Ice cap modeling constrained by empirical studies provides an effective way of reconstructing past climates. The former Patagonian ice sheet is in a climatically significant location since it lies athwart the Southern Hemisphere westerlies and responds to the latitudinal migration of climatic belts during glacial cycles. A numerical model of the Patagonian ice cap for the last glacial maximum (LGM) is developed, which is time-dependent and driven by changing the mass balance/altitude relationship. It relies on a vertically integrated continuity model of ice mass solved over a finite difference grid. The model is relatively insensitive to ice flow parameters but highly sensitive to mass balance. The climatic input is adjusted to produce the best fit with the known limits of the ice cap at the LGM. The ice cap extends 1800 km along the Andes and has a volume of 440,000 km3. During the LGM the equilibrium line altitude (ELA) was lower than at present by at least 560 m near latitude 40°S, 160 m near latitude 50°S, and 360 m near latitude 56°S. The latitudinal variation in ELA depression can be explained by an overall fall in temperature of about 3.0°C and the northward migration of precipitation belts by about 5° latitude. Annual precipitation totals may have decreased by about 0.7 m at latitude 50°S and increased by about 0.7 m at latitude 40°S. The ELA rises steeply by up to 4 m per kilometer from west to east as the westerlies cross the Andes and this prevents ice growth to the east. The limited decrease in temperature during the LGM could be related to the modest migration of the Antarctic convergence between South America and the Antarctic Peninsula.

The recessional history of Niagara Falls in the present Niagara Gorge during the postglacial period has been a focus of study throughout this century. Radiocarbon ages of clam shells suggest that Niagara Falls migrated very slowly around the narrowed gorge section at Niagara Glen from 10,500 to 5500 yr B.P., when upper Great Lakes water bypassed Lake Erie and flowed to the Ottawa River via the outlet at North Bay, Ontario. Prior to that interval, river discharge and recession rates were similar to those at present, and similar rates resumed after 5200 yr B.P. By about 4500 yr B.P., the present gorge had intersected a buried gorge of the pre-late Wisconsinan Niagara River (St. Davids Gorge). The sediment derived from the excavated buried valley fill may be present as a distinct marker horizon in the sediments in southwestern Lake Ontario.

Loess of the Driftless Area includes four distinct late Quaternary lithostratigraphic units: the Wyalusing (new), Loveland, Roxana, and Peoria formations. Erosion has removed parts or all of the pre-late Wisconsian loess at many sites. These formations consist largely of loess and retransported loess that typically occurs on uplands, terraces, and valley margins in the region. The oldest widespread formation (probably > 125,000 yr B.P.) is here formally named the Wyalusing Formation. The three younger formations correlate with the Loveland (probably > 125,000 yr B.P.), Roxana (55,000-25,000 yr B.P.), and Peoria (25,000-12,000 yr B.P.) loesses recognized elsewhere in the midcontinental United States. The soil/stratigraphic morphology of the Wyalusing-Loveland and Roxana-Peoria couplets appears to represent two distinct loess sedimentation sequences related to major expansion and contraction of the Laurentide Ice Sheet during oxygen isotope stages 6 and 3-2, respectively. The occurrence-frequency of loess units is inversely related to age, illustrating the erosional nature of the Driftless area landscape. The occurrence of four loess units at some sites in Driftless Area stratigraphic sections corroborates loess stratigraphy along the length of the Mississippi Valley, where typically not more than four or five loess units are found and they represent only the late and middle Pleistocene (<790,000 yr B.P.). Past climatic conditions, which favored erosion of loess, as well as a temporally erratic spatial extent of former continental ice sheets in North America, which provided the dust supply, probably account for the low number of loess units along the Mississippi Valley.

Buried soils are described from the floors of four kettles on Pinedale piedmont moraines of the southwestern Wind River Range, Wyoming, near the type locality of the Pinedale Till. The buried soils indicate that a previously unreported episode of slope erosion has occurred along adjacent catenas on some of the moraines in this region. Radiocarbon ages of the buried soils indicate that slope erosion occurred during the middle Holocene from 8540 ± 190 to 4800 ± 60 14C yr B.P. The presence of buried soils in moraine kettles indicates that profiles on the crests and backslopes of some of the moraines in this region are not the products of continuous post-Pinedale soil formation. Rather, the crest and backslope soils on Pinedale moraines result from two periods of soil development separated by an interval of erosion. Thus, soils on crests and backslopes are considered to be welded profiles. Relative-age studies that use soil data to estimate the age of moraines in this region must take into account the possibility that a mid-Holocene erosion episode affected the genesis and morphology of soils on the moraines. This study corroborates results of previous work that suggests that data from complete catenas are more useful for estimating the characteristic soil development on Quaternary moraines than are data derived from crests alone.

Soil catenas on three moraines in each of two areas with different climate were studied to determine (a) downcatena soil differentiation with climate and time and (b) their usefulness in estimating the relative ages of the underlying deposits. In the dry area (mean annual precipitation (MAP), ca. 0.5 m) all soils have A/Bw/C profiles formed in loess/till. Their similarity in morphology and in most chemical characteristics with catena position and age suggests that the low MAP does not result in much redistribution of water, elements, or sediment downcatena. This similarity also suggests ages close to each other and correlation with the Otiran Glaciation (oxygen-isotope stages 2 and 4). In the wet area (MAP ca. 3 m) the soils also formed in loess/till and with time (a) soil morphology progresses from A/Bw/C to A/E/B/C, (h) reduced properties intensify, especially in the downcatena profiles, (c) citrate-bicarbonate-dithionite-extractable Fe displays accumulation in the youngest catena followed by loss in the older catenas, and (d) downcatena trends in total chemical data display marked losses of the more mobile elements. These data demonstrate that although catena development is rapid in a wet climate, the downcatena contrast can be muted with time due to a change from processes dominated by oxidation to those dominated by reduction. Soil catena properties in the wet area are sufficiently different to not refute age estimations suggested by Suggate (1990): youngest moraine, oxygen-isotope stage 2; intermediate moraine, isotope stage 4; and oldest moraine, isotope stage 10. An unresolved problem in both areas is the possibility of soil erosion to foul soil-age relations.

The radiocarbon content and stable isotope composition of soil carbonate are best described by a dynamic system in which isotopic reequilibration occurs as a result of recurrent dissolution and reprecipitation. Depth of water penetration into the soil profile, as well as soil age, determines the degree of carbonate isotope reequilibration. We measured δ13C, δ18O and radiocarbon content of gravel rinds and fine (<2 mm) carbonate in soils of 3 .different ages (1000, 3800, and 6300 14 C yr B.P.) to assess the degree to which they record and preserve a climatic signal. In soils developing in deposits independently dated at 3800 and 6300 radiocarbon yr B.P., carbonate radiocarbon content above 40 cm depth suggests continual dissolution and reprecipitation, presumably due to frequent wetting events. Between 40 and 90 cm depth, fine carbonate is dissolved and precipitated as rinds that are not redissolved subsequently. Below 90 cm depth in these soils, radiocarbon content indicates that inherited, fine carbonate undergoes little dissolution and reprecipitation. In the 3800- and 6300-yr-old soils, δ13C in rind and fine carbonate follows a decreasing trend with depth, apparently in equilibrium with modern soil gas, as predicted by a diffusive model for soil CO2. δ18O also decreases with depth due to greater evaporative enrichment above 50 cm depth. In contrast, carbonate isotopes in a 1000-yr-old deposit do not reflect modern conditions even in surficial horizons; this soil has not undergone significant pedogenesis. There appears to be a lag of at least 1000 but less than 3800 yr before carbonate inherited with parent material is modified by ambient climatic conditions. Although small amounts of carbonate are inherited with the parent material, the rate of pedogenic carbonate accumulation indicates that Ca is derived primarily from eolian and rainfall sources. A model describing carbonate input and radiocarbon decay suggests that fine carbonate below 90 cm is mostly detrital (inherited) and that carbonate rinds have been forming pedogenically at a constant rate since alluvial fans were deposited.

Few of the marine terraces along the Pacific coast of North America have been dated using uranium-series techniques. Ten terrace sequences from southern Oregon to southern Baja California Sur have yielded fossil corals in quantities suitable for U-series dating by alpha spectrometry. U-series-dated terraces representing the ∼80,000 yr sea-level high stand are identified in five areas (Bandon, Oregon; Point Arena, San Nicolas Island, and Point Loma, California; and Punta Banda, Baja California); terraces representing the ∼125,000 yr sea-level high stand are identified in eight areas (Cayucos, San Luis Obispo Bay, San Nicolas Island, San Clemente Island, and Point Loma, California; Punta Bands and Isla Guadalupe, Baja California; and Cabo Pulmo, Baja California Sur). On San Nicolas Island, Point Loma, and Punta Bands, both the ∼80,000 and the ∼125,000 yr terraces are dated. Terraces that may represent the ∼105,000 sea-level high stand are rarely preserved and none has yielded corals for U-series dating. Similarity of coral ages from midlatitude, erosional marine terraces with coral ages from emergent, constructional reefs on tropical coastlines suggests a common forcing mechanism, namely glacioeustatically controlled fluctuations in sea level superimposed on steady tectonic uplift. The low marine terrace dated at ∼125,000 yr on Isla Guadalupe, Baja California, presumed to be tectonically stable, supports evidence from other localities for a +6-m sea level at that time. Data from the Pacific Coast and a compilation of data from other coasts indicate that sea levels at ∼80,000 and ∼105,000 yr may have been closer to present sea level (within a few meters) than previous studies have suggested.

May-June (MJ) and April-July (AJ) precipitation at Huashan in north-central China has been reconstructed for the period A.D. 1600 to 1988 using tree-ring density and width from Pinus armandii. MJ precipitation (based on ring width and maximum latewood density) calibrated and cross-validated against local instrumental data more strongly than AJ precipitation (based only on ring width). A major drought was reconstructed for the mid- and late 1920s, confirmed by local documentary sources. This drought (culminating in 1929) was the most severe of the 389-yr period for MJ and second most severe for AJ, after an event ending in 1683. Neither reconstruction shows much spectral power at frequencies lower than 1 in 10 yr, but both show concentrations of power between 2.1 and 2.7 yr and 3.5 to 9 yr. There are significant correlations between the two reconstructions and a regional dryness/wetness index (DW) based on documentary sources, particularly at high frequencies. These correlations are focused in the 7.6- to 7.3-, 3.8- to 3.6-, and 2.5-yr periods. Using singular spectrum analysis, quasiperiodic behavior with a period close to 7.2 yr was identified in the MJ precipitation reconstruction and in the DW index based on documents.

Major components of most southern African archaeological sites are stone, bone, and charcoal. A new technique for climate reconstruction utilizes measurements of vessel size and frequency in the cross-sectional xylem anatomy of archaeological charcoal from Collingham Shelter and Mhlwazini Cave in the Natal Drakensberg. Previous wood anatomical studies have shown that links exist among vessel diameter, vessel frequency and climate. The present study demonstrates that in relation to rainfall, vessel diameter in the species Protea caffra and Protea roupelliae correlated positively, whereas vessel frequency correlated negatively. In P. roupelliae, mean vessel diameter increases from 46 to 62 μm along a rainfall gradient ranging from 760 to 1665 mm. The significant correlations between rainfall and tangential vessel diameter for a charred sample of P. roupelliae suggest that such measures on an archaeological charcoal sample may be used to reconstruct rainfall patterns through time. Using nine assemblages of archaeological charcoal, generalized patterns of wetter and drier periods can be postulated. Comparison with contemporary values indicates that at 200 and 2400 yr B.P. the area near the archaeological sites was wetter than at present. A dry phase occurred between 1300 and 300 yr B.P. Values for the contemporary wood sample are the lowest observed, indicating that present conditions are much drier than those at any time within the last ca. 2000 yr. Dating resolution however, is insufficient to allow more-detailed interpretation of rainfall conditions over the past 2000 yr.