Climatologists generally agree that there is accumulating evidence that a warming trend has been occurring since the mid-1960s. There are rather distinct seasonal and regional climate variations in the lower 48 states of the United States. Livezey and Smith determined that the average national warming trend has been 0.0151F per year. Since 1964, this implies that average annual temperature has increased by approximately one-half of one degree.
Nevertheless, there are some dissenting views, such as that of Balling, who argues that temperature changes has not changed in any meaningful way since 1932. His analysis tests for linear trends in temperature and finds no statistically significant trends in the data. Nonlinear trends, however, are possible and the evidence suggests that this indeed is the case. The smooth lines are predicted temperatures from a simple nonlinear statistical trend model in which degree days are specified as a function of a constant, the time period, and the time period squared. The hypothesis that these latter two terms are individually or jointly zero could not be rejected. The probability levels, or the chances that this hypothesis is wrong, are zero for natural cooling degree days and slightly less than 2% for the commercial heating degree day trend terms. Notice that the trend for cooling degree days is downward from 1932 to approximately 1970 and then upward since then. Heating degree days display the inverse pattern, increasing during the early part of the sample but then decreasing during the 1980s and 1990s.
The highly volatile nature of temperature is cause for some concern because statistical models with a relatively small number of observations are prone to be heavily influenced or skewed by extreme observations, such as the severe cold during the winter of 1976–1977. Far more observations are required to place a high degree of confidence in detecting linear or nonlinear trends in temperature. With these caveats in mind, temperature trends during the past 30 years are consistent with the global warming hypothesis.
Morris found that climate trends vary by region and by season. Most regions of the eastern United States had average or slightly below average temperatures. In contrast, the western and Great Plains states experienced warmer temperatures. The western and upper Great Plains regions experienced the most warming during the winter months. Average temperatures during the fall are actually lower for most of the eastern and Midwest regions of the United States. This finding suggests that studies linking energy demand to temperature or degree days should capture some of these regional and seasonal patterns. The analysis presented next explicitly captures the seasonal dimensions. Although a regional climate energy demand model is not developed here, the findings of Enterline suggest that regional aggregation is not a serious problem.
