The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Natural Climate Variability on Decade-to-Century Time Scales
quality assurance. The degree to which precautionary measures have been taken to minimize data inhomogeneities varies considerably from country to country. In the United States, a fixed network of stations in the Historical Climatology Network is used (Karl et al., 1990), which largely consists of rural stations that have been adjusted when necessary for random station relocations, changes in instrument heights, systematic changes in observing times (Karl and Quinn, 1986; Karl et al., 1986), the systematic change in instruments during the mid—and late 1980s (Quayle et al., 1991), and increases in urbanization (Karl et al., 1988). The potential warm bias of the maximum introduced by the H083 series of thermometer (Gall et al., 1992) is not a factor in this network, since the H083 instrument was not used in the rural cooperative network.
In the former Soviet Union, the fixed network of 165 stations consists of rural stations (1990 populations less than 10,000 and local surroundings free of urban development). The data from the former Soviet Union have not been adjusted for any random or systematic inhomogeneities. Station histories, however, indicate that there have not been systematic changes in that network's operation over the course of the past 50 years.
In Canada, the results reported here are derived from a set of 227 rural stations (those in areas with population less than 10,000). These data were selected from a network of 373 principal stations, but a large number of urban areas and stations that relocated to airports were eliminated from the analysis.
In Alaska, a network of 39 stations was used. It included most stations that had been operating in that state since the early 1950s, with the exception of the stations in the major cities of Juneau, Fairbanks, and Anchorage. Once again no attempt was made to adjust for station relocations, and the stations consist of a mix of instrument types with some changes at specific sites.
Station histories from China do not reflect any systematic changes in instrumentation, instrument heights, instrument shelters, or observing procedures relative to the maximum and minimum temperature. Our analysis is based on a subset of the more than 150 stations available to us. No attempt was made to correct for random station relocations in the fixed network of 44 stations we finally selected from the larger network. The potential impact of urbanization precluded the use of many stations; we eliminated all stations that were in or near cities with populations over 160,000.
All stations in Australia have recently undergone a thorough homogeneity analysis (Plummer et al., 1995), but the results were not available for this analysis. Instead, stations were selected on the bases of length of record and distance from major areas of urbanization. All of the Australian stations used are in small towns or rural areas, many in post office ''back yards."
Fewer than half of the 154 stations available from Japan were used in this analysis. As with China, many stations were eliminated because of their proximity to major urban areas. An inspection of the station histories reveals a number of network "improvements" related to the automation of the temperature measurements in recent years. A full assessment of the homogeneity of the data awaits a detailed analysis. The station networks from Sudan and South Africa include some stations from urban areas, but countrywide decreases of the DTR are not overwhelmed by these stations. Incomplete information was available regarding systematic changes in instrumentation at these locations over the past several decades, but the data were inspected for station relocations and adjustments made when necessary on the basis of temperature differences from neighboring stations. In total, four stations in South Africa were adjusted using the procedures outlined by Jones et al. (1986a).
INFLUENCES ON DIURNAL TEMPERATURE RANGE
As more data become available from a variety of countries, it becomes difficult to dismiss the general decrease of the DTR over the past several decades as an artifact of data inhomogeneities. Observing networks are managed differently in each country. If local effects are significantly influencing the DTR, then at least three possible sources of change need to be explored: urbanization, irrigation, and desertification. Evidence to support or refute the impact of these human-induced local and regional effects are discussed in the subsections below.
Urban Heat Islands
It is well known that the urban heat island often tends to manifest itself most strongly during the nighttime hours (Landsberg, 1981). In mid-latitude North American cities the urban-rural temperature difference usually peaks shortly after sunset, then slowly decreases until shortly after sunrise, when it rapidly decreases and, for some cities, actually vanishes by midday. In many cities increases in urbanization would differentially warm the minimum relative to the maximum temperature. A number of precautions have been taken to minimize this effect.
In the contiguous United States, the corrections for urban development recommended by Karl et al. (1988) have been applied to the data, and any residual heat-island effect in this analysis should not be an issue. In Canada, only stations with population less than 10,000 were used in the analysis, and the average population of the cities in the proximity of the observing stations was slightly over 1,000. If the Canadian stations behave similarly to stations in the United States, the effect on the diurnal temperature range may be