6.2
Rational analysis of trends in extreme temperature and precipitation
Patrick J. Michaels, Univ. of Virginia and Cato Institute, Charlottesville, VA; and P. C. Knappenberger, R. E. Davis, and O. W. Frauenfeld
Previous studies have often assumed that climate changes accompanying an enhanced greenhouse effect would simply shift mean values while not changing relative frequency distributions. The resultant simplistic and naïve view of climate change has implications for environmental policy that could cost the U.S. several percent of its GDP. Our research shows that this view is scientifically unwarranted.
We analyzed daily maximum and minimum temperatures from the United States Historical Climate Network (HCN/D), sorting the days of the year from a high-quality subset of stations from coldest to warmest. We then examined the temporal behavior of each of these 365 records. In other words, we examined the behavior of the warmest day of each year, the second-warmest, etc…down to the coldest.
The U.S. annual temperature history of the 20th century is characterized by three distinct periods: a period of warming in the first 40 years followed by cooling through 1969 and warming again through the end of the century. In the early 20th century warming period there was a tendency for most of the increase in maximum temperatures to occur on the warmest days of the year. The mid-century period of cooling was marked by the greatest temperature decline being found on the already coldest days of the year. In these two cases, the extremes were becoming more severe. In contrast, during the period of late century warming, the largest warming trend occurred on the coldest days of the year, which is a climatic tendency toward less thermal extremes.
A comparable analysis of precipitation is confounded by its discrete nature. Therefore, we examined the distribution characteristics of daily precipitation rather than the individual events. We best-fit a Weibull distribution to the seasonal precipitation events in overlapping three-year periods, and then analyzed the historic behavior of the shape and scale parameters. There is a slight tendency for the amount of precipitation from events at both the low end as well as the high end of the distributions to be increasing, relative to the events near the center of the distribution. Back-calculating through these parameters reveals that, in general, the 1 to 1.5 inch (2.5-3.8cm)/24 hour rain events shows the largest increases.
We therefore conclude with confidence that the era of greenhouse enhancement is concurrent with a reduction in relative thermal extremes of in the United States, and an increase in a precipitation class that is a clear benefit to our society. Policies directed at reducing greenhouse emissions are likely to mute or reduce these salutary trends.
Session 6, Climate Trend Analyses
Tuesday, 14 May 2002, 10:30 AM-12:00 PM
Previous paper Next paper