Glacial landforms and drift stratigraphy in central Magellan Strait, southernmost Chile, document repeated fluctuations during the last glacial cycle of outlet lobes from an ice cap centered over the southern Andes. The lobes developed comparatively low-gradient profiles because of low basal shear stresses over soft deformable beds and this made them sensitive to even small-scale changes in the mass balance. Such low profiles and rapid calving in deep proglacial lakes during deglaciation may have made the Magellan ice lobe particularly responsive to climatic fluctuations during the last glacial cycle, and to advance and retreat over considerable distances. Study of the glacial landforms and drift stratigraphy has led to the identification of at least five glacier advances to limits at and south of the Segunda Angostura. Fragments of mollusc shells contained in basal till indicate marine incursions between some advances, thus documenting extensive deglaciation. A partial chronology based on amino acid studies and radiocarbon dating suggests that five of these advances occurred during the last glacial cycle. The most extensive advances may have culminated during substages of marine isotope stage 5 (substage 5b or 5d) and/or during stage 4. Slightly less extensive advances occurred between ca. 28,000 and 14,000 yr B.P.

The history of the Late Weichselian northwestern Barents Shelf, including western Svalbard, has been investigated by provenance/sedimentologist studies of five cores from the continental shelf and slope west of Svalbard. The chronostratigraphy of the cores is based on AMS 14C dates and oxygen isotope analyses. Interpretations of the cores suggest that the ice sheets of western Svalbard and northwestern Barents Sea experienced advances and retreats in two steps. The first significant ice advance beyond the present coastline occurred ca. 22,000 14C yr B.P. and was followed by an ice advance to the shelf edge ca. 18,000 14C yr B.P. Ice recession from the outer shelf and the southwestern Barents Sea began 14,800 14C yr B.P. and was followed by a second ice recession between 13,000 and 12,000 14 C yr B.P. during which ice withdrew from the inner shelf. A minor readvance of the ice sheet on the shelf west of Svalbard occurred close to 12,400 14C yr B.P. The first deglaciation event was associated with release of icebergs containing ice-rafted detritus, while the later episode also included significant amounts of meltwater and fine-grained sediment.

Ice sheet modeling is an essential tool for estimating the effect of climate change on the Greenland ice sheet. The large spatial and long-term temporal scale of ice sheet models limits the amount of data which can be used to test model results. A framework for the analysis of glacial geological data to test ice sheet models is illustrated by a case study in west Greenland. A geological scenario of ice margin positions since the last glacial maximum is based on a review of existing literature and new datings of moraine systems. Ages of moraine systems and associated accuracy ranges are interpolated to a 1 × 1 km grid over an area of about 57,500 km2. Resampling to a 20 × 20 km grid on an ice sheet model coordinates permits a quantitative comparison with modeled ice marginal positions. In view of the uncertainties in the geological scenario, with moraine system ages having an absolute uncertainty of 700 to 3300 cal yr, the Greenland ice sheet model of Huybrechts provides a reasonable simulation of the deglaciation pattern in central west Greenland. Modeled timing of the position of the ice margin generally precedes the geological record by 900 yr. The difference between geology and model is large for areas without proper geological information and small, but still about 500-950 yr, for well-dated moraine systems. Model deficiencies in west Greenland are probably related to the forcing function driving ablation estimates, and especially the forcing function for sea level and the description of calving processes should he reviewed critically. The simplified topography used by the model could also induce errors. Although topography used in models neglects possible important features such as ice streams, we do not find any significant trend between differences and relief amplitude. Missing moraines on the southern part of the shelf and absence of geological information from beneath the present ice sheet affect the character and quality of the geological scenario.

Particle-size measurements of some typical loess-soil samples taken in different localities of the Chinese Loess Plateau demonstrate that the grain size ratio of <2 μm/>10 μm (%) can be used as an indicator of variations in intensity of the East Asian winter monsoon winds. Grain-size curves of the Baoji and Weinan sections show that this proxy indicator is very sensitive to loess-soil alterations. Analytical results also suggest that during soil-forming periods, eolian dust accumulation was still substantial and, hence, loess deposition can be regarded as a nearly continuous process during the Quaternary period. In this study we compared the Baoji grain-size time series with the SPECMAP marine isotope record with the objective of elucidating the dynamic linkage between changes in global ice volume and the winter monsoon circulation. Both records show good agreement at both time and frequency domains. In particular, the winter monsoon variations are also dominated by a 100,000 yr period over the past 800,000 yr. It is thus inferred that direct local insolation forcing could be less important in driving the East Asian winter monsoon variability, and, alternatively, variations in glacial-age boundary conditions may have played a key role in modulating and pacing its strength and timing.

The Markhida Moraine is a broad zone of ice marginal landforms and deposits running E-W across the Pechora lowland in northern Russia. It was formed along the southern margin of an ice sheet advancing onto the coast from the Kara Sea. Previous researchers (Grosswald, 1980, 1983, 1994a: Grosswald et al., 1974) concluded that the moraine was formed during a readvance of the Barents-Kara ice sheet after 9000 14C yr B.P. This conclusion was based on the occurrence of deformed lacustrine sediments dated 9000 to 10,000 14C yr B.P. between two presumed basal tills in the type section at Markhida. A reinvestigation of the section by the authors revealed only one basal till overlain by lacustrine sediments and diamictons, yielding radiocarbon dates between 9000 and 10,000 14C yr B.P. Luminescence dating from the sand underlying the till gave Middle Weichselian ages. Deformation structures and fabric in the sediments overlying the till solely reflect gravity-driven processes, not glaciotectonic deformation as claimed by previous investigators. We conclude that most of these sediments were finally deposited during very slow (retarded) melting of a large body of buried glacier ice and the underlying permafrost. Thus, the Markhida Moraine, as an ice marginal feature, was formed prior to 10,000 14C yr B.P., probably during the Late Weichselian, but possibly as early as the Middle Weichselian.

As part of a regional investigation of the aminostratigraphy of central European loess deposits, we have sampled and analyzed the amino acid composition of fossil gastropod shells from stratotype loess profiles at Mende, Basaharc, and Paks, plus the profile at Süttö, in an effort to correlate independently loess-paleosol formations of the "Young Loess" of Hungary with loess units elsewhere in the region. The measured extent of isoleucine epimerization in fossil gastropod shells preserved in loess sediments was used to evaluate the relative ages of stratigraphic units and correlate between discontinuous exposures. Based on amino acid epimerization ratios, we propose a new chronology for the standard "Young Loess" profile of Hungary. Amino acid geochronological data suggest that paleosol complexes MF, BD1 + 2, and BA formed during interglaciations, rather than interstades. Correlations are suggested with PK 2 + 3 (last interglaciation), PK 4 (penultimate interglaciation), and PK 5, respectively, from previously described Czech and Slovak localities.

Measured 18O/16O ratios from the Greenland Ice Sheet Project 2 (GISP2) ice core extending back to 16,500 cal yr B.P. provide a continuous record of climate change since the last glaciation. High-resolution annual 18O/16O results were obtained for most of the current millennium (A.D. 818-1985) and record the Medieval Warm Period, the Little Ice Age, and a distinct 11-yr 18O/16O cycle. Volcanic aerosols depress central Greenland annual temperature (∼1.5°C maximally) and annual 18O/16O for about 4 yr after each major eruptive event. On a bidecadal to millennial time scale, the contribution of solar variability to Holocene Greenlandic temperature change is ∼0.4°C. The role of thermohaline circulation change on climate, problematic during the Holocene, is more distinct for the 16,500-10,000 cal yr B.P. interval. (Analogous to 14C age calibration terminology, we express time in calibrated (cal) yr B.P. (A.D. 1950 = 0 cal yr B.P.)). The Oldest Dryas/Bølling/Older Dryas/Allerød/Younger Dryas sequence appears in great detail. Bidecadal variance in 18O/16O, but not necessarily in temperature, is enhanced during the last phase of lateglacial time and the Younger Dryas interval, suggesting switches of air mass transport between jet stream branches. The branched system is nearly instantaneously replaced at the beginning of the Bølling and Holocene (at ∼14,670 and ∼11,650 cal yr B.P., respectively) by an atmospheric circulation system in which 18O/16O and annual accumulation initially track each other closely. Thermodynamic considerations of the accumulation rate-temperature relationship can be used to evaluate the 18 O/16O-temperature relationship. The GISP2 ice-layer-count years of major GISP2 climate transitions also support the use of coral 14C ages for age calibration.

Climatic changes inferred from lacustrine deposits and lake-level fluctuations in northwestern and central China are mainly based on paleoclimatic records from the Tibetan Plateau, while there is still a lack of data relating to the semiarid/arid desert regions of Inner Mongolia. In the Tengger Desert, different paleolake levels at Baijian Hu are documented by six paleoshorelines and stratified lake carbonates. The highest lake level occurred ca. 39,000 14C yr B.P. and prevailed over about 16,000 km2. From sediment structure, geochemical composition, and ostracods we infer humid/cool conditions until 23,000 14C yr B.P. In the northern Badanjilin Desert at Gaxan Nur/Sogu Nur, high lake levels can be deduced from mollusc-bearing paleobeaches and lake carbonates, which have been dated to 34,000 14C yr B.P. and indicate a lake that covered some 32,000 km2. After ca. 20,000 yr B.P. the climate became dry with increased eolian activity and decreasing lake levels. Reestablishment of wet conditions occurred ca. 13,000 yr B.P. The Holocene is represented by stratified lake deposits that alternate with fluvial and eolian deposits, indicating a longterm oscillating trend toward arid conditions. The existence of widespread freshwater lakes during the late Pleistocene indicates a semihumid climate without an accompanying fall in temperature to arctic conditions.

Variations in the amount and composition of sedimentary organic matter in glacial and interglacial deposits of the last 2.7 myr correlate with late Cenozoic climatic and oceanographic changes in the Norwegian-Greenland Sea. These variations are predominantly caused by the changing supply of terrestrial and reworked organic matter. The highest amounts of terrestrial organic particles (macerals) and of reworked coal clasts in glacial and early deglacial diamictons are closely related to glacial erosion of Mesozoic strata that crop out along the Scandinavian Shelf. The first occurrence of coal clasts at 2.53 myr demonstrates an initial advance of continental ice margins to these source areas. The establishment of anoxic conditions at the sea floor during diamicton deposition probably reflects an increased vertical flux of labile organic matter due to lithogenic adsorbtion followed by an almost complete mineralization at the water/sediment interface. The content of marine organic matter was continuously low over the past 2.7 myr except for past interglacial highstands (i.e., isotopic event 5.5.1), suggesting persistent diagenetic degradation of the labile organic fraction. The marine organic matter exclusively consists of residual dinoflagellate cysts and fragments of them.

A sequence of northern Lake Michigan beach ridges records lake-level fluctuations that are probably related to changes in late Holocene climate. Historically, episodes of falling and low lake level associated with regional drought led to the formation of dune-capped beach ridges. The timing of prehistoric ridge formation, estimated by radiocarbon dating of plant macrofossils from early-successional dune species, shows that return periods of inferred drought, averaged for time intervals of 100 to 480 yr, ranged between 17 and 135 yr per drought during the last 2400 yr. In five of ten of these time intervals, the average return period ranged between 17 and 22 yr per drought. These intervals of frequent ridge formation and drought were associated with the development of parabolic dunes, which is indicative of high lake level and moist climate. This seeming paradox suggests that unusually moist decades alternated with unusually dry decades during these time intervals. Regional water balance probably varied less during the time intervals when ridges formed less often and the lake produced no evidence of high level.

A stratigraphic analysis of buried soils within the Grand Sable Dunes, a dune field perched 90 m above the southern shore of Lake Superior, reveals a history of eolian activity apparently linked with lake-level fluctuations over the last 5500 yr. A relative rise in the water plane of the Nipissing Great Lakes initially destabilized the lakeward bluff face of the Grand Sable plateau between 5400 and 4600 14C yr B.P. This led to the burial of the Sable Creek soil by eolian sediments derived from the bluff face. Subsequent episodes of eolian activity appear to be tied to similar destabilizing events; high lake levels may have initiated at least four and perhaps eleven episodes of dune building as expressed by soil burials within the dunes. Intervening low lake levels probably correlate with soil profile development, which varies from the well-developed Sable Creek Spodosol catena to thin organic layers containing in-place stumps and tree trunks. Paleoecological reconstructions available for the area do not imply enough climatic change to account for the episodic dune activity. Burial of soils by fine-fraction sediments links dune-building episodes with destabilization of the lower lake-facing bluff, which is rich in fines.

New radiocarbon ages for Sierra Nevada deglaciation, the first 10 Be measurements from the Laurentide terminal moraine, and calculations based on paleomagnetic field strength have the potential to substantially improve the accuracy of cosmogenic age estimates. Specifically, three new constraints apply to the interpretation of measured abundances of in situ produced cosmogenic 10Be and 26Al: (1) A suite of minimum-limiting radiocarbon dates indicates that the Sierra Nevada was deglaciated at least several thousand years earlier than assumed when Nishiizumi et al. (1989) first calibrated 10Be and 26 Al production rates based on polished bedrock surfaces in the range, with retreat beginning by 18,000 cal yr B.P. and completed by 13,000 cal yr B.P. (2) Concentrations of 10Be in moraine boulders and glacier-polished bedrock in New Jersey show little variance (10%, 1σ) and can be used to calculate a preliminary 10Be production rate (integrated over the past 21,000-22,000 cal yr B.P. at 41°, 200-300 m altitude) that is about 20% lower than currently accepted. (3) Calculations of the effect of past geomagnetic field-strength variations on production rates suggest that the use of temporally averaged production rates may generate age errors of >20%; however, cosmogenic exposure ages can be corrected for this effect, although the corrections currently are imprecise. Many previously reported late-Pleistocene 10Be and 26Al exposure ages are probably too young and are less accurate and less precise than implied by reported uncertainties. The discrepancy between accepted production rates and those calculated from Laurentide exposures, when considered together with the Sierran deglacial chronology and the model results, suggest that correlations between cosmogenic and other numerical ages, especially for brief events like the Younger Dryas and Heinrich events, will not be robust until temporal variations and the altitude/latitude scaling of production rates are fully understood and quantified at levels comparable to current analytic uncertainties (∼3%).

New geochemical data on tephra samples from a layer present at several archeological sites in India support correlation of this layer to the Youngest Toba Tuff, erupted from northern Sumatra about 74,000 yr ago. The data show that the Indian tephra layer is not a correlative of older tephra erupted from Toba, as has been suggested on the basis of artifact assemblages. Previously published geochemical data on the Indian tephra beds was based on bulk ash samples containing mineral and clay contaminants, and the resulting variability in analyses did not allow identification or discrimination of individual eruptive events. Our new data were collected on individual glass shards and small, purified glass separates which have greater resolving power in fingerprinting. Acheulian and Paleolithic artifacts found at some of the Indian tephra sites do not reflect the antiquity of the tephra bed, as they occur in fluvial sediments and may be reworked.