of the sentence. L756: Delete “Nevertheless” L892: “physical relationships” - useful to provide one or two examples of such relationships, and how they provide “memory” for observations. L899: this paragraph is a little confusing. “Initial atmospheric conditions” were introduced in ~1970s, but what were the numerical weather predictions systems using before that? Also, the “detailed quantitative analyses” are obtained by the use of numerical models, not directly by using initial atmospheric conditions. I just find this paragraph unclear. L922: “evolution” should be replaced with “evolution potentially” L981: “of the quantities that” should be replaced with “of which quantities”; “those that” should be replaced with “which” L957: the bias should be defined more clearly (bias between …) L1030: “in principle” should be replaced with “ideally” L1030: “can forecast or simulate all aspects of the atmosphere”. “all” is very strong and should be replaced by “many”. L1052: Delete “about” Figure 2.7 caption, L1100: “The top panels are form the observations”, should be replaced by “The top panels are from the NCEP NCAR R1 observations”. It should also be kept in mind that reanalysis fields are not observations. L1126, 1127: an example would be very useful here. L1137-1139: a diagram would be very useful in making the point here. L1172: “new parameter estimation techniques” - since these methods are mentioned, perhaps the text should also give a very brief description of such techniques.

Review of Individual Chapters 21 Table 2.2, page 60: What is meant by link between atmosphere and ocean and how is this assessed? Consistency column should include citations. L1217: Concerning the title listed in the figure “Impact”, it would be better to show the percentage of explained variance to provide more of an impact. Figure 2.8: What season? What is the contour interval for the heights? Why is correlation used? Why not use regression to indicate amplitude? L1325: “reanalyses” should be replaced with “global reanalyses” L 1407 year missing in Madden and Julian citations. L1520: Delete “our” L1525: “they” should be replaced with “reanalyses” L1557-1562: Why is AMIP the only approach? What about pace-maker? What coupled efforts? Predictions? L1822: The GISS plot needs to be updated. Page 95: The need to deal with systematic errors in observations and the introduction of false trends into observations by changes in instrument systems both reflect deficiencies in the form of the measurement functional, the statistical model of measurement errors, or both. A similar comment applies to the inhomogeneities noted on lines 1927-1929, p97. L 1953-56: This is an encouraging example of diagnosis of systematic errors at their source. Page 99, Figure 2.19: The hemispheric asymmetry in number of observations is probably understated by this figure since all panels are for the austral summer. L1987-1990: Not clear. Wouldn’t one normally expect a data compilation covering 30 more years to have much more data in it? L1993: This error should be cast in terms of errors in some familiar quantity like thermocline depth, and compared to other sources of error. Page 100, Figure 2.20: Note from this figure the episodic nature of ocean observations: Note that the number of observations decays sharply after 1973, and again after 1992. Is there a specific reason for these changes in observational coverage? Compare this to figure 2.11 that indicates for the atmosphere that, at any given latitude, the number of observations increases with time.

22 Review of CCSP SAP 1.3 There are no counterparts of GARP or FGGE in operational oceanography. For this reason, analysis of the ocean will lag analysis of the atmosphere for some time to come. Figure 2.20 points out, if only indirectly, the scarcity of observations of the deep ocean. Diagnosis of the influence of the deep circulation on climate must remain in the realm of speculation. The influence is probably not among the greatest on decadal time scales, but errors in estimates of the deep circulation will not be diagnosed for some time. L2041-2044: What is a "reanalysis observation?" Please explain "merged dataset" L2064: What has been (or will be) the tangible benefit of improved reanalysis resolution for climate studies? L2071-2074: Delete sentence. Not relevant to paragraph on false trends. L 2072-2074: Formulation of forecast error models is particularly important in this context. L 2076-2089 The point of this paragraph is uncertain, particularly last sentence. One-way coupling is confusing language – does this mean forced ocean simulations with no feedback onto the atmosphere? L2079-2080: The question of how to do one-way coupling is far from settled. Two-way coupling is harder still. L2086-2087: Also, fully coupled systems have fairly coarsely resolved ocean model components due to resource limitations. L2089 needs to mention coupled activities at EMC, GFDL and JPL. Page 109: Besides assumptions 1) and 2), most data assimilation systems make assumptions of near linearity and Gaussianity Pages 109-110: The authors are correct in pointing out the need for bias correction and for better covariance models. L2130-2131 gives the impression that the state of the art of correction of systematic errors is more advanced than it actually is; methods for doing this are under development, and there are few examples. L2139-2149 more text about ongoing coupled efforts (e.g., at EMC, GFDL and JPL) needs to be included.

Review of Individual Chapters 23 APPENDIX A DATA ASSIMILATION General remarks: It should be noted that data assimilation is an exercise in the calculation of conditional probabilities. Assumptions of Gaussianity reduce the explicit evaluation of conditional probability to formulas involving covariances. It should be emphasized that all data assimilation methods are based on statistical error estimates. Specific remarks: L2847: "Observational increments" are known as "innovations" in the engineering literature, and are occasionally referred to as such in the data assimilation literature. L2854: Quadratic cost functions can be constructed without assumptions about the underlying distributions, but interpretation of the results is not so straightforward as it is in the Gaussian case. L2863-4: Straightforward implementations of the Ensemble Kalman Filter cannot incorporate future data; that's why it's called a filter, according to standard terminology in time series analysis. The analysis produced by 4DVAR at any given time can be influenced by observations at subsequent times. This property defines 4DVAR as a smoother. APPENDIX B AN EXAMPLE OF SOME OF THE OUTPUT FIELDS FROM REANALYSIS Specific remarks: L2906: replace “/s” with “1/s” L2942, 2959, 2954, 2958: clarify units. L2976, 2984: clarify “layers” L2975: for consistency, replace “m**2” with “m2”.

24 Review of CCSP SAP 1.3 CHAPTER 3 ATTRIBUTION OF THE CAUSES OF CLIMATE VARIATIONS AND TRENDS OVER NORTH AMERICA DURING THE MODERN REANALYSIS PERIOD General Remarks: The committee understands that the goal of chapter 3 is to document how reanalysis is currently an essential tool for rigorous attribution of regional climate variations, and could be used in the future for climate-change attribution. This message should be stated clearly at the beginning of chapter 3 in order to provide a bridge with the previous chapter for the multiple intended audiences. For example, while the chapter will be of considerable interest to climate scientists, as it is presently written, the committee is unsure what policy makers could take away from it. This chapter relies heavily on original, non-peer reviewed work. The authors should emphasize that although much of the work in this chapter has not been done before, that they are drawing on previous work (especially in sections. 3.4 and 3.5, which are primarily a review of the relationship between drought and climate shift). The authors should clearly identify what is their own original work. In general the committee believes that the authors should rebalance their work by including more the peer reviewed literature. The authors are encouraged to add relevant references, especially with respect to climate variations that are included in the attribution sections. The chapter would be greatly improved by referring to a detailed appendix that explains the methodology of the non-peer reviewed material, such as how smoothing was accomplished, identification of which years were generated by original research or if details can be obtained from a website; how the “obs” figures were constructed, how the PDSI was computed, how the “natural variability” time series were constructed. The committee believes that this Appendix should be peer reviewed. This peer review could be conducted either prior to or during the public comment period. Some suggestions follow to help improve this chapter. It would be useful to document studies that have made use of reanalysis data for analyzing climate shifts. The so-called transition around 1976 might be an instructive example. On L4405-4409 the authors state “There is evidence of abrupt changes of ecosystems in response to anthropogenic forcing that is consistent with tipping point behavior over North America (Adger et al. 2007), and some elements of the physical climate system including sea ice, snow cover, mountainous snow pack, and streamflow have also exhibited rapid change in recent decades (IPCC, 2007).” It would valuable to summarize and critique these lines of evidence, especially in view of the difficulties in detecting purely meteorological shifts. Has reanalysis data been used in an auxiliary role? What might be its likely potential? Quantifying the ability of reanalyses to reproduce droughts ought to be a key part of this report. The authors state on page 214: “The indications for drought itself, such as

Review of Individual Chapters 25 the PDSI or precipitation, are not derived from reanalysis data, but from the network of surface observations.” Why is this so? How can one have objective confidence in determining the mechanisms of drought from reanalyses if one doesn’t even know if the reanalysis captures the drought? This should be addressed. From a statistical point of view, it would be worth stressing the extreme difficulty of detecting and defining a rapid climate shift from very short noisy climatic time series. It may be helpful to distinguish between the statistical signal detection aspect of the definition and the physical aspects. Statistical significance has to be assessed against a red-noise null-hypothesis. The intractability of the purely statistical problem makes it imperative to consider the physical plausibility of any “shift” detected, and it would be here that suitable reanalysis data could play a potentially important role. This point might well be illustrated by a figure or a table. There are some poorly-worded (and therefore incorrect) attribution statements, such as “the spatial variations in observed North American surface temperature change since 1951 are unlikely due to anthropogenic forcing alone” (p. 178). The statement should be revised, since SST’s and natural variability are known to influence spatial variations of North American surface temperatures, so it is impossible (or at best exceptionally unlikely) that the spatial variations of the change are due to anthropogenic forcing alone. In formulating these attribution statements, the authors have ignored sources of error in the observed record, such as observation uncertainty (changes in siting, instrumentation, etc.); analysis uncertainty (as discussed in the example shown in Ch. 2 of differences among analyses); and sampling bias (carrying out a trend analysis partially because the last 10 years have been so unusual). In essence, the authors have neglected the uncertainty of the observed trend and the uncertainty that models have as much or more century-scale natural variability as the real climate system. These factors should be addressed. While the committee appreciates the need for a non-parametric confidence test, the standard for “detecting a change” is so weak that an observed fall of 0.05 degrees C would merit an inference of “moderate confidence” that an upward change had been detected. Any statements of confidence from non-parametric tests should include reference to the results of application of such tests to a well-behaved time series, to which conventional tests of statistical significance could be applied. This would facilitate critical evaluation of the level of confidence that a change had, or had not been detected. Spatial variations in summertime surface temperature change are unlikely the result of anthropogenic forcing alone. This chapter is predominantly oriented toward treating drought as an “event”, so it fails to discuss the importance of long-term local precipitation trends in altering the rainfall PDF and thereby producing more or fewer drought events of greater or lesser severity. In addition, Section 3.5.4.2 fails to consider/discuss any anthropogenic influences besides greenhouse gases, such as changes in irrigation,

26 Review of CCSP SAP 1.3 deforestation/reforestation, and radiative and microphysical effects of aerosols. The analysis of Indian/WPac SST’s in 3.5.4.2 seems to leave unconsidered the likely relevance of SST changes in that region independent of other changes (Rossby wave forcing) compared to SST changes occurring simultaneously everywhere (no Rossby wave forcing, but strong anthropogenic influence). Specific remarks: The source of “obs” analyses should be identified. More complete definitions of Atmospheric Model Intercomparison Project (AMIP) and Coupled Model Intercomparison Project (CMIP) are needed throughout the chapter. The authors should avoid the tendency to “explain away” differences between observed and AMIP long-term trends, while ignoring other possibilities in instances where the observed and AMIP agree (for example p. 177, p. 192). In the Attribution summary of Chapter 3, the authors use italics to highlight “likely – unlikely” but the definitions of these terms are not explained. It would be helpful to refer back to the table in the preface with footnote, or add the numbers parenthetically and explain how these numbers were estimated. L4301-4303: “A retrospective assessment of [abrupt shifts] may offer insights on mitigation strategies that are consistent with the known frequency and severity of impacts related to rapid climate shifts.” Due to their rarity, any retrospective analysis of impacts would be very difficult. This sentence understates its complexity. L4356: Do the authors mean “proxy” climate records rather than historical? L4361: “3.4.4.1 Abrupt Natural External Forcings Since 1950” Are these external forcings (aerosols, GHGs etc) included in any of the reanalyses? The discussion of abrupt natural external forcings such as volcanic eruptions needs to be framed in the context of reanalyses. Specifically, the report should clearly state which “external” forcings, including natural and anthropogenic aerosols and greenhouse gases, are included in current reanalyses, and what are the potential implications for the role of reanalysis datasets in attribution. The question of uncertainties in estimating these forcings also needs to be addressed. Implications and recommendations for future reanalyses should also be given. L4432-4433: “Some rapid climate transitions in recent decades appear attributable to chaotic natural fluctuations.” Again definition of what is meant by a “rapid transition” is problematic: one person’s transition is another person’s climate noise. A “wave-particle duality” analogy between episodic and oscillatory views of atmospheric variability has been discussed recently, and this may be helpful here to the intended audience (Ghil and Robertson 2002).

Review of Individual Chapters 27 L4446: An “apparent” rapid transition might be more accurate. Page 219: Mentioning “billion-dollar weather disasters” should not be done without discussing increased vulnerability and inflation. L3034: “2 C warming” Is this a linear trend? Is it per century? The number should be given. L3042-3045: See comment for L3908 and L3916. L3056-3061: See comments for L3908 and L3916. L3071 & 3075: The terms “short-term” and “long-term” are not defined anywhere. L3087: “may be” should be replaced with “are” L3093: “record-setting 2006 US warmth”: This term should be used with caution. The committee suggests “unusual” instead. L3096: Delete “the source for” L3117: “gold-standard”: This term is confusing and potentially ambiguous. More explanation is needed. Does this imply that this standard is something that is assumed to be error-free by definition or does this mean the best available measurement? {This last sentence isn't clear in itself (distinction between what and what?), and probably isn't necessary.} L3133: “immediate cause(s)”: The committee disagrees with this terminology. The immediate cause of a temperature change pattern is some combination of changes in advection, land surface characteristics, cloud cover, etc. A teleconnection is at best an intermediate cause. L3139: Figure 3.1 does not have a clear flow and the relationship of the graphs to the rest of the figure is unclear. The figure also brings up the potential for confusion between the term “attribution” defined in the broader sense in this report, and its narrower but, by now, familiar usage in the climate-change community. This needs to be kept in mind throughout the document. L3222: Use consistent method of citing IPCC reports. L3370: See 1217. L3377-3378: Some text is missing here. L3458: “jointly” has a specific statistical meaning that is probably not intended here. L3793: Figure 3.6 and many figures that follow use non-conformal projections. This should generally be avoided, but it should especially be avoided here because the

28 Review of CCSP SAP 1.3 spatial average of the plotted field and how much different areas contribute to that spatial average are very important. L3871-3872: Delete “the observed”; replace “detected,” with “detected in observations,” L3875-3877: The text gives the impression that where the observed pattern agrees with models with greenhouse gas forcing, the models are correct, and where the observed pattern disagrees, the models are deficient. No, the models are deficient everywhere, and the disagreement in the Southeast US suggests that part of the agreement elsewhere may be fortuitous. L3907-3908: Given what we know about the climate system, it is impossible that any sub-century spatial variations in observed surface temperature change could be due to anthropogenic forcing alone. The authors must mean to say something different, such as it is unlikely that the spatial variations are due predominantly to anthropogenic forcing. L3915-3918: Given what we know about the climate system, it is certain that any sub- century spatial variations are influenced by observed SST variations. The authors must mean something different, such as, it is likely that the spatial variations are predominantly associated with sea surface temperature variations. L3924: “much” should be replaced with “many” L3951: “in producing should be replaced with “to produce” L3958: Delete “explaining” L4000-4002: See comment regarding L3907-3908. L4034-4039: It is equally true that the U.S. also experienced warm conditions during the end of the 20th Century, and it is partly for that reason that the 1951-2006 observed trends are not smaller. The passage could perhaps be justified if the trend starting at 1951 is less than one would obtain starting earlier or later, however this is not the case. It appears that any trend starting between 1925 and 1950 would yield an even lower trend. L4081-4087: If “natural cooling” can explain the discrepancy with anthropogenically- forced warming in the Southeast, then it is equally plausible that “natural warming” can explain part of the apparent agreement with anthropogenically- forced warming elsewhere. In fact, since the cooling is related to teleconnection patterns, there must be natural warming elsewhere. L4097: “20004” should be replaced with “2004” L4137: “High” should be replaced with “Very high” ???

Review of Individual Chapters 29 L4140-4144: See comments for L4034-4039. Texas, for example, has smaller trends 1921-2006, 1931-2006, and 1941-2006 than 1951-2006. Of all the available starting dates for trend assessment, 1951 produces a trend estimate that falls close to the median of the other estimates. L4144: “mid-spread” should be replaced with “wide-spread” L4208: Spell checker Freudian slip. L4241: “decadal-like”? L4466: “appearnence” should be replaced with “appearance” L4494: “phenomena’s” should be replaced with “phenomenon’s” L4509-4510: Moisture demand from PET exceeding supply from precipitation is the definition of a “dryland”, not a “drought”. Otherwise, the Colorado Basin would be in drought even in the wettest year. L4553: “and” should be replaced with “and an average of” 4554: “drought” should be replaced with “severe drought” L4554: Delete “index”. L4558: In Fig. 3.20, what is the red line? L4588: Droughts (6) and (7) are a single event. L4596: The fractional “variability relative to the average precipitation should be shown in Fig. 3.22”, because this is the key parameter in the discussion. L4610: “conditons” should be replaced with “conditions” L4651: “influence for the” What does this phrase mean? L4656-4659: Is this in reference to the western US or the northwestern US? It seems to hop around. L4700: “upstream” should be replaced with “downstream”? L4712: “have also been linked” should be replaced with “have been linked to” L4769: What is the bottom panel of Fig. 3.23? L4783-4794: The discussion is missing a logical link: a demonstration that the absolute magnitude of the Indian Ocean temperature is the factor that matters, rather than an anomaly with respect to surrounding SSTs. The correlation found by Lau et al. might be solely due to short-term variability, based on the information presented.

30 Review of CCSP SAP 1.3 L4804: This line of text is awkward. L4872-4873: If there is an initial soil moisture deficit, the drought has already started. Change to “…subsequently amplified by local soil moisture conditions, and in some…” L4877: Delete “an” Page 137 – move last bullet to second bullet on page 136. L3094-3103: This text appears to be a discussion of the prediction/predictability question. How does attribution differ from predictability? L3149 “places” Figure 3.1 has too many arrows and the point being made is unclear, however the text on page 3274-3284 is quite clear. Figure 3.2 contour interval. L3392: It is not clear what aspect of figure 3.1 is referenced. Page 170: Use of AMIP does not include changes in forcing other than SST. May explain why magnitude weak, but also a weakness in the comparisons. L3822-3825: This assessment seems incomplete for the CMIP models. L3852 “And I remains unclear how SST” should be “And it remains unclear how SST.” L4018-4022: This needs to be stated earlier also. L4101: “series in” should be changed to “series is” L4459: Latif and Barnett not the best reference here. L4493: This statement does not appear to be accurate; we do prediction all the time without understanding the mechanisms. Figure 3.22 shouldn’t a ratio be plotted? L4761: Yeh and Kirtman (2007) reference should be cited L4763-4788: non-Gaussian behavior assumed – no change in La Niña? Figure 3.23: The committee does not understand the bottom panel.

Review of Individual Chapters 31 CHAPTER 4 RECOMMENDATIONS General remarks: The title provides a nice paradigm for the chapter as it suggests that issues, opportunities and recommendations will be discussed. The committee finds that the opportunities and recommendations are apparent in the organization and presentation of the chapter material; however the connection between reanalysis and attribution needs to be strengthened. The committee offers the following suggestions to improve this connection: Introduce this chapter with a restatement of what the scope of this SAP is, why the scope has been so defined (what was seen to have highest priority and why; what it was possible to do at the time, what was not done and why), and describe the motivation for this SAP. Refer back to text added to chapter 1 (Introduction) in which the connections between reanalysis and attribution are described; highlight the steps taken/model process and use examples from chapter 3 (perhaps even show one of the figures) to highlight findings/conclusions drawn in this SAP. Mention that the goal of the chapter is to provide high-level recommendations aimed at improving the scientific and practical value of future climate analysis and reanalysis. This discussion should clearly state that these recommendations will help reduce uncertainties in climate attribution and will develop ways of realizing the benefits of reanalysis data in supporting policy decisions. The introduction of Chapter 4 needs to better explain variability and trends. The meaning of attribution should be clearly defined and should be discussed/interpreted in a probabilistic manner. This will further enhance the education function of the document. The introduction should also mention that although some researchers prefer the use of all available data in reanalysis, there is a basic, unavoidable need for verification of reanalyses using an independent data set. L5838-5843: Questions such as “What was the cause for the Nation’s record setting 2006 warmth?” are ill-posed. Since the intended audience of this report includes policy-makers, the report misses an opportunity to explain why such questions are ill- posed. As mentioned somewhere else in the report, the key policy-relevant questions are: “How much has the probability of warmth such as 2006 changed, and would this probability be expected to undergo further change in the future?” or a related question: “What is the net contribution of anthropogenic forcings to the 2006 warmth, and what is the marginal contribution of each forcing?” Both questions are relevant for adaptation, and the latter question is also relevant for mitigation. As mentioned earlier, reanalysis should also include reanalysis of the chemical state of the atmosphere. A skillful reanalysis using a global air quality model is

32 Review of CCSP SAP 1.3 necessary for various reasons: understanding of aerosol-climate interactions, understanding of global transport of air pollution, provision of boundary conditions for regional photochemical simulations, etc. This issue should be addressed in the final chapter as well as in the introduction. Specific remarks: Some of the recommendations need further clarification. The designation of recommendations by “R” and “A” for reanalysis and attribution, respectively, further enhances the separation/discontinuity between these techniques. The committee suggests that all recommendations that blend reanalysis and attribution issues be combined. The biggest challenge with reanalysis is the model and the authors are silent on the fidelity of the model. A discussion of how the model should assimilate temperature and precipitation should be included in the recommendations. R2 needs to be more specific. What does “optimized for climate purposes” mean? Does this mean detection and attribution? There should be some discussion of the fact that there is a range of climate purposes, and different purposes demand different, incompatible, reanalysis configurations. For example, if one wants a trend-free reanalysis, one uses a sparse subset of the current data, but if one wants the most accurate representation of the atmospheric state at any given time, one uses as much data as possible. This discussion would tie reanalysis and attribution sections together better. R6 states that it is beneficial to go beyond present ad hoc project efforts to a more coordinated and effective national program in climate analysis and reanalysis. How this approach would be beneficial to improve coordination is unclear from this recommendation. What is the scope of this coordination? What is the rationale behind this recommendation? Would the goal be to coordinate better, have a better use of existing resources? There is a US national interest to continue to do reanalysis – does this recommendation mean that a program should go beyond the current program? Will this approach make better use of existing climate data sets? For example, surface trend problems are evidence that we need to do better – will this be accomplished through a national program? Capability may be a better word that does not necessarily imply new infrastructure. L5400 “… efforts should include a focus on …” L5454: Recommendation A1: Is Program for Climate Model Diagnosis and Intercomparison (PCMDI) addressing this need?

Next: References »
Review of the U.S. Climate Change Science Program's Synthesis and Assessment Product 1.3: Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The U.S. Climate Change Science Program is in the process of producing 21 draft assessments that investigate changes in the Earth's climate and related systems. These assessments are designed to inform decisionmakers about the scientific underpinnings of a range of environmental issues, such as models of past climate conditions. This National Research Council book reviews one of these assessments, Synthesis and Assessment Product 1.3 "Reanalyses of Historical Climate Data for Key Atmospheric Features: Implications for Attribution of Causes of Observed Change." The committee commends the authors for clearly stating their goals and their intended audience and for their fidelity in following the prospectus. However, the current draft needs revision to better link reanalysis and attribution. In addition, the document needs to better explain how reanalysis fits into climate science and include a general description of how climate science is done and how the models, observations, and theories are related to the ultimate goal of reanalysis, especially for the benefit of non-specialists.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!