by Tim Storer
Given the constantly fluctuating nature of weather both seasonally and annually, it is often difficult for scientists to show or describe long-term climate changes in a succinct manner. Even if one accepts the hypothesis that humans can have no major effects on global climate change, it is still useful to study climate patterns for predictive purposes. Certain studies have shown how various climate extremes such as rainfall and temperature are expected to increase in future decades, and others have shown overall warming of the planet. In South Africa, there have been strong trends showing heightened extremes of both the lowest and highest temperatures in all stations studied, though the degree of this amplification varied by location (Kruger and Sekele 2013). This has implications for both the wild ecosystems in South Africa and the human populations. Uncharacteristically high or low temperatures can easily catch humans (and other species) off guard, and it is useful to help predict and prepare for these conditions.
Kruger, A. C. and Sekele, S. S., 2013. Trends in extreme temperature indices in South Africa: 1962–2009. Int. J. Climatol. 33, 661–676. http://goo.gl/VFQ82G
A.C. Kruger and S.S. Sekele did a comprehensive data collection of temperatures in South Africa between the period of 1962–2009, compiling their data from 28 weather stations around the country. From each station, the highest and lowest temperatures from each day in that time period were recorded. The RClimDex software they used calculates peak and minimum temperatures per year, the temperatures of the 10th/90th percentiles per year, and the number of consecutive days in these extreme high percentiles. By measuring changes in the percentage of days in the 10th/90th percentiles as years go by, the researchers could see trends in weather intensity. The researchers also used cluster analysis to identify areas of similar climatological characteristics. Six different clusters, labeled A–F, were identified based on similar trends and mean temperature highs and lows per year.
The data analysis yielded several extraordinary results. First, 22 of the 28 stations showed statistically significant increases in the percentage of days above the 90th percentile of peak daily temperature, and none of the stations reported decreases. These trends were particularly high in the southern coastline, with a mean increase of 2.97% per decade. This indicates an increase in the number of extremely hot days at most of the stations throughout the decades studied. Similarly, the percentage of days with unusually cold temperature highs showed significant decreases in 18 of the 28 stations. This shows the decreasing frequency of moderate weather. Strong trends in 90th percentile peak and 10th percentile peak are shown in the same regions, though with opposite sign in trends, as expected.
Comparable results were found for the percentage of extremely cold days, with 13 of the 28 stations reporting statistically significant increases in the percentage of days with temperature in the lowest 10% (most of the other areas showed no significance, and one showed a decrease). Also, the percentage of days with unusually warm minimum temperatures showed significant decreases at 18 of the 28 stations, and no significant increases. The most sizeable amplification of weather extremes, both hot and cold, were seen in the southwestern and northeastern parts of the country.
Almost no significant trends were found in the absolute minimum and maximum yearly temperatures from the stations. Similarly, the attempts to show trends in extreme weather streaks, both hot and cold, fell short due to statistically insignificant findings. Researchers chose to study this by counting the number of 90th percentile days within streaks of 90th percentile days at least 6 days long. They speculate that by having such a long duration, any trends weren’t correctly observed.