Impact of rainfall variability on streamflow: Case of Crocodile River, Mpumalanga

Climate change threatens water security in semi-arid transboundary river basins facing intense agricultural demands and strict international treaty obligations. The Crocodile River Catchment in Mpumalanga, South Africa, represents a critically stressed system where river discharge is regulated by the Kwena Dam to secure downstream irrigation and meet statutory flow requirements to Mozambique. While progressive warming and recurrent droughts have increased hydroclimatic volatility across the region, localized dam operations and landscape buffering cushion river channels, making it unclear how directly atmospheric drying trends translate into changes in downstream river discharge. This study conducts a non-parametric trend analysis of long-term hydroclimatic variables within the catchment. Specifically, it aims to detect and compare monotonic trends in historical precipitation (P) and river discharge (Q) records spanning 1970 to 2025. Chronological records are evaluated using the rank-based, non-parametric Mann–Kendall (MK) statistical test, which is robust against outliers and non-normally distributed time series. The standardized test statistic (Zs) is calculated to determine the significance and direction of monotonic changes at a 95% confidence level (|Zcrit| = 1.96). The MK test reveals a major hydroclimatic decoupling within the catchment. Precipitation exhibits a statistically significant decreasing trend (S = -384.00), (Zs = -2.71), verifying severe long-term atmospheric drying. Conversely, river discharge shows no statistically significant trend (S = -96.00), (Zs = -0.67), failing to reject the null hypothesis. This structural mismatch based on the trend analysis is further influenced by Kwena Dam's (situated upstrem of the catchment) operating rules and international obligations, which artificially sustain and stabilize downstream river channels despite collapsing rainfall inputs. The catchment demonstrates a modified hydrological regime where human intervention sometimes masks natural climatic forcing and these trend-detection insights provide a verified empirical baseline to optimize regional water allocation policies and manage transboundary drought risks.