Projected changes in daily temperature extremes for selected locations over South Africa

Abstract

Extreme events, particularly very high temperatures, are expected to increase because of climate change. It is thus essential that localised studies be done to quantify the magnitude of potential changes so that proper planning, especially effective adaptation measures, can be affected. This study analysed annual extreme daily maximum temperatures for future climate change scenarios at 22 locations in South Africa, through analysis of a subset of the Coordinated Regional Downscaling Experiment (CORDEX) model ensemble datasets. The multi-model simulations were validated against observational data obtained from the South African Weather Service for the period 1976–2005. Two study periods of mid- (2036–2065) and far-future (2066–2095) were analysed for two Representative Concentration Pathways, i.e., RCP4.5 and RCP8.5. Bias correction was done on the model data to correct simulated historical climate data, to be more characteristic of observed measurements. While the method included adjustment for variance, systematic underestimations of extremes were still evident. The Generalized Extreme Value distributions were fitted to the bias-corrected projections, and 10-, 50- to 100-year return periods quantile values were estimated. The return period quantile values are likely to increase under both Representative Concentration Pathways in the mid- and far-future periods, with the largest increase in return period quantile values set to occur towards the end of the century under the highest emission scenario. All stations showed an increase in the frequency of days with maximum temperatures above specific critical thresholds, with some stations under the RCP8.5 scenario projected to experience temperatures of greater than 32°C (35°C) for more than 200 (100) days per year by the end of the century, an increase from a baseline of approximately 70 to 150 (14 to 83). For the same scenario, Return periods for 38°C for most stations are projected to be shorter than a year. From the above and considering the likely underestimation in the severity of the projected changes, i.e. too low return period quantile values, the general implication is a strong likelihood that most places in South Africa is likely to experience a strong increase in the intensity, duration, and frequency of very hot extremes in future, with potentially dire consequences to relevant socio-economic sectors. We suggest that future research, comprised of the full set of CORDEX data be conducted to optimise the results of this study.