Figure 2.25(i): Trends for 1900 to 1999 for the four seasons. Precipitation trends are represented by the area of the circle with green representing increases and brown representing decreases. Annual and seasonal trends were calculated using the following method. Precipitation anomalies in physical units were calculated for each station based on 1961 to 1990 normals and averaged into 5° x 5° grid cells on a monthly basis. The 1961 to 1990 monthly mean precipitation for each grid cell was added to the monthly anomalies and the resulting grid cell values summed into annual and seasonal totals. This series was converted into percentages of normal precipitation, and trends calculated from the percentages. Average trends within six latitude bands (85°N to 55°N, 55°N to 30°N, 30°N to 10°N, 10°N to 10°S, 10°S to 30°S, 30°S to 55°S) are shown in the legend of each map. The 1961 to 1990 monthly mean precipitation for the latitude band was added to the anomaly time-series and the resulting values totalled across all months within the season or year. The significance of each trend (based on a 5% level) was determined using a t-test and a non-parametric test statistic. Trends found to be significant under both tests are indicated with an asterisk.
Figure 2.25(ii): As in Figure 2.25(i) except annual trends for the three periods of changing rates of global temperature (shown in Figure 2.9) and the full period, 1900 to 1999. During the 100-year periods, calculation of grid cell trends required at least 66% of the years without missing data and at least three years of data within each decade except the first and last. During the shorter periods, calculation of grid cell trends required at least 75% of the years without missing data. Stations with more than one sixth of their data missing during the normal period and grid cells with more than one season or year without any measurable precipitation during the normal period were excluded from consideration. Due to the nature of trend estimation, it is not possible to cumulatively sum the trends for each of the three periods to obtain an overall trend.
Overall, global land precipitation has increased by about 2% since the beginning of the 20th century (Jones and Hulme, 1996; Hulme et al., 1998). The increase is statistically significant but has been neither spatially nor temporally uniform (Karl and Knight, 1998; Doherty et al., 1999). Dai et al. (1997b) found a global secular increase in precipitation separate from ENSO and other modes of variability. Data from over 20,000 stations contributed to the changes since 1900 shown in Figure 2.25. The effects of changes in windshields on winter precipitation measurements were taken into account for most mid- and high latitude observations. Dai et al. (1997b) indicated that instrumental discontinuities are unlikely to significantly impact other observations.
Mid- and high latitudes
Over the 20th century, annual zonally averaged precipitation increased by between 7 to 12% for the zones 30°N to 85°N and by about 2% between 0°S to 55°S (Figure 2.25(ii)). The increase in the Northern Hemisphere is likely to be slightly biased because adjustments have not been made for the increasing fraction of precipitation falling in liquid as opposed to frozen form. The exact rate of precipitation increase depends on the method of calculating the changes, but the bias is expected to be small because the amount of annual precipitation affected by this trend is generally only about a few per cent. Nevertheless, this unsteady, but highly statistically significant trend toward more precipitation in many of these regions is continuing. For example, in 1998 the Northern Hemisphere high latitudes (55°N and higher) had their wettest year on record and the mid-latitudes have had precipitation totals exceeding the 1961 to 1990 mean every year since 1995.
Figure 2.25(i) shows mostly increasing precipitation in the Northern Hemisphere mid- and high latitudes, especially during the autumn and winter, but these increases vary both spatially and temporally. For example, precipitation over the United States has increased by between 5 to 10% since 1900 (Figure 2.25(ii)) but this increase has been interrupted by multi-year anomalies such as the drought years of the 1930s and early 1950s (Karl and Knight, 1998; Groisman et al., 1999). The increase is most pronounced during the warm seasons. Using data selected to be relatively free of anthropogenic influences such as ground water pumpage or land use changes, several recent analyses (Lettenmaier et al., 1999; Lins and Slack, 1999; Groisman et al., 2001) have detected increases in streamflow across much of the contiguous United States, confirming the general tendency to increasing precipitation. However, Lins and Michaels (1994) found in some regions that increased streamflow did not relate well to an increase in rainfall. This has been further evaluated by Groisman et al. (2001) who show that changes in snow-cover extent also influence the timing and volume of streamflow.
Regionally, Mekis and Hogg (1999) showed that precipitation in Canada has increased by an average of more than 10% over the 20th century. Zhang et al. (2000) report an increase in Canadian heavy snowfall amounts north of 55°N and Akinremi et al. (1999) found rainfall significantly increasing in the Canadian prairies from 1956 to 1995. Multi-decadal streamflow data in Canada are not extensive, but there are no apparent inconsistencies between observed changes in streamflow or precipitation (Zhang et al., 2000).
Over the last 50 years there has been a slight decrease in annual precipitation over China (Zhai et al., 1999a), which is supported by a significant (5% confidence level) decrease in the number of rainy days (3.9%/decade; Figure 2.25 (ii)). In contrast, the area affected by the upper 10% of heaviest precipitation has significantly increased. Zhai et al. (1999b) show a significant increase in precipitation over the middle and lower reaches of the Yangtze River and west China during the latter part of the 20th century, while also detecting a declining trend in precipitation over northern China.
There have been marked increases in precipitation in the latter part of the 20th century over northern Europe, with a general decrease southward to the Mediterranean (Schönwiese and Rapp, 1997; Figure 2.25(i)). Dry wintertime conditions over southern Europe and the Mediterranean (Piervitali et al., 1998; Romero et al., 1998) and wetter than normal conditions over many parts of northern Europe and Scandinavia (Hanssen-Bauer and Førland, 2000) are linked to strong positive values of the North Atlantic Oscillation, with more anticyclonic conditions over southern Europe and stronger westerlies over northern Europe (Section 2.6.5).
Based on recent research (Bogdanova and Mescherskaya, 1998; Groisman and Rankova, 2001), the precipitation trend for the last century over the former USSR as reported by the SAR was slightly overestimated. The new results indicate that precipitation has increased since 1891 by about 5% west of 90°E for both warm and cold seasons. Georgievsky et al. (1996) also noted increases in precipitation over the last several decades over western Russia, accompanied by increases in streamflow and a rise in the level of the Caspian Sea. In eastern Russia a negative precipitation trend since 1945 is embedded in the century-long positive precipitation trend (Figure 2.25(ii); Gruza et al., 1999). Soil moisture data for large regions of Eurasia (Robock et al., 2000) show large upward trends. The rate of increase is more than 1 cm/decade in the available soil moisture in the top 1 m of soil. These large positive trends occur simultaneously with positive trends in temperature that would normally reduce soil moisture. Increases in precipitation (and cloud cover, Section 2.5.5) are believed to have more than compensated for the increased losses due to evapotranspiration.
An analysis of rainfall data since 1910 by Haylock and Nicholls (2000) reveals a large decrease in total precipitation and related rain days in south-western Australia. Annual total rainfall has increased over much of Australia with significant increases of 15 to 20% in large areas. The increase in total rainfall has been accompanied by a significant 10% rise in the average number of rain days over Australia (Hennessy et al., 1999). Elsewhere in the Southern Hemisphere, a long-term increase in precipitation in Argentina has been observed for the period 1900 to 1998 (Figure 2.25(i); Dai et al., 1997b).
Tropics and sub-tropics
The increase in precipitation in the mid- and high latitudes contrasts with decreases in the northern sub-tropics (with marginal statistical significance) which were largely responsible for the decade-long reduction in global land precipitation from the mid-1980s through the mid-1990s. Since the SAR, record low precipitation has been observed in equatorial regions, while the sub-tropics have recovered from their anomalously low values of the 1980s.
Regionally positive but non-significant trends have occurred in the rainy season rainfall in north-east Brazil and northern Amazonia (Marengo et al., 1998). River data from northern Amazonia indicate wetter periods in the mid-1970s, and in 1990, as well as drier periods between 1980 to 1990, consistent with rainfall anomalies. Northern Amazonian rainfall appears to be modulated by multi-decadal climate variations.
There is little evidence for a long-term trend in Indian monsoonal rainfall but there are multi-decadal variations (Kumar et al., 1999a,b). From 1906 to about 1960, monsoonal rainfall increased then decreased through 1974 and has increased since (see Section 2.6). In central America for much of the period from the early 1940s to present, western Mexico has experienced an increasingly erratic monsoonal rainfall (Douglas and Englehart, 1999).
Since 1976, increases in precipitation in the South Pacific have occurred to the north-east of the South Pacific Convergence Zone (SPCZ) while decreases have occurred to its south-west (Salinger et al., 1996). Manton et al. (2001) found significant decreases in rain days since 1961 throughout Southeast Asia and western and central South Pacific, but increases in the north of French Polynesia and Fiji.
Streamflow data for major rivers in south-eastern South America for the period 1901 to 1995 show that streamflow has increased since the mid-1960s, and was accompanied by a significant decrease in the amplitude of the seasonal cycle of most of those rivers (Garcia and Vargas, 1998; Genta et al., 1998). Figure 2.25(i) shows increases in precipitation since 1900 along the South American eastern coastal areas, with less extensive increases since 1976.
There has been a pattern of continued aridity since the late 1960s throughout North Africa south of the Sahara. This pattern is most persistent in the western region. The driest period was in the 1980s with some recovery occurring during the 1990s, particularly in the easternmost sectors where rainfall in some years was near or just above the long-term mean (Nicholson et al., 2000). Southern Africa was relatively moist in the 1950s and 1970s (Nicholson et al., 2000); but Hulme (1996) found significant decreases in precipitation being observed since the late 1970s. Early 2000, however, has seen flood-producing rains in the eastern part of southern Africa.
Other reports in this collection