Hardy et al. 2003, Vuille et al. 2003a). The local warming observed at the Quelccaya ice cap (Thompson et al. 2000a, Mark and Seltzer 2005) does correlate with the ice isotopic ratio there. In both Tibet and the Andes, there is no clear relationship between ice isotopic ratio and local accumulation on the ice caps. Thus, the ice isotopic ratios in these locations are not heavily influenced by local precipitation, but hydrologic influences retained from the lowland regions upwind may still be very important. Controls on ice isotopic ratio in equatorial Africa are not known.
The correlation of the ice isotopic ratio with temperature is very strong in the cold interiors of the polar ice sheets. Calibration is nonetheless necessary because factors like the seasonal timing of precipitation, the warm-weather bias of storms, and the atmospheric mixing of air masses may change with time. Calibration of the ice isotopic ratio thermometer is achieved over a range of timescales (Alley and Cuffey 2001 and references therein) by using weather station and satellite records for annual cycles, by using gas isotopic ratios for decadal-scale rapid climate changes, and by using borehole temperatures for centennial-to-millennial-scale climate changes. Such studies have shown polar ice isotopic ratios to be reliable thermometers. Before calibration, temperature changes inferred from ice isotopic ratio changes are accurate to within a multiple of approximately 2. Temperatures recorded by ice isotopic ratio in the polar ice sheets are representative of a broad region in the ice sheet interior and also include an imprint of temperature at much larger scales.
The time resolution of ice isotopic ratio temperature reconstructions varies from place to place. Many sites from Greenland, the Canadian Arctic, and the tropics have nearly annual resolution, whereas sites from the very dry interior of East Antarctica have only decadal resolution. Diffusional smoothing reduces the resolution from annual to a few years or more in most places, and this reduction of resolution increases backward in time. The time span covered by ice isotopic ratio records is greatest where snowfall rates are small and glacier thicknesses are greatest, and varies from several hundred thousand years in central East Antarctica to 100,000 years in central Greenland to 10,000 years in the high-altitude tropics and coastal Antarctica and Greenland. Most glaciers in the midlatitude mountains cannot provide long records of this sort because the ice mass is too rapidly removed by flow.
The four available ice cores from Tibet (Figures 6-1 and 6-2) together show that 20th century climate is anomalous relative to the preceding 1,900 years for this region (Thompson et al. 1989, 1997, 2000a, 2003, 2006, in press).2The anomaly is some combination of apparent warm conditions and weak monsoon precipitation. That a warming is part of this signal is clear, given that the anomaly is seen in the northern Tibetan records and that the monsoon-dominated southern records correlate with instrumental trends in the region (Thompson et al. 2000a; the Dasuopu core). A