of the range decrease simulated by GCMs, however, is only 10 percent of the magnitude of the simulated mean average temperature increase, whereas the observational record shows nearly equal changes of the daily temperature range and the mean temperature. The clear-sky daytime forcing by sulfate and carbonaceous aerosols, and other indirect effects such as increased cloudiness and albedo, operate in a sense that may reduce the daily temperature range; however, the decadal changes of this geographic and seasonally varying forcing are not well known. At present we do not have an adequate explanation for the observed decrease in the temperature range, or for the lack of daytime warming, yet these changes are clearly important, both scientifically and practically.
The mean monthly maximum and minimum temperature is derived from an average of the daily maximum and minimum temperatures. The monthly mean diurnal temperature range (DTR) is defined as the difference between the mean monthly maximum and minimum temperatures. The dearth of appropriate data bases that include information on the daily or mean monthly maximum and minimum temperature has previously impeded our ability to investigate changes in these quantities. The problem has historical roots. It arises because the climatological data that are made available internationally by national meteorological and/or climate data centers are usually derived from the monthly climate summaries (CLIMAT messages) on the Global Telecommunications System (GTS), which do not include information on the maximum or minimum temperatures. The GTS is the principal means by which near-real-time, in situ, global-climate data are exchanged. Moreover, the problem has been exacerbated because the World Meteorological Organization's retrospective data collection projects such as World Weather Records and Monthly Climatic Data of the World have always been limited to mean monthly temperatures. This has forced climatologists interested in maximum and minimum temperatures either to develop historical data bases on a country-by-country basis (Karl et al., 1991 a) or to try to work with the hourly GTS synoptic observations. The former is a slow, painstaking process, and the latter has been limited by poor data quality and metadata (information about the data) and records of short duration (Shea et al., 1992).
The first indication that there might be important large-scale characteristics related to changes of the mean daily maximum and minimum temperatures was reported by Karl et al. (1984a). Their analysis indicated that the DTR was decreasing at a statistically significant rate at many rural stations across North America. Because of data accessibility problems, subsequent empirical analyses continued to focus on data from North America over the next several years (Karl and Quinn, 1986; Plantico et al., 1990). By 1990, however, a United States/People's Republic of China (PRC) bilateral agreement organized by the U.S. Department of Energy and the PRC Academy of Sciences provided the opportunity to analyze maximum and minimum temperatures from the PRC. Also about this time, the Intergovernmental Panel on Climate Change (IPCC) made arrangements with the Australian National Climate Centre to analyze maximum and minimum temperature data from southeastern Australia. The IPCC (1990) reported a significant decrease in the DTR from both of these regions. Meanwhile, work from another data exchange agreement, a bilateral between the United States and the former Soviet Union, came to fruition as a data set of mostly rural maximum and minimum temperatures was developed for the former Soviet Union. Karl et al. (1991a) reported on the widespread decrease of the DTR over the former Soviet Union, China (not including Tibet), and the contiguous United States, as was highlighted by the IPCC (1992).
Additional data from other countries and updates to previous analyses have now been analyzed. Data from the eastern half of Australia, Canada, Alaska, Sudan, Japan, some Pacific island stations, South Africa, and a few other long-term stations in Europe have now been included. Figure 1 shows the area of the globe that has now been analyzed for differential changes of the maximum and minimum temperature. This area now covers over 50 percent of the Northern and 10 percent of the Southern Hemisphere land masses, but still only about 37 percent of the global land mass.