Palaeoclimate proxy evidence (tree rings, lake sediments and pollen evidence) has been used to estimate variability in drought and precipitation patterns in past centuries. Much of the recent research has emphasised the North American region (e.g., Cook et al., 1999a), where a key conclusion is that the range of regional drought variability observed during the 20th century may not be representative of the larger range of drought evident in past centuries (Laird et al., 1996; Woodhouse and Overpeck, 1998). Hughes and Graumlich (1996) and Hughes and Funkhouser (1999) provide evidence of multi-decadal mega-droughts in the western Great Basin of North America in the 10th to 14th centuries. Nonetheless, the 20th century dust bowl still stands out as the most extreme drought of the past several centuries, the period when North American continental scale reconstruction is possible. Swetnam and Betancourt (1998) argue that recent spring wetness in the American south-west is greater than that observed in at least the last thousand years. Evidence of significant changes in regional hydroclimatic patterns is not limited, however, to North America. Stine (1994) argues that enhanced drought conditions occurred synchronously in South America. Ice accumulation at Quelccaya in the Andes, and on the Dunde Ice Cap on the Tibetan Plateau (Thompson, 1996) was slower in the first half of the last millennium than the last 500 years, but 500-year averages are not easily related to the palaeo-temperature data (Figure 2.21). Pollen evidence indicates significant changes in summer rainfall patterns in China in the earlier centuries of the past millennium (Ren, 1998). The relationship between such past changes in regional drought and precipitation patterns, and large-scale atmospheric circulation patterns associated with ENSO, for example, is an area of active current research (e.g., Cole and Cook, 1998).
The strong spatial variability inherent in precipitation requires the use of estimates based on satellite observations for many regions. Thus satellite data are essential to infer global changes in precipitation, as the oceans account for 70% of the global surface area. Since adequate observations were not made until the early 1970s, no satellite-based record is sufficiently long to permit estimates of century-long changes. The first satellite instrument specifically designed to make estimates of precipitation did not begin operation until 1987. At this time three data sets are available: (a) the Global Precipitation Climatology Project (GPCP) product, which spans the period from 1987 to the present (Huffman et al., 1997); (b) the CPC Merged Analysis of Precipitation (CMAP) product, covering the period from 1979 to 1998 (Xie and Arkin, 1997); and (c) MSU-derived precipitation estimates since 1979 (Spencer, 1993). While the period from 1987 appears to be well observed, it is too short to draw conclusions regarding decadal-scale variations. The longer CMAP data set assumes that the various satellite-derived estimates have no trend over the period, and hence no longer time-scale conclusions are possible. Nonetheless, analyses of the CMAP product and associated data from the NCEP reanalysis project indicate that there have been substantial average increases in precipitation over the tropical oceans during the last twenty years, related to increased frequency and intensity of ENSO (Trenberth et al., 2001). ENSO conditions are not related to positive precipitation anomalies everywhere over the tropical oceans (e.g., south-western Tropical Pacific).
Other reports in this collection