Temporal Sensitivity of Observed Surface Air Temperature Trends Across South Africa over the Last 30 Years

Observed temperature trends provide a basis for understanding climate change and planning for adaptation to its impacts. However, estimated trends can vary substantially according to the tem-poral and spatial characteristics of the observations analysed. Here, an assessment is made of the spatial and temporal variation of observed surface temperature trends across South Africa using daily observations from a fixed network of 123 stations for 1996–2025. The analysis involves mean temperature, mean daily maximum and minimum temperatures, and warm and cold extremes, i.e., annual maximum and minimum temperatures. The methods comprise trend estimations of progres-sively shorter and moving 10-year periods, and a quadratic assessment of trend curvature. The re-sults indicate widespread warming overall, but with substantial spatial heterogeneity, including localised areas where temperature trends are negative, suggesting cooling over the analysis period. For mean temperature, the median long-term trend is 0.28 °C/decade across the stations, with 55.3% showing statistically significant warming. However, for daytime maximum temperature the median trend is 0.51 °C/decade with 72.4% of stations showing statistically significant positive trends. Sub-stantial spatial and temporal sensitivity of estimated trends to the analysis period were found, with short-term trends varying markedly among periods and regions. The median of the average tem-perature trend was +0.44 °C/decade for 2011–2025 but declined to essentially zero for 2016–2025. The strongest long-term warming occurs predominantly in the northern interior of South Africa, but the spatial distribution of trend curvature differs from that of the overall warming trend. Evidence of significant positive trend curvature was found for a relatively small number of stations. The findings demonstrate that a single trend estimate can provide an incomplete representation of observed cli-mate change and that trend magnitude, temporal variability and changes in rate should be consid-ered separately. The results provide an observational basis for more robust climate-risk assessment and planning in various climate-sensitive sectors.