date: Wed Dec 18 14:03:03 1996 from: Keith Briffa to: Paul Sinclair Report of the PEP III Stream 1 Working Group convened at the IGBP-PAGES Workshop, Bierville, Sept. 12-15, 1996. Introduction This group was charged with reviewing the potential sources of qualitative, high-resolution evidence of climate variability and climate change on interannual, decadal and century timescales along the PEP III transect within the Stream 1 timeframe of the last 2000 years. The group discussed the specific strengths and limitations of various types of climate proxies and reviewed their temporal and spatial coverage. A principal aim of the group was to re- examine and, where possible, redefine major foci for PEP III Stream 1 activities in the medium term, taking account of ongoing initiatives, identifying new opportunities and highlighting possible geographical imbalances. The following is a brief description of the outcome of the discussions around these topics, including a summary of the major conclusions and recommendations of the group. An expanded review and discussion document, in the form of a revised implementation plan for PEP III Stream 1 activities, is currently being prepared. Rationale and General Objectives The group recognised that the particular importance of general Stream 1 activities, within all PEP transects, arises from the need to define accurately the history of climate variability on interannual and decadal, as well as longer, timescales. Well-dated, high-resolution climate information from before the period of possible human modification and for sites where human modification over the last 2000 years has been minimal, is essential if we are to understand the mechanisms of natural climate change and rigorously ascribe causes. It follows that there exists a concomitant need to establish quantitatively, and with equal accuracy, the histories of potential climate forcing agents - such as changes in solar output, volcanic activity, etc. The reconstruction of both climate and climate-forcing histories within the Stream 1 timeframe is also highly relevant for attempts at rigorous attribution of recent climate change to anthropogenic causes. In parallel with the general ambitions of all PEP transects, the group therefore defined the ultimate aims of PEP III Stream 1 activities as follows: 1. To document the full range of climate variability that has taken place across the Western Eurasian and African continents over the last 2000 years. 2. To identify factors responsible for causing climate change. 3. To provide datasets for the parameterization and validation of climate models. 4. To explore the influence of climate change on human activity and the influence of human activity on climate change. To accomplish these objectives requires sensitive methods because the likely magnitude of change is considerably less in recent millennia than on longer, e.g. Milankovitch, timescales. It was also agreed that studies within Time Stream 1 should not be constrained by the 2000-year limit if longer timescales within the Holocene could be embraced with equivalent precision and accuracy. Important ‘Systems’ and Proxies The history, current state and prospects for future development of a number of major sources of paleoclimate and palaeoenvironmental sources were discussed. The group focussed attention on their major strengths and known weaknesses and explored the immediate prospects for better exploiting these sources and extending the geographical range within the PEP III transect. These sources and the specific proxies that they provide will be discussed in more detail in the forthcoming Stream 1 implementation plan. They include 1. Instrumental Records 2. Historical and Documentary Records 3. Tree Rings 4. Ice Cores 5. Speleothems 6. Lake Sediments 7. Marine/Large Lake/Inland Sea Sediments 8. Peats 9. Archaeological Records A PEP III Perspective Though the primary axis of interest in PEP III activities is defined north to south, the east/west dimension is significant in terms of the large distances across Europe and Africa and because of the existence of large climate gradients. Localised records that provide well-calibrated climate information are valuable in their own right, the more so where they are of high-resolution (seasonal or annual) and where the dating is good. Even more valuable are similar records or data sources that occur along one or more parts of the PEP III transect, or better still, that have extended spatial or network coverage across large regions. While recognising these optimum data attributes, the group felt strongly that the development of data sources that meet only some of these criteria was still to be strongly encouraged. All proxy data sources are limited in different ways in their ability to represent climate variations and a multi-proxy approach to palaeoclimate research is most likely to yield maximum results. Different records will respond to different climate parameters. Comparison of different records will identify similarities and differences - providing either mutual support for, or raising questions and suggesting alternative interpretations of, the different data. The multiproxy approach should highlight individual data limitations; in some cases allowing refinement of the dating control; provide better understanding of the time- averaged or lag responses, and generally lead to a more accurate and complete picture of climate changes than can be achieved using individual data sets in isolation. Regional Foci To optimise resources, taking account of ongoing research projects and recognising the current geographical imbalance in traditional regions of high-resolution palaeoclimate research, the group recommended a number of areas where future research might be usefully concentrated. Above all, these are regions of major importance within the global climate system. Instrumental records and modelling studies indicate that climate variability in each of these areas is influenced by characteristic large-scale modes of variability of the general atmosphere or the interaction of the atmosphere with the oceans. The regions identified were 1. The mid-to-high-latitude section (~north of 60 N) of Fennoscandia and western and central Russia 2. Western Europe and the Mediterranean regions under the influence of the North Atlantic Oscillation 3. Monsoon Effected areas of West Africa 4. Northeast Africa and the Nile Valley 5. Southern Africa and Madagascar An Archaeological Emphasis Until now, there has been little mention in Stream 1 discussions of the potential for generating past climate information through the use of high-resolution historical and, particularly, archaeological data. Archaeology and PEP III Stream 1 studies in general, will gain reciprocal benefits through a greater emphasis on selected archaeological studies and increased collaboration between archaeologists and palaeoclimate scientists. Both require an understanding of how people have impacted on the natural environment, not least in order to interpret proxy climate data. Initiating the development and study of these data in areas of high archaeological potential will be invaluable for understanding human responses to climate and other environmental change. Though such studies are feasible throughout the PEP III transect, they should be especially encouraged in areas where there is known potential to generate high-resolution data in space (local and regional) and time (annual- decadal) and that could be used to make comparisons with changes in human activities in other regions. Areas of proven archaeological potential occur widely in all parts of Europe, the Mediterranean and many parts of Africa. However, the group identified the following selected list. 1. The Nile Valley 2. East African Highlands and Coast 3. Madagascar Highlands and Coast 4. Southern Africa The background and specific rationale for highlighting these regions will be outlined in detail in the forthcoming PEP III Stream 1 Implementation Plan. These regions coincide with most of the regions already identified as important potential regional research foci. Important Methodological Issues The group reviewed a number of issues of relevance to the interpretational ‘value’ of palaeoclimate sources in general, and particularly in the context of Stream 1 objectives. The group consensus was to encourage explicit recognition among research workers of the importance of these topics. It was felt that experimentation with different palaeoclimate sources should attempt to address each of these issues explicitly and that direct reference to their particular significance be included in reports or research publications dealing with those sources. 1. Resolution The desirability of high resolution palaeodata has been stressed already. True annual resolution and, where it can be achieved, subannual resolution offer the prospect of specific seasonal reconstruction of climate and evidence of the annual cycle range. Some records may, however, be annually or sub-annually resolved, but may only reflect climate forcing at a specific time of the year. Conversely, annually registered data may still only reflect longer- term climate forcing, integrated over a number of years. Where annual resolution cannot be achieved, efforts should be made to establish the true resolving power of the data and the extent to which the resolution varies in time. While recognising the desirability of annual or subannual resolution, the group systems were decadal resolution can be attained e.g. many lake and mire systems, particularly where tephrochronology might provide additional precise links to other time series. 2. Dating and Chronology The importance of strong dating control, particularly within the context of Stream 1 studies, cannot be overemphasised. The value of statistical techniques for testing the association between different climate reconstructions, or between deduced climate variability and evidence of forcing agents, is highly dependent on accurate chronology. Even where palaeorecords are deposited annually or display annual layering, simple counting of the layers does not guarantee absolute dating accuracy. Some individual year boundaries may be ambiguous. In any individual record, depositional discontinuities or hiatuses or post- depositional disruption can occur. Where practical, comparisons of duplicate records at a site, or from adjacent sites, can confirm the dating accuracy or identify anomalies and enable the timescale to be corrected. When annual dating control is not feasible for a particular record, comparison with features of other, better-dated records may improve the chronology. Again, where feasible, multiple approaches to dating should be adopted. Where there is uncertainty in the dating, this must be clearly expressed. The group laid great stress on the desirability of adapting multiple approaches to dating time series in situations where absolute chronology is not routinely attainable. In this respect it was felt that the continued development of tephrachronologies and studies intended to explore the potential of using specific tephras as absolute dating horizons or time markers in many ice, lake, peat or other palaeodata sources should be noted. 3. Replication Analysing multiple or replicate samples of some proxy data record, besides helping to establish good dating control in the timeseries, is highly desirable because it provides a basis for quantifying the underlying signal strength represented in that record. Even perfectly dated, duplicate, records will not agree perfectly. The degree of similarity, however, will give an indication of how strongly the underlying forcing is represented in each series. The degree of difference is an indication of ‘noise’. Averaging replicate series (provided they have good dating control) will reduce the noise and produce a record which better expresses the underlying forcing. Because so much more information can be gleaned from duplicated records and higher, quantifiable confidence achieved by producing a mean series, duplicate sampling should be undertaken wherever possible. 4. Calibration Climate inferences drawn from palaeodata should be supported by rigorous statistical comparisons with observed meteorological data. Statistical regression equations or transfer functions applied to high-resolution palaeodata series to estimate past climate variability, should be calibrated in the time domain and not be based only on spatially-derived relationships. Where possible, empirical or statistical approaches to inferring past climate variability should be supported by theoretical modelling studies. Realistic confidence limits on palaeoclimate estimates should be provided. 5. Signal Representativeness Researchers should strive to make explicit the potential weaknesses or limitations in the representativeness or accuracy of their past climate inferences based on their detailed knowledge of the palaeodata and systems with which they work. For example, some records may represent high-frequency (interannual) climate forcing with good fidelity, but be limited in their ability to represent the same forcing on longer timescales, perhaps because the systems are able to adapt to gradual changes in forcing. Longer-timescale variations may, of necessity, be removed in the production of a palaeorecord perhaps because the primary data are known to be influenced on these timescales by processes other than the climate forcing of interest. Separate calibrations of a palaeorecord after spectral decomposition might highlight differences in the fidelity of the climate interpretation on different timescales. However, the short length of many available climate records will limit the extent to which even decadal-to-century timescale fidelity can be demonstrated. 6. Human Impact It is possible, even probable, that instrumentally recorded climate data are already registering anthropogenic disturbance of the ‘natural’ climate system. In addition, many other aspects of the natural environment have been modified by humans. In some localised areas this may have occurred over thousands of years. On much larger, even global scales, changes in atmospheric and terrestrial environments have certainly occurred over the last century. These include increasing concentrations of greenhouse gasses, heavy metal pollution, and nitrogen and sulphur deposition. Reductions in stratospheric ozone and increased surface ultraviolet radiation levels have been dramatic in recent decades. The potential for these and other factors associated with human activities to complicate the calibration of palaeodata against modern instrumental records must be appreciated. The degree to which any or a combination of these might affect the long-term validity of modern palaeodata/climate associations will be difficult to establish. The relative importance of different factors will obviously vary greatly according to the data source, and the specifics of the physical or biological systems involved. Still, future work must aim to provide quantitative answers. A combination of statistical, theoretical and experimental approaches will almost certainly be required. Conclusions and Recommendations The following is a list of the major conclusions arising out of the Stream 1 group discussions and includes a brief outline of the recommendations agreed by those present. The list is not prioritised. A major focus of PEPIII Stream 1 research should be on establishing the degree to which 20th century climates are unprecedented. This must involve quantitative reconstruction of past mean climates on multidecadal and century timescales as well as interannual variability and the frequency of extremes. There remains a widespread preconception that the Little Ice Age and Medieval Warm Epoch were ubiquitous features of the climate history of the last two millennia. There is still a need to further clarify the definition of these concepts in terms of their character, extent and precise timing, even within Europe. More research is required to establish the extent to which the concepts of the Little Ice Age and Medieval Warm Epoch are valid or relevant in other areas of the transect. Future research must attempt to clarify whether, if real, these phenomena represent unique events within the last 2000 years, and, if possible, in the earlier Holocene. Traditionally, high-resolution studies within the Stream 1 timeframe have been very much concentrated in the mid-to-high latitudes of the northern part of the PEPIII transect. In part, this is simply a result of proximity to established research situations. However, it also reflects genuine difficulty in locating datable, high- resolution records in low latitudes and in the African part of the transect. The existence of numerous, long tree-ring chronologies in Morocco is an important exception. Some potential for other dendroclimatological studies in north and east Africa has been clearly demonstrated and the future development of this potential should be explored. Given the problems of identifying annually resolved palaeosources in Africa, the group laid stress on the need to explore other less-well-resolved sources, particularly where they might be represented across wide areas of the transect. The concept of a specific research initiative aimed at exploring climate proxies in the sediments of a series of African crater lakes along the East African section of the transect was considered worthy of prioritisation. Even in Europe, there is important potential for identifying and processing historical and early meteorological records. Some palaeoseries, produced perhaps decades ago, now require updating. More research is required in order to identify and gauge the significance of anthropogenic environmental disturbance and the implications for palaeoclimate estimates calibrated against modern climate data. Collaborative projects, which involve comparison and an integrated approach to the interpretation of different climate proxy sources in areas where these are available, are not common. More projects should be initiated that bring together palaeoclimate scientists working with different systems (e.g. with ice core data, tree rings, lake sediments). Archaeological data have been underutilised in a palaeoclimate context. Collaboration between archaeologists and palaeoclimatologists, especially in areas with a tradition of detailed high-resolution archaeological work and historical and palaeoclimate proxies, should be promoted. Several regions, such as in the Mediterranean and in monsoon areas (particularly in Egypt) and in the south of Africa, are potential foci for such efforts. More attention to formal calibration studies and a routine quantitative evaluation of palaeoclimate data and reconstruction confidence limits, are encouraged, as is the establishment of standard ‘calibration training sets’, such as could be used for the interpretation of widespread data, such as chironomids. Intense attention to accurate chronology is encouraged in situations where absolute dating is not feasible. This might involve multiple dating proxies. Work to establish a detailed network of tephra histories in the circum North Atlantic Region and the possibility of developing a widely applicable ‘tephra chronology’ to provide fixed dating points for many less-well-resolved data sources is noted. The further development of the European palaeodata archive in Marseilles (linked to the World Data Center-A for Paleoclimatology in Boulder) is seen as fundamental to the feasibility of achieving the aims of PAGES PEPIII Stream 1. Universal submission of palaeodata and the free interchange of these data through the medium of the PEPIII data archive is crucial for the future success of Stream 1 research.