 8. A Program for Action by the U.S. Committee for the Global Atmospheric Research Program. Section 8. A National Climatic Research Program. Part 3. The Plan. Our recommendations for the planning and execution of the climatic research program outlined above are given here in terms of what we believe to be the appropriate sub-programs, the necessary facilities and support, and the desirable timetable for both the short-term and long-term phases. We also offer some observations on the program's administration and coordination, although we recognize that a program of this scope will require much further planning and that the support and cooperation of many persons and agencies will be necessary for its successful execution. Sub-program identification. In a program as broad as that envisioned here, it is convenient to think in terms of a number of components or sub-programs, each concerned with a specific portion of the overall effort. Such sub-programs also represent the necessary division of effort for the practical execution of the program. The NCRP itself should ensure the coordination of the various sub-programs and maintain an appropriate balance of effort among them. Climatic Data Analysis Program, or CDAP. In order to promote the extensive assembly and analysis of climatic data outlined above, we recommend that a Climatic Data Analysis Program, CDAP, be established as a sub-program of the NCRP. The purposes of the sub-program are to facilitate the exchange of data and information among the various climatic data depositories and research projects, and to support the coordinated preparation, analysis, and dissemination of appropriate climatic statistics. Climatic Index Monitoring Program, CIMP. In order to promote the monitoring of the various climatic indices outlined above, we recommend that a Climatic Index Monitoring Program, CIMP, be established as a second sub-program of the NCRP. The purposes of this sub-program are to support and coordinate the collection of data on selected climatic indices and to ensure their systematic dissemination on a timely and sustained basis. Climatic Modeling and Applications Program, CMAP. In order to promote the construction and application of the climatic models outlined above, we recommend that a Climatic Modeling and Applications Program, CMAP, be established as a third sub-program of the NCRP. The purposes of this sub-program are to support and coordinate the development of a broad range of climatic models, to support necessary background scientific research, and to ensure the systematic application of appropriate models to the problems of climatic reconstruction, climatic prediction, and climatic impacts. Facilities and Support. The availability of adequate facilities and support and the design of coordinating mechanisms are necessary to carry out the various sub-programs recommended for the NCRP and should be given careful consideration. Of primary importance are the roles of climatic data analysis facilities and research consortia, the needed high-speed computers, and the required levels of funding. Climatic Data Analysis Facilities. To assist in the implementation of both the Climatic Data Analysis Program, CDAP, and Climatic Index Monitoring Program, CIMP, we recommend the development of new Climatic Data Analysis Facilities at appropriate locations, including linkage to the various specialized data centers and climatic monitoring agencies by a high-speed data transmission network. Such facilities should have access to machines of the highest speed and capacity available and be staffed by specialists in data analysis, transmission, and display. Collection of certain climatic data by a group of specialized facilities appears more desirable than does collection of all data by a single centralized facility. We envisage these facilities as performing the bulk of the recommended CDAP. This would include the inventory, compilation, processing, analysis, and documentation of both conventional and proxy climatic data. Close working cooperation is envisaged with specialized data depositories. For conventional atmospheric and oceanic data, these include NOAA's National Climatic Center and National Oceanographic Data Center. For satellite data, the National Environmental Satellite Service. For glaciological data, the Geological Surveys Data Center A in Tacoma. For ice core data, the Army's Cold Regions Research and Engineering Laboratory. For marine cores, Columbia University's Lamont-Doherty Geological Observatory. And for pollen and tree ring data, the universities of Wisconsin and Arizona. We also envisage the data analysis facilities as playing a prominent role in the CIMP and in the processing analysis and dissemination of the results on as nearly a real-time basis as possible. Certain of the facilities could serve as global climatic watchdogs and might have a resident scientific staff to perform diagnostic research as appropriate. Climatic research consortia and manpower needs. We envisage the broad range of research and analysis recommended here as being best performed by a number of institutions and groups. This is desirable in order to ensure the breadth of viewpoint and diversity of approach necessary in a problem as close to the unknown as is climatic variation. An attempt to carry out all the recommended activities and research by a single institution would in any case be a practical impossibility. Research on climate and climatic variation at the present time is principally performed in governmental laboratories and in a variety of research projects in universities and other institutions, usually with the support of the federal government. Chief among the laboratories concerned with elements of the climatic problem are NOAA's Geophysical Fluid Dynamics Laboratory, NOAA's National Environmental Satellite Service and Environmental Data Service, NSF's National Center for Atmospheric Research, and NASA's Goddard Institute for Space Studies. More specialized research on problems related to climate is also performed by the U.S. Geological Survey and by the operational services and laboratories of the U.S. Army, Navy, and Air Force. Many of the climate related research projects in universities and other institutions are supported by the National Science Foundation through its programs for atmospheric, oceanic, and polar research by DOT's Climatic Impact Assessment Program and by ARPA's Climate Dynamics Program. These include the various quaternary research groups, geological and oceanographic laboratories, numerical modeling groups, and polar studies and environmental institutes. Each of these efforts makes a contribution to the National Climatic Research Picture and they represent a valuable reservoir of experience and talent. In order to promote greater cooperation and exchange to ensure an appropriate balance of effort and to give such research the needed stability and coherence, we recommend that efforts be made to coordinate present research more effectively as parts of a National Climatic Research Program. We believe that this can be achieved best by the formation of cooperative associations of existing climatic research groups and the initiation of whatever new research efforts may be required as parts of such associations. We accordingly recommend the formation of a number of climatic research consortia among various research groups as appropriate to their interests with each such consortium having links to computing facilities of the highest speed and capacity available. Such research consortia would serve as valuable coordinating mechanisms for the broad range of climatic research envisaged in the climatic modeling and applications program, CMAP, as well as giving both coherence and flexibility to the NCRP as a whole. The present mode, NORPACS, and CLIMAP programs may serve as useful examples for such consortia. As the National Program develops, the possible need for new institutional structures or facilities should become clear. Our recommendations reflect the consensus that maximum use should be made of existing institutions while further consideration is given to the possible need for their expansion. Aside from institutional arrangements, however, we believe that the proposed research program unquestionably calls for the initiation and support of new mechanisms to provide an expanded base of appropriately trained scientific and technical manpower. We accordingly recommend that programs for technical training be developed and that both pre-doctoral and post-doctoral fellowships in the broad area of climatic research be established as soon as possible. Computer requirements The required access to high-speed computers has been alluded to several times in the discussion of the recommended program. Although it is difficult to make precise projections, the volume of data processing involved in the analysis and monitoring portions of the program alone indicate that a dedicated machine of at least the CDC 7600 class is required for the implementation of the CDAP and CIMP. The computer needs of the research consortia and of the other research groups involved in the modeling portions of the program are even more demanding in view of the variety of the needed climatic models and tests and the number and the length of the necessary climatic simulation experiments and applications. Our estimates of the NCRP's overall computer requirements are given in table 6.2 and call for a very significant increase over present levels of computer usage. If anything, these estimates may be too low. In its computer planning, NCAR has estimated a climate-related usage of several CDC 7600 units by 1980 for the needs of NCAR and the university community it serves, WM Washington personal communication. While the installation of the TIASC system at GFDL in 1974 will likely significantly raise their machine usage for climatic studies. As shown in table 6.2, it is estimated that climatic data analysis and monitoring will require the full-time use of at least one fourth generation machine and that climatic modeling and applications will require the full-time use of at least one fifth generation machine. We therefore recommend that machines of the CDC 7600 class be secured as soon as possible for the use of the data analysis facilities and the associated elements of the CDAP and CIMP and that planning begin for the acquisition of computers of the TIASC or ILLIAC-4 class for the use of the Climatic Research Consortia and the associated elements of the CMAP. It will also be necessary to provide broadband communication links among the various facilities and cooperating groups and with the climatic research community as a whole. Estimated Costs The cost of the recommended national climatic research program is difficult to determine accurately without a great deal of information on observational computing and support costs from the various agencies and institutions presently engaged in the many aspects of climatic research. Rather than seeking such detailed data, we have restricted ourselves to gross projections on the basis of estimates of the costs of present efforts. Our estimates of the expenditures for climatic research, not including the costs of instruments, observing platforms, or operational and service-related activities, are given in table 6.3. Our projections of the growth of these direct costs during the early phases of the program, i.e. to the year 1980, are shown in Figure 6.1. Along with the percentage increases over the preceding year, these estimates, of course, depend directly on the base figures that are used and are subject to further refinement. These figures are intended for order of magnitude guidance only and will require revision as the program develops. We recognize that the ultimate distribution of resources among the various sub-programs of the NCRP will be determined by the sense of priorities of the government and by the capabilities of the research community. The estimates shown in Figure 6.1 for the year 1980 are based on our perception of the needed increases over present efforts in the areas of data analysis and monitoring, CDAP and CIMP, especially those concerning satellite data and the monitoring of oceanic climatic indices. In the area of climatic modeling and applications, CMAP, the largest increases over present efforts are envisaged for the development and application of coupled global climate models and climatic impact studies. The relatively rapid growth rate during the program's third and fourth years are projected to include the acquisition of the necessary computers and networks. Overall, the recommended program calls for an approximate four-fold expansion of the support of research on climatic variation by the year 1980. The program's costs beyond this time are more difficult to estimate and will depend on the progress and opportunities developed prior to that time. It is useful to compare these cost projections with the direct and indirect costs of present GARP efforts and those of closely related programs. In fiscal year 1973, the direct GARP expenditures totaled $13.2 million, about 54% of which represented expenditures by the Department of Commerce and NASA, directed toward the improvement of weather forecasting, with the remainder expended by NSF for research on both forecasting and general circulation studies. Some of these costs are included in the estimates in Table 6.3, insofar as they can be identified as directed toward climatic research. The indirect costs associated with GARP amounted to $29 million in fiscal year 1973 and are not reflected in the present climatic research estimates. In addition to these efforts, there are other current programs that contribute to GARP and whose costs should not be overlooked. The implementation of the World Weather Watch, WWW, and its satellite system represented $1.5 million direct costs and $54.5 million indirect costs in fiscal year 1973, while systems design and technological development represented $2.4 million direct costs and $50.1 million indirect costs in the same period. The extent to which elements of the recommended CDAP and CIMP sub-programs of the NCRP may be considered as add-ons to such existing programs needs further consideration, as does the extent to which the future costs of GARP itself may be merged with those envisaged for the NCRP. Also in need of further study are the United States contributions to the costs of the various sub-programs recommended as part of the International Climatic Research Program, ICRP, described below, as well as the impacts of inflation. We also note that funds will be required for the training of additional scientific manpower in all aspects of the research program. Timetable and priorities within the program. We recognize the need for flexibility in a research program of this kind and that future technological and research discoveries may have important impacts on the direction of climatic research. In spite of these unknown factors, however, some consideration of goals and priorities is useful. Here we present our recommendations for the objectives of the initial phase of the program, 1974 through 1980, and the necessary sequence of planning activities for both these goals and those of the long-term phase, 1980 through 2000. Our recommendations for a coordinated international program are considered subsequently. The initial phase, 1974 through 1980. Once the decision is made to develop a national climatic research program, we recommend that planning begin immediately for the implementation of its component activities and sub-programs. Our specific recommendations for both the immediate and subsequent objectives during this phase of the program are shown in table 6.4 in terms of the data analysis, index monitoring, and modeling sub-programs identified earlier. Here our sense of relative priorities is given implicitly by the ranking into immediate and subsequent objectives. These timescales refer to the expected times of the achievement of first useful results with the recognition that initial development must in some cases begin earlier and that further development and application will continue later. This ranking also reflects a balance between the relative ease of accomplishment and the relative potential for initial practical usefulness. We believe that progress toward these subsequent objectives will require the support of all immediate objectives of the program with new priorities evolving as a function of achievement and opportunity. Relationship to the FGGE 1978 through 1979. The first GARP Global Experiment FGGE now planned for 1978 through 1979 is primarily an attempt to collect a definitive global dataset for use in the improvement of weather prediction by numerical atmospheric models. The potential value of these data for climatic research lies not so much in their display of seasonal and inter-hemispheric variations, valuable as that will be, but in the fact that many of the short period physical processes to be intensely measured or parameterized in FGGE are also important for the understanding of climate. Among these are the processes of convection, boundary layer dynamics, and the atmosphere's interaction with the surface of the ocean. The observational requirements during the FGGE call for measurement of the atmospheric temperature, water vapor, cloud cover, elevation, wind, and surface pressure together with the surface boundary variables of sea surface temperature, soil moisture, precipitation, snow depth, and sea ice distribution. To enhance their value for climatic studies, we recommend that these data be supplemented during FGGE insofar as possible by observations of the global distributions of ozone, particulates, surface and planetary albedo, incoming solar and outgoing terrestrial radiation, vegetal cover, and the continental freshwater runoff. We recommend that special observations also be made in conjunction with regional programs such as Norpax and Polex, which are expected to be in operation during the FGGE. The long-term phase, 1980 through 2000. The long-range goals and full-scale operation of the NCRP in the period beyond 1980 are portrayed in the upper part of Figure 6.2. During this period, the full interaction among the observational, analysis, modeling, and theoretical components of the program will occur, leading to the development of an operational global climatic data system, and, it is hoped, to the acquisition of an increasingly accurate theory of climatic variation. Although priorities cannot be set at such long range, the eventual practical payoffs of this program will be the determination of the degree to which climatic variations on seasonal, annual, decadal, and longer time scales may be predicted, and the degree to which they may be controlled by man. Administration and Coordination The administrative structure and coordination of the recommended program are the responsibility of the federal government and were not given extensive consideration. However, noting the concern with the problem of climatic variation in many parts of the government and the widespread participation of many governmental and non-governmental groups in climatic research, we believe that the program should be administered in such a way that the interests of all are effectively represented and coordinated. It is particularly important that the advice of the scientific community be used in the design and development of the major elements of the research program. Both the short-term and long-term goals of the NCRP are also shared by the International Climatic Research Program, ICRP, recommended below. The development of this international program should proceed in parallel with the NCRP and should be closely coordinated with GARP. The principal activities within GARP, up to the present time, have focused on the problem of improving the accuracy and extending the range of weather forecasts, and the United States contributions to GARP in particular have emphasized the development and use of numerical models for this purpose. These efforts are necessary steps in the development of an adequate modeling capability for both weather prediction and climate, and were they not already underway as part of GARP, they would have had to be undertaken through some other means as a prelude to the climatic research program. A coordinated international climatic research program, ICRP. Many of the efforts envisaged within the NCRP are of an obvious international character, and the degree to which these should be regarded as national, as opposed to international activities, is not of critical importance for our purposes. The important point is that there are international efforts now underway within GARP of direct relevance to the climatic problem, of which we note especially the International Study Conference on the Physical Basis of Climate and Climate Modeling held in Sweden in July and August 1974 under the auspices of the ICSU slash WMO, GARP Joint Organizing Committee. The recommendations and programs resulting from this and subsequent planning conferences should be closely coordinated with the U.S. national program. We offer here our recommendations for an appropriate international climatic research program, and some observations on how such a program might best be coordinated with GARP itself. Program motivation and structure. The observational programs planned in support of GARP offer an unparalleled opportunity to observe the global atmosphere, and every effort should be made to use these data for climatic purposes, as well as for the purposes of weather prediction. The climatic system, however, consists of important non-atmospheric components, including the world's oceans, ice masses, and land surfaces, together with elements of the biosphere. While it is not necessary to measure all of these components in the same detail with which we observe the atmosphere, their roles in climatic variation must not be overlooked. In addition to the fundamental physical differences discussed in Chapter 3, the problem of climatic variation also differs from that of weather forecasting by the nature of the datasets required. The primary data needs of weather prediction are accurate and dense synoptic observations of the atmosphere's present and future states. While the data needed for studies of climatic variation are longer term statistics of a much wider variety of variables. When climatic variations over long time scales are considered, these variables must be supplied from fields outside of observational meteorology. Thus, an essential characteristic of climate studies is its involvement of a wide range of non-atmospheric scientific disciplines. The types of numerical models needed for climatic research also differ from those of weather prediction. The atmospheric GCMs, which represent the ultimate in weather models, do not need a time-dependent ocean for weather forecasting purposes over periods of a week or two. For climatic change purposes, on the other hand, such numerical models must include the changes of the oceanic heat storage. Such a slowly varying feature may be regarded as a boundary or external condition for weather prediction, but becomes an internal part of the system for climatic variation. International Climatic Research and GARP In view of these characteristics, we suggest that while the GARP concern with climate is a natural one, as indicated above, the problem of climate goes much beyond the present basis and emphasis of GARP. Accordingly, we recommend that the global climate studies that are underway within GARP be viewed as leading to the organization of a new and long-term international program devoted specifically to the study of climates and climatic variation, which we suggest be called the International Climatic Research Program, ICRP. International Climatic Decades 1980-2000 We suggest that the observational programs of GARP, and especially those of the FGGE, be viewed as preliminary efforts later to be expanded and maintained on a long-term basis. In particular, we recommend that the special data needs of climatic studies be supported on an international scale through the designation of the period 1980-2000 as the International Climatic Decades, or ICD, during which intensive efforts would be made to secure as complete a global climatic database as possible. The general outline of the envisaged international program, ICRP, is sketched in the lower part of Figure 6.2, and the program's scientific elements are discussed in more detail below. Program Elements Climatic Data Analysis The main thrust of the International Climatic Program should be the collection and analysis of climatic data during the ICDs, 1980-2000. During this period, the participation of all nations should be sought in order to develop global climatic statistics for the broad set of climatic variables. We urge that these efforts include international cooperation in the systematic summary of all available meteorological observations of climatic value, including oceanographic observations in the waters of coastal nations. International Paleoclimatic Data Network, IPDN We urge the development of an international cooperative program for the monitoring of selected climatic indices and the extraction of historical and proxy climatic data unique to each nation, such as indices of glaciers, rainforest precipitation, lake levels, local desert history, tree rings, and soil records. Specifically, we recommend that this take the form of an International Paleoclimatic Data Network, IPDN, as a sub-program of the ICRP. The cooperation of such organizations as SCAR, SCOR, and the International Union for Quaternary Research, INQUA, should be sought in this program. The contents of these international observational efforts might possibly broadly follow those recommended for the U.S. national effort with modifications as appropriate to each nation's needs and capabilities. In addition, we recommend that the ICRP undertake the following. The international collection of special climatic data sets on such events as widespread drought and floods, and following major environmental disturbances, such as volcanic eruptions, programs to encourage international exchange of climatic data and analyses. Climatic Research Although cooperative research studies are desirable, we recognize that the large-scale numerical simulation of climate with CGCMs can now be carried out in only a relatively few countries. To promote wider international participation in climatic research, we therefore recommend that the ICRP include the following. Programs and activities to encourage international cooperation in climatic research, and to facilitate the participation of developing nations that do not yet have adequate training or research facilities. Internationally supported regional climatic studies in order to describe and model local climatic anomalies of special interest. The contents of these and other research activities of the ICRP might also broadly follow those recommended for the U.S. national effort with appropriate modifications for each nation's interests and capabilities. Global Climatic Impacts While all nations are tied in some fashion to the world pattern of climate, some are more vulnerable to climatic variations than others by virtue of their locations and the delicacy of their climatic balance. We therefore recommend that the ICRP include the following. International Cooperative Programs to assess the impacts of observed climatic changes on the economies of the world's nations, including the effects on the water supply, food production, and energy utilization. This should include the impacts of variations of oceanic climate for those nations whose economies are dependent on the sea, the cooperation of appropriate international agencies of the United Nations, and of other groups such as the International Federation of Institutes of Advanced Study should be sought. Co-operative analyses of the regional impacts of possible future climates. Such studies could be of great importance to many countries, particularly emerging nations making long-range policy decisions concerning the development of their resources. Program Support The question of the details of support of the ICRP was not dealt with. It seems clear, however, that an appropriate balance of effort should be maintained among ICRP, the various national climatic research programs, and other international programs such as the World Weather Watch, WWW, and the United Nations Environment Program, UNEP. The services of groups performing the function of the present GARP Joint Organizing Committee and its Joint Planning staff will also be necessary for the success of the International Program. In order to assist in the coordination of the ICRP, we urge that support be made available by the appropriate agencies of the United Nations on a scale commensurate with the breadth and importance of the problem. This should include a budget adequate for the effective international coordination of the ICRP on a scale significantly greater than that of GARP and on a continuing long-term basis. We also urge that scientific assistance be sought from the International Council of Scientific Unions in support of selected ICRP sub-programs. End of Section 8, Recording by Warren Coddy, Gurney, Illinois. Section 9 of Understanding Climate Change This is a Librebox Recording. All Librebox recordings are in the public domain. For more information or to volunteer, please visit Librebox.org. Understanding Climate Change, a Program for Action, by the U.S. Committee for the Global Atmospheric Research Program. Section 9, Appendix A, Survey of Past Climates, Part 1. Introduction The Earth's climates have always been changing, and the magnitude of these changes has varied from place to place and from time to time. In some places, the yearly changes are so small as to be of minor interest. While in others, the changes can be catastrophic, as when the monsoon veils or in seasonable rain delays the planting and harvesting of basic crops. On a longer time scale, certain decades have striking and anomalous characteristics, such as a severe droughts that affected the American Midwest during the 1870s, 1890s, and 1930s, and the high temperatures recorded globally during the 1940s. And on still longer time scales, the climatic regimes that dominated certain centuries brought significant changes in the global patterns of temperature, rainfall, and snow accumulation. For example, northern hemisphere winter temperatures from the mid-15th to the mid-19th centuries were significantly lower than they are today. The late 19th century represented a period of transition between this cold interval, sometimes known as the little ice age, and the thermal maximum of the 1940s. Some idea of the magnitude of the climatic changes that characterize the little ice age can be gained from a study of proxy or natural records of climate, such as those of alpine glaciers. As shown in Figure 8.1, as late as the mid-19th century, the termini of these glaciers were still advanced well beyond their present limits. The practical, as well as the purely scientific value of understanding the processes that bring about climatic change, is self-evident. Only by understanding the system can we hope to comprehend its past and to predict its future course. This objective can be achieved only by studying the workings of the global climate machine over a time span adequate to record a representative range of conditions in nature's own laboratory. And for this, the record of past climates is indispensable. From the evidence discussed below in Summarized in Figure 8.2, we conclude that a satisfactory perspective of the history of climate can be achieved only by the analysis of observations, spending the entire time range of climatic variation, say, from 10 to the minus one to 10 to the ninth years. Near the short end of this range, there is a rich instrumental record to collate and analyze, although, as discussed elsewhere in this report, awkward gaps exist in our knowledge of many parts of the ARCI system during even the past hundred years. As the time scale of observations is lengthened to include earlier centuries, a direct instrumental record becomes less and less adequate. A continuous time series of observations as far back as the 17th century is available for only one area. For earlier times, the instrumental record is blank, and indirect means must be found to reconstruct the history of climate. The science of paleoclimatology is concerned with the Earth's past climates, and that branch which seeks to map the reconstructed climates may be referred to as paleoclimatography. So defined, the science of paleoclimatology does far more than satisfy man's natural curiosity about the past. It provides the only source of direct evidence on processes that change global climate on time scales longer than a century. When calibrated and assembled into global arrays, these data will be essential in the reconstruction of paleoclimates with numerical models. Nature of paleoclimat evidence The subject of ancient climates may conveniently be approached in terms of the nature of the climatic record, whether from human, historical recordings, or from proxy or natural climate indicators. It is therefore convenient to identify historical climatic data and proxy climatic data as sources of paleoclimatic evidence. Prior to the period of instrumental record, historical climatic data was found in books, manuscripts, logs, and other documentary sources, and provide valuable, although fragmentary, climatic evidence before the advent of routine meteorological observations. Lam, 1969, has pioneered the collection of such data, and has charted the main course of climate over western New York during the past 1,000 years, figure A.2B. Where the historical or manuscript record overlaps the instrumental record, the climatic reconstructions may be confirmed and calibrated by the latter. In contrast, the proxy record of climate makes use of various natural recording systems to carry the record of climate back into the past. Records from well-dated tree rings, annually layered, or varbid, lake sediments, and ice cores resemble the historical data in that values can be associated with individual years, and may be calibrated with modern data to extend the climatic record for many centuries, and in certain favorite sites for as long as 8,000 to 10,000 years. Other recording systems, such as the pollen concentration, lake sediments, and fossil organisms, and oxygen isotopes in ocean sediments, have less resolution but may provide continuous records extending over many tens of thousands of years. These and other characteristics of proxy climatic data sources are summarized in table A.1. In general, the older geological records provide only fragmentary and generally qualitative information, but constitute our only records extending back many millions of years. For the past 1 million years, however, and especially for the past 100,000 years, the record is relatively continuous, and can be made to yield quantitative estimates of the values of a number of significant climatic parameters. These include the total volume of glacial ice, and its inverse, the sea level, the air temperature and precipitation over land, the sea surface temperature and salinity for much of the world ocean, and the general trend of air temperatures over the polar ice caps. Like sensing systems made by man, each natural paleoclimatic indicator must be calibrated, and each has distinctive performance characteristics that must be understood if the data are to be interpreted correctly. In discussing these sources, it is useful to distinguish between those paleoclimatic indicators that are more or less continuous recorders of climate, such as tree rings and valves, and those whose records are episodic, such as mountain glaciers. We should also consider the minimum attainable sampling interval that is characteristic of a particular paleoclimatic indicator, c-table A.1. Thus, tree rings, valves, and some ice cores can be sampled at intervals of one year. Paulin and other sedimentary fossil samples only rarely represent less than about 100 years, and many geological series are sampled over intervals representing 1,000 years or more. These figures reflect differences in the resolving power of each proxy indicator. Climate-induced changes in the plant community, as reflected in Paulin concentrations, for example, are relatively slow. The high frequency information is lost, but low frequency changes are preserved. In contrast, tree ring records and nasotopic records in ice cores respond yearly and even seasonably in favored sites. Each proxy record also has a characteristic chronologic and geographic range over which it can be used effectively. Tree ring records go back several thousand years at a number of widely distributed continental sites. Paulin records have the potential of providing synoptic coverage over the continents for the past 12,000 years or so, and a nearly complete record of the fluctuating margins of the continental ice sheets is available for about the last 40,000 years. Planktonic and benthic fossils from deep ocean cores can, in principle, provide nearly global coverage of the oceans going back tens of millions of years, although sampling difficulties have thus far limited our access to sediments deposited during the last several hundred thousand years. Although instrumental records best provide the framework necessary for the quantitative understanding of the physical mechanisms of climate and climatic variation with the aid of dynamic models, the increasingly quantitative and synoptic nature of paleoclimatic data will add a much-needed perspective. As discussed elsewhere in this report, it is therefore important that the historical and proxy records of past climate be systematically assembled and analyzed in order to provide the data necessary for a satisfactory description of the Earth's climates. Instrumental and Historic Methods of Climate Reconstruction Over the past three centuries, the development of meteorological instruments and appropriate platforms for sensing the state of the atmosphere, hydrosphere, cryosphere system has produced an important storehouse of quantitative information pertaining to the Earth's climates. Time series of these records show that climate undergoes considerable variation from year to year, decade to decade, and century to century. From a practical viewpoint, much of this information mirrors the economically more important climatic variations, as found, for example, in the changes of crop and animal production, the patterns of natural flora and fauna, and the variations of the levels of lakes and streams and the extent of ice. Generally, the term climate is understood to describe in some fashion the average of such variations. As discussed in Chapter 3, a complete description of a climatic state would also include the variants and extremes of atmospheric behavior, as well as the value of all parameters and boundary conditions regarded as external to the climatic system. In discussing the reconstruction of climates from instrumental data, several characteristics of past and present observational systems should be considered. First, instrumental observations have been obtained for the most part for the purpose of describing and forecasting the weather. Hence, although extensive records of such weather elements as temperature, precipitation, cloudiness, wind and observations are available, they are inadequate for many climatic purposes. There exist few direct measurements related to the thermal forcing functions of the atmosphere, hydrosphere, cryosphere system, such as the solar constant, the radiation, heat and moisture budgets over land and ocean surfaces, the vegetative cover, the distribution of snow and ice, the thermal structure of the oceanic surface layer, and atmospheric composition and turbidity. Second, observational records may be expected to contain errors due to changes in instrument design and calibration, and to changes in instrument exposure and location. There is therefore a need to establish and maintain conventional observations at reference climatological stations and a need to identify, insofar as possible, benchmark records of past climate. Such observations are needed to supplement the climatic monitoring program described elsewhere, C Chapter 6. Third, the time interval over which portions of the climatic system need to be described are very different. If, for example, the fluctuations in the volume and extent of the polar ice caps are to be studied, a time interval of order 100,000 years, the maximum residence time of water in the ice caps, is required, or if atmospheric interaction with the deep oceans is to be considered, then a time interval of the order 500 years, the residence time of bottom water, is required. It is therefore apparent that the period of instrumental records covering the past century or two is long enough only to have sampled a portion of such climatic responses, and that our information on older climates must come from historical sources and from the various natural, proxy, indicators of climate described earlier. Although such records will always be fragmentary, we should recognize their unique value in describing the past behavior of the Earth's climatic system. For practical reasons, it has been convenient to compute climatic statistics over relatively short intervals of time, such as 10, 20, or 30 years, and to designate the 30-year statistics as climatic normals. It is important to note, however, that the most widely accepted climatic normals, for the period 1931 to 1960, represent one of the most abnormal 30-year periods in the last 1,000 years, Bryson and Hare, 1973. As noted elsewhere in this report, the entire last 10,000 years are themselves also abnormal in the sense that such interglacial climates are typical of only about one-tenth of the climatic record of the last million years. While continuous observations of atmospheric pressure, temperature, and precipitation are available at a few locations from the late 17th century, such as the record of temperature in central England assembled by Manley, 1959, it is only since the early part of the 18th century that the spatial coverage of observing stations has permitted the mapping of climatic variables on even a limited regional scale. These and other scattered early observations of rainfall, wind direction, and sea surface temperature have been summarized by LAM, 1969. Only since about 1850 are reliable decadal averages of surface pressure available for most of Europe, and only since about 1900 are the reliable analyses for the mid-latitudes of the northern hemisphere as shown in figure 8.3, and only since about 1950 does a surface observational network begin to approach adequate coverage over the continents. Large portions of the oceans, particularly in the southern hemisphere, remain inadequately observed. For the climate of the free atmosphere, the international radio-saud network permits reliable analyses for the mid-latitudes of the northern hemisphere only since the 1950s, and less than adequate coverage exists over the rest of the globe. Beginning in the 1960s, routine observations from satellite platforms have begun to make possible global observations of a number of climatic variables, such as cloudiness, the planetary albedo, and the planetary heat budget. Yet many important quantities, such as the heat and moisture budgets at the Earth's surface, and the thermal structure and motions of the oceanic surface layer remain largely unobserved even on a large scale. Biological and Geological Methods of Climate Reconstruction During the first three decades of the 19th century, Vanessa in Switzerland and Esmark in Norway inferred the existence of a prehistoric ice age from the study of vegetation-covered moraines and other glacial features in the lower reaches of mountain valleys. After a century of effort, the literature of paleoclimatology has become so diverse and so burdened by stereographic terminology that it is useful to provide a summary of paleoclimatic techniques. The quantitative description of past climates as determined by biological and geological records requires the development of paleoclimatic monitoring techniques and the construction of timescales by suitable chromatic or dating methods. In general, the second of these problems is the more difficult. Beyond the range of carbon-14 dating, the past 40,000 years, it has only since about 1970 that the main chronology of the climate of the past 100,000 years has become clear. And only since 1973 that the main features of the chronology of the past million years have been established. Key discoveries in these time ranges have been in the sea level records of oceanic islands and in the sedimentary records of deep sea cores. In preparing the survey, the chronology of these records has been used as a framework into which the data from more fragmentary or poorly dated records have been fitted. Monitoring techniques The problem of developing a paleoclimatic monitoring technique or finding something meaningful to plot may be broken down into three sub-problems. A natural climatic record must be a. Identified, b. Calibrated, and c. Obtained from a stable recording medium. Identification of natural climatic records A number of different monitoring techniques that can provide data for paleoclimatic inference are summarized in Table A.1 and are based on observations of fossil pollen, ancient soil types, lake deposits, marine shorelines, deep sea sediments, tree rings, and ice sheets, and mountain glaciers. The techniques that are emphasized here are those that in general yield more or less continuous time series. Other types of proxy data are also useful in the reconstruction of climatic history. See, for example, Flint, 1971, or Washburn, 1973. Calibration of paleoclimatic records Many proxy records must be calibrated to provide an estimate of the climatic parameter of interest. The elevation of an ancient coral reef, for example, is a record of a previous sea level, but before it can be used for paleoclimatic purposes, the effect of local crustal uplift or subsidence must be removed. Bloom, 1971, Matthews, 1973, Walcott, 1972. Another example may be cited from paleontology, where the taximatic composition of fossil assemblages and the width of tree rings are known to reflect the joint influences of several ecological and environmental factors of climatic interest. Here appropriate statistical techniques are used to define indices that give estimates of the individual paleoclimatic parameters, such as air temperature, rainfall, or sea surface temperature and salinity. In the case of tree rings, although each tree responds only to the local temperature, moisture, and sunlight, for example, by averaging over many sites, the tree's response may be related to the large-scale distribution of rainfall and surface temperature. In this way, a statistical relationship may be established with a variety of parameters, even though they may not be direct causes of tree growth. When such tree ring data are carefully dated, they can thus provide estimates of the past regional variations of climatic elements such as precipitation, temperature, pressure, drought, and stream flow, for its et al., 1971. These methods yield what are called transfer functions, which serve to transform one set of time-varying signals to another set that represents the desired paleoclimatic estimates. In addition to their application to tree ring data, multivariate statistical analysis techniques have been successfully applied to marine fossil data, Embry and Kip, 1971, Embry, 1972, Embry et al., 1973, and do fossil pollen data, Bryson et al., 1970, Webb and Bryson, 1972. Difficult results indicate, for example, that average winter sea surface temperatures 18,000 years ago in the Caribbean were about 3 degrees centigrade lower than today, while those in the mid-latitudes of the North Atlantic were about 10 degrees centigrade below present levels. The oxygen isotope ratio, oxygen 18 to oxygen 16, as it is preserved in different materials, is used in three separate paleoclimatic monitoring techniques. Although the results are interpreted differently, in each technique, the ratio is measured as a departure delta-oxygen 18 from a standard, with positive values indicating an excess of the heavy isotope. One technique examines the ratio in polar ice caps, where the values of delta-oxygen 18 are generally on the order of 30 parts per thousand lower than in the Oceanic Reservoir, because of the precipitation and isotopic enrichment that accompanies the transport of water vapor into higher latitudes. As shown by Densgard, 1954, and by Densgard et al., 1971, the value of delta-oxygen 18 in each accumulating layer of ice is closely related to the temperature at which precipitation occurred over the ice. Although complicating effects make it impossible to convert the delta-oxygen 18 curve into an absolute measure of air temperature, the isotopic time series are extraordinarily detailed. Another isotopic technique records delta-oxygen 18 in the carbonate skeletons of plantonic marine fossils, Emiliani, 1955, 1968. Here the ratio is determined by the isotopic ratio and temperature in the near-surface water in which the organisms live. Work by Shackleton and Upadike, 1973, demonstrates that the observed ratio is predominantly influenced by the isotopic ratio in the seawater. Hence the isotopic curve reflects primarily the changing volume of polar ice, which, upon melting, releases isotropically light water into the ocean. A third technique measures the isotopic ratio in benthic fossils, whose skeletons reflect conditions prevailing in bottom waters. By making the assumption that the temperature of bottom water underwent little change over the past million years, the difference between the isotropic ratio observed in benthic and plantonic fossils can be used to estimate changes in surface water temperatures. Initial application of this technique, Shackleton and Upadike, 1973, provides an independent confirmation of the previously cited estimate of glacial age Caribbean temperatures obtained by paleological techniques. Over time spans on the order of tens of millions of years, measurements of delta-oxygen and 18 in benthic fossils offer a means of tracing changes in bottom water in which the effects of changing polar temperatures and ice volumes are combined. Douglason, 7, 1973. Evaluation of the recording medium All paleoclimatic techniques require that ambient values of a climatic parameter be preserved within individual layers of a slowly accumulating natural deposit. Such deposits include sediments left by melting glaciers on land, sentiments accumulating in peat bogs, lakes, and on the ocean bottom, soil layers, layers accumulating in polar ice caps, and the annual layers of wood formed in growing trees. Ideally, a recording site selected for paleoclimatic work should yield long, continuous, and evenly spaced time series. The degrees to which these qualities are realized vary from site to site, so that distortions and non-uniformities in each record must be identified and removed. The stratigraphic techniques by which this screening is accomplished will not be discussed here, although the reader should be aware that, with the exception of tree rings, some degree of chronological distortion will occur in all paleoclimatic curves where chronic control is lacking. To enable the reader to form his own judgments as to the chronology of past climatic changes, most of the paleoclimatic curves given in this report show explicitly the time control points between which the data are spaced in proportion to their relative position in the original sedimentary record. This procedure assumes that accumulation was constant between the time controls, which is a reasonable assumption in favorable environments. In other cases, this assumption introduces a distortion in the signal and inconsequent uncertainty in the timing of the inferred climatic variations. Each of the recording media used in paleoclimatography has characteristic limitations and advantages. As summarized in table A.1, their reconstruction of past climates requires evidence from a variety of techniques, each yielding time series of different lengths and sampling intervals and reflecting variations in different regions. The tree ring record, for example, provides evenly spaced and continuous annual records, but only for the past few thousand years. The ice margin record of both valley and continental glaciers is discontinuous, especially prior to about 20,000 years ago, because each major glacial advance tends to obliterate, or at least to conceal, the earlier evidence. Records of lake levels and sea level are also discontinuous. The former rarely extend back more than 50,000 years, although the latter extend back several hundred thousand years. Soil sequences display great variability in sedimentation rate, but provide continuous climatic information for sites on the continents where other records are not available, or are discontinuous. In favorite sites, the soil record extends back about a million years. Pollen records are usually continuous, but are rarely longer than 12,000 years. Deep sea cores provide material for the study of fossils, oxygen isotopes, and sedimentary chemistry. These records are relatively continuous over the past several hundred thousand years and are distributed over large parts of the world ocean. They are relatively uniform but low deposition rates, however, generally limit the chronological detail obtainable. Cores taken in the continental ice sheets provide a detailed and generally continuous record for many thousands of years, although their interpretation is handicapped by the lack of fully adequate models of the ice flow with its characteristic velocity, temperature, feedback. Chronometric techniques The problem of constructing a paleoclimatic chronology has been approached by four direct methods and one indirect method. Dental chronology The most accurate direct dating is achieved in tree ring analysis, in which many records with overlapping sets of rings are matched. With sufficient samples, virtual certainty in the dates of each annual layer may be obtained, and a year by year chronology can be established for periods covered by the growth records of both living and fossil trees. Such records are especially valuable for studying variations of climate during the past few hundred years and can be extended to many of the land regions of the world. Analysis of annually layered sediments In favored locations, lakes with annually layered bottom sediments provide nearly the same time control as to tree rings. Some ice cores and certain marine sediment cores from regions of high deposition rates also contain distinct annual layers. These data, along with tree rings and historical records, are the only source of information on the high frequency portion of the spectrum of climatic variation. Radiocarbon dating The advent of the Carbon-14 method in the early 1950s was a major breakthrough in paleoclimatography, for it made possible the development of a reasonably accurate absolute chronology of the past 40,000 years in widely distributed regions. Prior knowledge was essentially limited to dated tree ring sequences, for the past several thousand years, and to varve counted sequences in Scandinavia, extending back to about 12,000 years. The Carbon-14 method has an accuracy of about plus or minus 5% of the age being determined. That is, material 10,000 years old could be dated within the range 9,500 to 10,500 years. The calibration of Carbon-14 ages against those determined from dental chronology gives insight into the variations of atmospheric Carbon-14 production rates over the past 7,000 years, Zeus 1970. Decay of long-lived radioactives These methods employed daughter products of uranium decay or the production of argon-40 to a potassium decay. Used under favorable circumstances, one of the uranium methods, the decay of thorium-230, can provide approximate average sedimentation rates in deep-sea cores. The other method, the growth of thorium-230, can be used successfully on fossil corals to provide discrete dates for shoreline features of recording ancient sea levels. Together, these techniques have provided a reasonably satisfactory chronology of the past 200,000 years, with a dating accuracy of about plus or minus 10%. Our chronology for older climatic records is based on the well-known potassium argon technique applied to terrestrial lava flows and ash beds. This technique has provided, for example, the important dates for paleomagnetic reversal boundaries. Stratograph correlation with dated sequences Much of the absolute chronology of climatic sequences is supplied by an indirect method, namely the stratigraphic correlation of specific levels in an undated sequence with dated sequences from another location. For example, a particular glacial moraine that lacks material for carbon-14 dating may be identified with another form at the same time that has datable material. Such correlation by direct physical means is limited to relatively small regions, however, and stratigraphic correlation techniques must be used. Three such methods form the backbone of the chronology of paleoclimate, biostratigraphy, isotope stratigraphy, and paleomagnetic stratigraphy. The techniques of biostratigraphy use the levels of extinction or origin of selected species as the basis for correlation. This method has enabled Bergen, 1972, for example, to devise a time scale of the last 65 million years that is widely used as the basis for historical interpretation. Isotope stratigraphy, applicable only to the marine realm, makes use of the fact that the record of oxygen isotope variations, which reflects chiefly the global ice volume, has distinctive characteristics that permit the correlation of previously undated sequences. The application of paleomagnetic correlation techniques has revolutionized our approach to the climatic history of the past several million years. They are important stems from the fact that the principal magnetic reversal boundaries, which have occurred irregularly about every 400,000 years, are recorded in both marine and continental sedimentary sequences. Regularities in Climatic Series On the assumption that climatic changes are more than just random fluctuations, paleoclimatologists have long sought evidence of regularities in proxy records of the Earth's climatic history. Many have found what they believe to be firm evidence of order and refer to the chronological patterns as cycles. Although the number of records is limited, and hard statistical evidence is sometimes lacking, it is nevertheless convenient to describe some of the larger climatic changes in terms of quasi-periodic fluctuations, or cycles, with specified mean wavelengths or periods, in the sense that they describe the apparent repetitive tendency of certain sequences of climatic events. For example, many aspects of the global ice fluctuations during the past 700,000 years may be summarized in terms of a 100,000-year cycle, C-figure A.2e. Each such period is marked by a gradual transition from a relatively ice-free climate, or interglacial, to a short, intense glacial maxima and followed by an abrupt return to ice-free climate. No two such cycles are the same in detail, however, and should not be construed as indicating strict periodicities in climate. Some paleoclimatic cycles may be periodic, or least quasi-periodic, and rest on evidence that is exclusively or mainly chronological. The best example is the approximate 100,000-year cycle formed from the spectral analysis of time series, such as that shown in figure A.4. For the 100,000-year cycle, as well as some of the higher frequency fluctuations that modify it, there is circumstantial evidence to suggest that these have in some way been induced by secular variations of the Earth's orbital parameters, which are known to alter the latitudinal pattern of the seasonal and annual solar radiation received at the top of the atmosphere. For the 2,500-year, and shorter, fluctuations suggested by some proxy data series, the causal mechanism, is unknown. With the possible exception of the approximately 100,000-year quasi-periodic fluctuation referred to above, the quasi-biannual oscillation of 2-3-year period is the only quasi-periodical oscillation whose statistical significance has been clearly demonstrated. This is not to say that other such fluctuations in climate data are absent, but rather that much further analysis of proxy records is required. A question of equal importance is the shape of the continuum variant spectrum of climatic fluctuations. A uniform distribution of variance as a function of frequency, or white noise, would imply a lack of predictability in the statistical sense or a lack of memory in prior climatic states. A red noise spectrum, on the other hand, in which the variance decreases with increasing frequency, implies some predictability in the sense that successive climatic states are correlated. The existence of nonzero autocorrelations in such a spectrum implies that some portion of the climatic system retains a memory of prior states. In view of the relatively short memory of the atmosphere, it seems likely that this is provided by the oceans on timescales of years to centuries and by the world's major ice sheets on longer timescales. An initial estimate of the variance spectrum of temperature has been made from the fluctuations on timescales from 1 to 10,000 years by Kutzbach and Bryson, 1974, and is shown in Figure A.5. This spectrum has been constructed from a combination of calibrated botanical, chemical, and historical records, along with instrumental records in the North Atlantic sector. As may be seen in Figure A.5, A, the variance spectral density increases with decreasing frequency, increasing period, over the entire frequency domain, but is most pronounced for periods longer than about 30 years. In Figure A.5, B, the spectrum of the same time series is shown with frequency on a logarithmic scale and the ordinate as spectral density, V, times frequency, F, so that equal areas represent equal variance. Again, for periods longer than about 30 to 50 years, the observed temperature spectrum is seen to depart significantly from the white noise continuum associated with the higher frequency portion of the spectrum. The determination of the character of the variance spectrum of the various climatic elements remains largely a task for the future. We will use the term cycle in the following paragraphs to designate such quasi-periodic sequences of climatic events, since there appears to be no other word or phrase to convey the concept of a series of generally similar events spaced at reasonably regular intervals in time. Although our knowledge of the record of past climates has improved greatly during the last decade, a much broader paleoclimatic database is clearly required. Only then can adequate spectral analyses be performed and the spatial and temporal structure of paleoclimatic variations firmly established. Chronology of Global Climate Period of Instrumental Observations A variety of meteorological indices have been used to characterize the climate and its temporal variations during the past century or more of extensive observations. Global or hemisphere averaged indices, such as the surface temperature index shown in Figure A.6, are often used for this purpose. This index clearly suggests a worldwide warming beginning in the 1880s, followed by a cooling since the 1940s. The warming may be recognized as the last part of a complex but recognizable trend that has persisted since the end of the 17th century, Figure A.2b. The geographic patterns of temperature change during these overall warming and cooling epics show considerable variability with the largest changes concentrated in the polar regions of the northern hemisphere. Mitchell, 1963, has shown that the pattern of temperature change during recent decades is consistent with concomitant changes in the large-scale atmospheric circulation as reflected in sea-level pressure. Less attention has been given to the more complex relationships between circulation variation and changing precipitation patterns, although Krause, 1955a, 1955b, and Lamb, 1969, have considered this aspect of the problem. Lamb and Johnson, 1959, 1961, 1966, and Lamb, 1969, have made an extensive analysis of certain features of atmospheric circulation based on the observed and historically reconstructed surface pressure maps for individual months since about 1750. They have extracted such indices as the strength of the zonal and meridional flow, the position and wavelength of trough bridge patterns, and the position and strength of subtropical pressure systems. The year-to-year and decade-to- decade changes in these indices reflect changing large-scale circulation patterns, which in turn are associated with changing patterns of temperature and precipitation. From the instrumental era for the North Atlantic sector, the typical variability over 20 to 30-year intervals of the low-level westerlies is plus or minus 1 to 2 meters per second, and that for the planetary-scale circulation features, such as large-scale troughs and ridges, is plus or minus 1 to 2 degree latitude and 10 to 20 degrees longitude. Although changes in the position, pattern, and intensity of the general circulation are interrelated, such empirical studies suggest that longitudinal shifts have the most significant effects on the climatological temperature and precipitation patterns, at least for middle and higher latitudes. Examples of such shifts are shown in Figure A.7. In tropical and subtropical latitudes, on the other hand, latitudinal shifts appear to be more closely related to regional climatic variations, as indicated by the data of Figure A.8. The Last Thousand Years To obtain an indication of the climate in the Northern Hemisphere for the Last Thousand Years, LAM, 1969, has compiled manuscript references on the character of European weather and has developed an index of winter severity as shown in Figure A.2b and A.9. Although different longitudes show somewhat different results, the trends shown by this index, the excess number of unusually mild or unusually cold winter months over months of opposite character, for the period since about 1700, have been validated by comparison with thermometer records. Other portions of the record have been cross-checked with data on glacial fluctuations, oxygen isotope variations, and tree growth, so that the main characteristics of European climate during this period are reasonably well known. La Marche, 1974, has constructed temperature and moisture records from the ring-width variations in trees at high altitude arid sites in California, C Figure A.9c. Comparisons of his data with those from Europe shown in Figure A.9d indicate a degree of synchrony with the major fluctuations of temperature between the West Coast of North America and Western Europe during the Last Thousand Years. The early part of the Last Millennium, about AD 1100 to 1400, is sometimes called the Middle Age's Warm Epic, but was evidently not as warm as the first half of the 20th century. The period from about 1430 to 1850 is commonly known as the Little Ice Age, and some records indicate that this period had cold maxima in the 15th and 17th centuries. From such evidence, we inferred that the atmospheric circulation may have been more meridional than it present, and characterized in Western Europe and Western North America by short, wet summers and long, severe winters. During the Little Ice Age, many glaciers in Alaska, Scandinavia, and the Alps advanced close to their maximum positions since the last major ice age thousands of years ago. A visual impression of these events in the French Alps was shown in Figure A.1. The expansion of the Arctic pack ice into North Atlantic waters caused the North's colony in southwest Greenland to become isolated and perished, and in Iceland, grain that had grown for centuries could no longer survive. The Last 5,000 Years As indicated in Figure A.2C, the period from 7,000 to 5,000 years ago was marked by temperatures warmer than those that prevail today, and is thus sometimes known as the Hypes of Thermal Interval, Flint, 1971. The last 5,000 years is characterized by generally declining temperatures and a trend toward more extensive mountain glaciation, but not ice sheets, in all part of the world, Porter and Denton, 1967. Close examination of the records of mountain glaciers, tree lines, and tree rings suggests that this general cooling trend was itself punctuated in many parts of the world by cold intervals centered at about 5,300, 2,800, and 350 years ago as shown in Figure A.10. Much further analysis of proxy climatic records during this period is needed, including the evidence available from historical sources. The Last 25,000 Years The climatic record of the last 25,000 years is largely concerned with the present interglacial interval, or Holocene, and the terminal phases of the last major glaciation, C Figure A.2D. Although the maximum ice extent occurred between about 22,000 and 14,000 years ago, C Figure A.11, the curves of ice accumulation and decline are not identical for the various ice sheets. The Laurentide ice sheet, which covered parts of Eastern North America, and the Scandinavian ice sheet, which covered parts of Northern Europe, reached their maximum extent between 22,000 and 18,000 years ago, while the Cordillarian ice sheet achieved its maximum only 14,000 years ago. The maximum areas of the Northern Hemisphere ice sheets during the past 25,000 years were about 90% of the maxima during the last million years of the Pleistocene, C Table A.2. Wide spread deglaciation began rather abruptly about 14,000 years ago, and the waning phases of the continental ice sheets were characterized by substantial marginal fluctuations, Dreminis and Carrow, 1972, as shown in Figure A.11. The Cordillarian ice sheet, which had just attained its maximum extent, melted rapidly and was gone by 10,000 years ago. The Scandinavian ice sheet lasted only slightly longer and retreated at the rate of about one kilometer per year between about 10,000 and 9,000 years ago. The climatic instability suggested by these fluctuations in the margins of the Northern Hemisphere's major ice sheets is corroborated by the records from fossil pollen, deep-sea cores, ice cores, and sea-level variations, as shown in Figure A.12, and by Lockestein records in Western North America and Africa. By 8,500 years ago, the ice conditions in Europe had reached essentially their present state, and in North America, the ice sheets had shrunk to about their present extent by about 7,000 years ago. How widespread and synchronous these fluctuations were is not yet known, but evidence is growing that there were several periods of widespread cooling and glacial expansion in the regions bordering the Atlantic Ocean, C Figure A.2, C, spaced about 2,500 years apart. One of these glacial advances, the Younger Dryas event, about 10,800 to 10,100 years ago, was a climatic event of unparalleled roughness in Europe, establishing itself within a century or less and lasting for some 700 years. Northern forests that had advanced during the preceding warm interval were destroyed in many places. Such vegetation records suggest that by the end of the Younger Dryas event, European climate had returned to about its present state. The rise in sea-level during the last 18,000 years, indicated in Figure A.12, D, is generally ascribed to the melting of northern hemisphere continental ice sheets. Details of the sea-level curve, however, do not correspond to the chronology of deglation just described. While the continental ice sheets had essentially disappeared by about 7,000 years ago, the worldwide stand of sea-level had reached its maximum only during the last few thousand years, or is still slowly rising, Bloom, 1971. One possibility is that the West Antarctic ice sheet is unstable and has been disintegrating during the entire interval in question. Further research is clearly needed to settle this question, although it serves to illustrate the global interrelationships among the elements of the climatic system. The last 150,000 years In order to find an ancient counterpart to the warm, ice-free conditions of the past 10,000 years, the Holocene, or present interglacial, it is necessary to go back some 125,000 years to an interval known as the Eman Glindeglacial, C Figure A.2. As shown by the proxy data of Figure A.13, the warmest part of this period lasted about 10,000 years and was followed abruptly by a cold interval of substantial glacial growth lasting several thousand years. The interval between this post-Eman event, about 115,000 years ago, and the most recent glacial maximum, 18,000 years ago, was characterized by marked fluctuations so were imposed on a generally declining temperature. An intense glacial event about 75,000 years ago is sometimes used to separate the interval into an older and generally non-glacial regime and a more recent glacial one. A remarkable feature of the climatic record of the past 150,000 years is that both the present and the Eman interglacials began with an abrupt determination of an intensely cold, fully glacial interval. Because these catastrophic episodes of deglaciation have left such a strong imprint on the climatic record, they have been named, in order of increasing age, Termination I and Termination II, C Figure A.14, and Brooker and Bendonk, 1970. The Last Million Years For at least the last one million years, the Earth's climate has been characterized by an alteration of glacial and interglacial episodes marked in the Northern Hemisphere by the waxing and waning of continental ice sheets and in both hemispheres by periods of rising and falling temperatures. How clearly these fluctuations are stamped on the various proxy data records is shown in Figure A.14. The most prominent features of the isotope curves shown here are seven terminations, marking a transition from full glacial to full interglacial conditions. All but one, Termination III, of these changes are relatively rapid monotonic swings and provide an objective basis for defining a climatic cycle for at least the last 700,000 years. As shown in Figure A.14, these same fluctuations can be recognized in diverse and widely distributed records, including the chemical composition of Pacific sediments, fossil plankton in the Caribbean, and the soil types in Central Europe. These cycles, identified as A to E by Cookla, 1970, are found in each of the records shown in Figure A.14 and may be grouped into a climatic regime covering the last 450,000 years, designated alpha. The earlier records, regime beta, show higher frequency fluctuations with less coherence among the various proxy climatic recorders. The last 100 million years. Although continuous and detailed records are lacking for these earlier times, at least a broad outline of this period of climatic history may be discerned. From the climatic point of view, perhaps the most striking aspect of world geography at the beginning of this interval was the essentially meridial configuration of the continents and shallow ocean ridges, which must have prevented a circumpolar ocean current in either hemispheres. In the south, this barrier was formed by South America and Antarctica, which lay in approximately their present latitudinal positions, by Australia, then a northeastward extension of Antarctica, and by the narrow and relatively shallow ancestral Indian Ocean, decent hold in 1970. About 50 million years ago, the Antarctica Australian passage began to open, can it at all, 1973. And as Australia moved northeastward, the Indian Ocean widened and deepened. Both paleonautological and sedimentary evidence suggest that about 30 million years ago, the Antarctic circumpolar current system was first established. This must be considered a pivotal event in the climatic history of the past 100 million years. And when the evidence of global plate movements is complete, it may well be possible to account for much of the secular climatic changes of this period as a response to the changing boundary conditions imposed by the distribution of land and ocean. During the last part of the Mesozoic era, from 100 million to 65 million years ago, global climate was in general substantially warmer than it is today, and the polar regions were without ice caps. About 55 million years ago, numerous geological records, Attencourt 1970, Flint 1971, make it clear that global climate began a long cooling trend known as the Sinozetic Climate Decline, C-Figure A.15. Evidence from the marine record indicates that about 35 million years ago, and Antarctic waters underwent a substantial cooling, Douglason 7, 1973, Shackleton and Kennet 1974A 1974B. There is direct evidence that ice reached the edge of the continent in the Ross Sea area some 25 million years ago, and during the Oligocene epic, roughly 35 million to 25 million years ago, global climate was generally quite cool, more 1972. During early Myocene time, 20 million to 15 million years ago, evidence from low and middle latitudes indicates a warmer climate, but isotope evidence and faunal data indicate that this warming did not affect high southern latitudes. About 10 million years ago, there is widespread evidence of further cooling, substantial growth of Antarctic ice, Shackleton and Kennet 1974A 1974B, and growth of mountain glaciers in the northern hemisphere, then at all, 1971. For general descriptive purposes, the present glacial age may be defined as the beginning of this time. Indirect evidence from marine sediments indicates that about 5 million years ago, the already substantial ice sheets on Antarctica underwent rapid growth and quickly attained essentially their present volume, Shackleton and Kennet 1974A 1974B. This evidence is generally consistent with direct records from the Antarctic continent, which showed that between 7 million and 10 million years ago, a large ice sheet existed in western Antarctica, and that by about 4 million years ago, the ice sheet in east Antarctica had developed to essentially its present volume, Denton et al., 1971, Mojewski, 1973. The present Antarctic ice mass is equivalent to about 59 meters of sea level. Although the behavior of the smaller west Antarctic ice sheet is incompletely known, the available evidence indicates that it has undergone considerable fluctuation, and that its variation are roughly synchronous with the northern hemisphere glacial interglacial cycle. This may be due to the fact that while the east Antarctica ice sheet is solidly grounded on the continent, much of the west Antarctic ice mass is grounded on islands or on the sea floor, and could therefore be significantly influenced by sea level variations due to glacial changes in the northern hemisphere. Such continental ice sheets first appeared in the northern hemisphere about 3 million years ago, occupying lands adjacent to the North Atlantic Ocean, Bergeron, 1972b, and during at least the last million years, the ice cover on the Arctic Ocean was never less than it is today, Hunkins et al., 1971. The last one billion years. Our knowledge of the climatic events over this time range consists principally of evidence of glaciations as preserved in the geological record. This may be seen in perspective with that for the more recent periods discussed above in figure A.15. The present glacial age is seen to be at least the third time that the planet has suffered widespread continental glaciation. The permo-carboniferous glacial age, about 300 million years ago, occurred at a time when the Earth's land masses were joined in a single supercontinent, Pangaea. The area of the continent was distributed in roughly equal proportions between the hemispheres, with a concentration of land in the midlatitudes, Deeds and Holden, 1970. Glaciated portions of Pangaea included parts of what are now South America, Africa, India, Australia, and Antarctica. One or more early glacial ages are known from late Precambian times, about 600 million years ago, from the indication of glaciation and deposits now widely scattered over the globe, including Greenland, Scandinavia, Central Africa, Australia, and Eastern Asia, Holmes, 1965. Although other glacial ages may have occurred besides those recognized in figure A.15, none has left such a clear and widespread impact on the geological record, Steiner and Grilmer, 1973. While the evidence is far from complete, it may be that each of the Earth's major glacial ages, including the present one, resulted from crustal movements that permitted the development of sharp thermal gradients over the continental landmass that includes a pole of rotation. To establish this or other hypotheses of long period climatic changes, however, will require the assembly of a much more complete geologic record and the performance of appropriate climate modeling experiments. End of section 9. Recording by Todd