In a subsequent story, William Stevens reported that 1995 had been the "hottest year on record." But Stevens was apparently relying on faulty data that only covered the year through November. Once December's data were included, it turned out to be an average year after all—0.05 degrees above average in North America, cooler than 1991, 1990, 1988, 1987, 1983, 1981, and 1980. 1996 was also 0.09 degrees below average worldwide.
Rising temperatures due to the greenhouse effect, a phenomenon that remains undetected by NASA satellites, has also been blamed for each year's remarkable weather patterns, including hurricanes, tornados, and floods—a trend that itself forms a pattern.
An article in Nature magazine attempted to discredit the satellite data from NASA and the University of Alabama at Huntsville, which showed a slight cooling trend over the 18 years they had been taken, by arguing that the satellites last only ten years and the scientists had miscalibrated the newer satellites. (Temperature readings from satellites are considered far more accurate than surface readings, which are prone to all manner of variability. These variations include the "urban heat island effect," in which weather stations once set in rural areas become enveloped by the hot pavement of growing suburbs.) The authors of the article based their claim of the satellites' inaccuracy on estimates of average global temperatures derived from general-circulation climate models. But unfortunately these computer models rely on relatively crude mathematical approximations of the climate's complex physical behavior, often overlooking relevant variables—such as the heat sequestered in oceans and radiated by clouds. In fact, the models have yet to accurately reproduce known climate trends for the past 100 years from first physical principles.
University of Virginia climatologist Patrick Michaels notes that the arguments used to explain discrepancies between observed data and predictions based on theory have often been novel. Chris Folland of the United Kingdom Meteorological Office, for example, commented to a stunned gathering of climatologists that "the [observed] data don't matter," meaning that the predicted temperature increase was likely offset by a random but relatively long-lasting temperature decrease due to the inherently complex and nonlinear nature of the Earth's climate—a natural variation. Such a random fluctuation, of course, may just as easily have gone the other way and increased temperatures to further magnify existing increases from global warming. And assuming that the temperature was increasing at all, there would be no way of knowing whether it was due to a random fluctuation or increases in greenhouse gases. In such a view, chaos theory trumps the supposition of causation, rendering observed data meaningless. It may be added that this chaotic fluctuation of the data, though certainly possible in the real world, was also the product of computer modeling.