Drivers of drought in Cape Town: Explaining past and future regional precipitation changes through large-scale dynamics

Climate models consistently project a decrease in winter rainfall in South Africa’s Southwestern Cape (SWC) in the coming decades. However, the model spread in future precipitation projections remains large, and climate models exhibit substantial biases in simulating regional rainfall. Improving projections requires a better understanding of the physical processes governing precipitation variability and change in the SWC, as well as their representation in climate models. Here, we link large-scale zonal and meridional wind patterns to SWC precipitation and derive dynamical rainfall drivers tailored to the region of interest. Using a multilinear regression framework, we dynamically reconstruct precipitation variability on interannual to multidecadal timescales and show that zonal and meridional wind in specific regions explain approximately 40% of observed historical rainfall variability. We further show that a substantial fraction of the projected precipitation decline in CMIP6 models (around 75%), as well as a large portion of the inter-model spread in future projections, can be attributed to changes in the identified dynamical drivers. Moreover, we show that models with more realistic links between circulation and regional precipitation also exhibit smaller rainfall biases. Finally, we assess the potential and limitations of applying observational constraints to SWC precipitation change and derive observation-informed precipitation projections. These projections indicate end-of-century precipitation declines exceeding the CMIP6 multi-model median, suggesting that future winter rainfall declines in the SWC may be underestimated by unconstrained model projections.