A Cooperative Holocene Mapping Project (COHMAP) has assembled a database of Holocene paleoclimatic data from 1000 locations with ¹⁴C dates (Table I). The data are stored on computer tapes at Brown University and will be available for verification of computer simulations of Holocene climates. Pollen, lake-level, and marine plankton data are the main sources of paleoclimatic Information (Reference Webb and HechtWebb in press). Multivariate statistical techniques (Reference Imbrie, Kipp and TurekianImbrie and Kipp 1971, Reference Webb and BrysonWebb and Bryson 1972) and energy-budget models (Reference KutzbachKutzbach 1980) exist that can transform each of these types of data into quantitative estimates of past temperature and precipitation.

Sites with pollen data are the most numerous (750 sites) within the database (Table I), and dense networks of pollen sites exist in eastern North America (Reference Bernabo and WebbBernabo and Webb 1977), Europe (Reference Huntley and BirksHuntley and Birks 1983), the Soviet Union (Reference PetersonPeterson unpublished) and Alaska (Reference AndersonAnderson unpublished). Pollen data from New Zealand, Africa, and South America are also being assembled. Work is now in progress to gain climatic estimates from this pollen information, and Reference Webb and HechtWebb (in press) has presented maps of past temperatures and precipitation in the midwestern United States.

Lake-level data are the next most numerous source of Holocene paleoclimatic data (200 sites). Dense networks of sites exist in western North America, Africa and Australia (Reference Street and GroveStreet and Grove 1979). The changing lake levels directly record variations 1n the local water budget from which energy-budget and hydrological models can estimate past changes in precipitation (Reference KutzbachKutzbach 1980). Water levels were high in many African and north-west Indian lakes at 9 and 6 ka BP, and the model calculations indicate that the rainfall in Chad and north-west India was about 200 mm higher at 6 ka BP than at present (Reference KutzbachKutzbach 1980, Reference Swain, Kutzbach and HastenrathSwain and others 1983).

Table I Distribution of Terrestrial and Marine Sites with Radiocarbon-Dated Pollen, Lake-Level, and Plankton Data in the Cohmap Database

Marine plankton data are mainly available from the Atlantic and western Indian oceans (Reference Ruddiman and McIntyreRuddiman and McIntyre 1981, Reference Prell, Hansen and TakahashiPrell in press). For most of the ocean regions, sedimentation rates are too low for deep-sea cores to resolve differences within the Holocene. Isolated samples from the Southern Ocean, from the margins of the Pacific Ocean, and from several marginal seas fill out the network of marine data. Patterns are evident in the maps of ocean-surface temperatures estimated from these data for 6 ka BP, but most temperatures are similar to those for today.

Recent climate-model simulations have shown how important Milankovitch factors are in influencing the global climatic patters at 9 ka BP (Reference KutzbachKutzbach 1981). The model simulations for rainfall in the subtropics agree well with the paleoclimatic data for high lake levels in this region at 9 ka BP (Reference Street and GroveStreet and Grove 1979, Reference Prell, Hansen and TakahashiPrell in press). The presence of the Laurentide ice sheet also affects the simulated climatic patterns at 9 ka BP (Reference Kutzbach and Otto-BliesnerKutzbach and Otto-Bliesner 1982). At 6 ka BP the extent of terrestrial ice sheets is approximately similar to the ice extent today, but summer insolation in the northern hemisphere was about 5% larger than it is today. The patterns of pollen-derived temperature estimates for eastern North America and Europe agree well with initial model results for July temperatures at 6 ka BP. These comparisons between model results and the paleoclimatic data will be a continuing focus for COHMAP research.