Dependence of long-term annual rainfall trends in South Africa on analysis period: 1921-2022

Abstract

The focus of the paper is to address the temporal discrepancies in long-term rainfall trend results in South Africa, by evaluating a systematic range of observation periods over the century-long analysis period of 1921–2022. Available long-term climate projections show that significant parts of South Africa are expected to experience progressively drier conditions, mainly in the west. To assess long-term rainfall trends over the country, historical trends should be determined – the results of published studies for which are not consistent. Arguably the most significant reason for these inconsistencies is the length of analysis period. This study investigates this effect on the magnitude and statistical significance of historical annual rainfall trends over hydrological years, with the data of 94 homogeneous rainfall districts. Trends of annual rainfall were determined for all periods up to 2022, from the 1921–2022 period up to the last 30 years, i.e. 1993–2022. The annual rainfall trends over the longest analysis period show significantly positive trends over most of the central and western parts and significant drying over extensive parts of mainly the far north-east. However, the most recent period (1993–2022) shows significantly drying trends over extensive parts of the west, south and east. Among the main findings is that the change in trends shows a consistent spatial pattern of significant negative change over most of the south- western half, and mostly significantly positive trends over the north-eastern sector. These results are in large agreement with future rainfall projections, particularly for the western half. Ultimately, the study emphasizes the wide range of long-term trends in South Africa, both spatially and temporally, and the importance of considering a range of historical analysis periods in the detection of long-term rainfall changes. In effect, the eventual results provide increased confidence to predicted future rainfall scenarios.