Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection

2.5.3 Water Vapour

Although measurement problems hinder the analysis of long-term water vapour changes (Elliott, 1995; Rind, 1998), several recent studies tend to confirm and extend the findings of lower tropospheric water vapour increases reported in the SAR. Furthermore, new analyses indicate upward trends in near-surface humidity. Knowledge about changes in water vapour at upper tropospheric and lower stratospheric levels is of great importance because strong alterations in radiative forcing can result from small absolute changes in water vapour at these levels (Chapters 6 and 7). New data presented here from the SPARC WAVAS (Stratospheric Processes and their Role in Climate / Water Vapour Assessment) project (Kley et al., 2000) are starting to cast light on changes at these levels. Note that water vapour pressure, and specific humidity (for a constant relative humidity) increase non-linearly with increasing temperature. Surface water vapour

Water vapour pressure, dew-point or relative humidity at the surface is conventionally measured using wet and dry bulb thermometers exposed in thermometer screens at climate stations. The quality of these data has been little studied. Wet bulb thermometers are not usually aspirated, so that the cooling of the wet bulb, and therefore the deduced specific or relative humidity, depends on the flow rate of air within the screen. This may often differ from the assumed airflow. Occasionally wet bulbs may dry out. Thus it is not possible to judge fully the accuracy of surface vapour pressure trends presented here.

Schönwiese et al. (1994) and Schönwiese and Rapp (1997) found small increases in surface vapour pressure over most of Europe from 1961 to 1990. The annual trends are weak. Statistically significant changes are confined to increases of about 0.5 to 1.5 hPa (relative to mean values of 12 to 15 hPa) in the southern and eastern Mediterranean region (with the largest increase in summer) and decreases of about 0.5 hPa over parts of Turkey (mainly in springtime).

Specific humidity trends over the United States were overwhelmingly positive for the period 1961 to 1995, with magnitudes of several per cent per decade, and with the largest and most statistically significant trends in spring and summer (Gaffen and Ross, 1999). Night-time specific humidity trends were generally stronger than daytime trends. Relative humidity showed smaller increases, especially in winter and spring. The specific humidity and derived dew point trends are broadly consistent, both spatially and in their day-night differences, with temperature trends. Schwartzman et al. (1998) found that the diurnal dewpoint cycle is changing over North America, with a relative decline in late afternoon and a small rise at midday.

Figure 2.26:
Trends in annual mean surface water vapour pressure, 1975 to 1995, expressed as a percentage of the 1975 to 1995 mean. Areas without dots have no data. Blue shaded areas have nominally significant increasing trends and brown shaded areas have significant decreasing trends, both at the 5% significance level. Biases in these data have been little studied so the level of significance may be overstated. From New et al. (2000).

Increases in water vapour over the former Soviet Union, Eastern China, the United States and tropical Western Pacific islands have been found in some seasons by Sun et al. (2000) in the second half of the 20th century, but with decreases in Canada in autumn. The selective character of the findings prevents any assessment of statistical significance. Wang and Gaffen (2001) found that specific humidity trends over China were overwhelmingly positive over 1951 to 1994, with the largest and most statistically significant trends in north-west China north of 35°N and west of 105°E. Trends were larger in summer and night-time trends were generally larger than daytime ones.

Recently New et al. (2000) have estimated linear trends for annual and seasonal values of surface vapour pressure over land using calculated monthly vapour pressure data from climate stations. Figure 2.26 shows trends for the 21 years from 1975 to 1995, corresponding to much of the recent period of global warming described in Section Although the uncertain quality of the data prevents any definitive conclusions about statistical significance, nominal significance of trends at the 5% level was estimated after smoothing the annual data to reduce the influence of outliers at the beginning and end of this short series. Few Southern Hemisphere data have been analysed, but Figure 2.26 shows that there have been widespread nominally significant increases in annual mean water vapour in the Northern Hemisphere. These increases are reflected in the individual seasons, although nominally significant annual mean increases are more extensive. Regional decreases over eastern Canada are explained by colder conditions in the winter half year associated with the increasingly positive phase of the North Atlantic Oscillation (Section 2.6.5). Lower-tropospheric water vapour

Radiosonde and satellite observations of water vapour above the surface have been analysed for evidence of long-term change. Both data sources have had serious data quality and temporal homogeneity problems (Elliott, 1995), although recent work to determine trends in water vapour from the surface to 500 hPa since 1973 has been based on radiosonde data judged to be largely unaffected by these problems (Ross and Elliott, 2001). Published satellite data are insufficiently homogeneous or too short in length to deduce reliable trends or low-frequency variations.

Figure 2.27:
Annually averaged trends in surface to 500 hPa precipitable water at 0000UTC for the period 1973 to 1995. Positive trends are indicated by triangles and negative trends by circles. Filled symbols indicate the trends were statistically significant at the 5% level according to the Spearman test. The two sizes of symbols give an indication of the magnitude of the trend. From Ross and Elliott (2001).

Radiosonde observations
Ross and Elliott (1996, 1998) analysed surface-to-500 hPa precipitable water over the Northern Hemisphere for 1973 to 1995 using quality-controlled data. Increases in precipitable water were found over North America except for north-east Canada. Over Eurasia, only China and the Pacific islands show coherent regional increases. The remainder of Eurasia shows a mixture of positive and negative trends, with a tendency for negative trends over Eastern Europe and western Russia. Mid-tropospheric water vapour trends tend to be of the same sign as temperature trends over North America, China, and the Pacific, but elsewhere the temperature trends are more consistently positive than the water vapour trends. Figure 2.27 summarises the results. Lower-tropo-spheric dew-point data for the period 1961 to 1995 also show increases, though smaller than those for the 1973 to 1995 period, and few are statistically significant (Ross and Elliott, 2001).

Zhai and Eskridge (1997) found increases of about 1 to 3%/decade in surface-to-200 hPa precipitable water over China for 1970 to 1990. Increases were most significant in spring. Percentage trends were larger over the 700 to 400 hPa layer than the surface-700 hPa layer. Gutzler (1996) found that specific humidity data at 1,000, 700, and 300 hPa at four western tropical Pacific radiosonde stations from 1973 to 1993 gave increases of 3 to 9%/decade, with larger percentage increases at increasing height above the surface. In contrast, Peixoto and Oort (1996) found decreases in zonal mean relative humidity between 1974 and 1988. The decreases are more marked at 300 hPa, where they are more likely to be associated with instrument changes than at lower levels, and are more pronounced at higher latitudes than in the tropics.

Other reports in this collection