Traditionally, vulnerability assessments for climate change risks rely on randomized precipitation possibilities and lack a physical science basis. Additionally, limitations in representing local precipitation fields in General Circulation Models (GCM) hinder the usefulness of precipitation projections of future hydroclimatic extremes. To address these challenges in climate risk management, this study investigates the ability of large-scale atmospheric circulation patterns, or weather regimes (WRs), to explain local precipitation and regional precipitation drivers. Utilizing a Non-Homogeneous Hidden Markov Chain approach, we identified six primary WRs for South Africa, each exhibiting distinct seasonality. Three WRs, associated with higher precipitation near Cape Town, dominate in winter, while two WRs linked to lower precipitation are prevalent in summer. The WR-precipitation relationship in South Africa appears to be influenced by topographic features (e.g., The Great Escarpment and Cape Fold Mountains) and ocean currents (Agulhas and Benguela), leading to distinct spatial precipitation responses to regional WR configurations. The seasonal frequency of WRs in the South African region has shifted dramatically in the past 20 years. Changes in WR composition, particularly the replacement since 2010 of a WR associated with rain in Cape Town with a dry WR, may help to explain the worsening drought conditions in the past decade. During the 2015-2017 Day Zero drought in Cape Town, the regional WR associated with dryness in Cape Town occurred more frequently than the historical average. The insights from the WR-precipitation analysis can be used to inform WR-based stochastic weather generators for vulnerability assessment and climate change adaptation planning.

Tritium (³H) has become synonymous with modern groundwater and is used in a myriad of applications, ranging from sustainability investigations to contaminant transport and groundwater vulnerability. This study uses measured ³H groundwater activities from 722 samples locations across South Africa to construct a ³H groundwater distribution surface. Environmental co-variables are tested using geostatistical analysis to constrain external controls on ³H variability, namely: [1] depth to the water table, [2] distance from the ocean and [3] summer vs winter rainfall proportion. The inclusion of co-variables in the ‘fit’ of residual variograms improved prediction variance significantly, yet does not mitigate issues with sample density. The distribution of ³H in groundwater surface agrees well to expected controls, with proximal (<100km) coastal regions, winter rainfall zones and deeper groundwater tables predicted to have lower ³H activities. Conversely, inland localities with shallower water tables and/or summer rainfall are predicted to have elevated ³H activities. High groundwater ³H anomalies could potentially be attributed to uranium-bearing deposits, as geogenic production of ³H amplifies the activity contributed through recharge. Some ³H high and low anomalies cannot be explained by known phenomena and may simply be regions of variable recharge and/or longer isolated groundwater flow paths. Regions of active recharge are more vulnerable to climate change as well as modern pollution. Less actively recharged groundwater may be more resilient to climate change, yet represents a potentially non-renewable resource for abstraction. The application of ³H distributions in the assessment of hydrological resilience is pertinent to effective groundwater management studies.