Naar klimaatresponsieve planning in Zuid-Afrikaanse steden: het bevorderen van het local climate zones-kader voor de mitigatie van stedelijke hitte-eilanden

As global temperatures continue to rise and urban areas become more developed and densely populated, challenges related to environmental quality and human well-being have intensified. One critical manifestation of this is the Urban Heat Island (UHI) effect, where urban areas experience higher temperatures than surrounding rural regions. This effect has severe implications for human health, including heightened risks of respiratory diseases, heat stress, and, in extreme cases, heat stroke and mortality. With projections indicating that over 60% of the global population will reside in urban areas by 2050, the need to integrate UHI mitigation into urban planning is urgent. However, the scarcity of UHI-focused studies in African cities has limited the development of localised, evidence-based interventions. This dissertation investigates the dynamics of urban heat across the surface and atmospheric profile, focusing on how local land cover and morphology shape microclimates in South African cities. The goal was to assess the thermal performance of different urban forms and land cover types and evaluate their potential to support or undermine UHI mitigation. Given the resource constraints in the South African research and policy context, this study relied on weather station data and satellite imagery to provide high-resolution insight into land-climate interactions at the micro- and local scales. In Chapter 2, the Local Climate Zones (LCZ) framework was adapted for three South African cities: Thohoyandou, East London, and Cape Town. Using machine learning classification on Sentinel-2A imagery, we observed that several urban features did not neatly align with the standardised LCZ typology. Two classification approaches were tested: a direct application of the LCZ framework and a modified approach with custom sub-classes tailored to the local urban morphology. The customised classification outperformed the standard approach, with sub-classes exhibiting distinct spectral signatures, laying the groundwork for LCZ-based thermal analyses. Chapter 3 evaluated whether the LCZ classes could meaningfully differentiate surface thermal profiles. Using a 1°C temperature difference threshold, we found that classes with similar built densities or vegetation proportions often fell below this threshold. In the absence of building height data, LCZs that differ only in vertical structure were also thermally indistinct. We applied two clustering methods: one based on physical similarity and another on thermal characteristics, both of which yielded nearly identical groupings. These reclassified zones exhibited thermal differences above the 1°C threshold, offering a refined typology for analysing urban thermal behaviour. In Chapter 4, we deployed 13 weather stations in Cape Town and 15 in Thohoyandou to collect meteorological data, focusing on air temperature and wind speed. Using Random Forest models, we simulated UHI mitigation scenarios, such as increasing vegetation or replacing impervious surfaces. In Cape Town, converting paved surfaces to high vegetation yielded the greatest cooling benefits, while in Thohoyandou, replacing the same with low vegetation was most effective. Notably, the least temperature increase in Thohoyandou occurred when built areas were simulated over bare land, whereas in Cape Town, the least warming resulted from development over low vegetation. These differences underscore the need for context-specific UHI strategies. In Chapter 5, we added citizen-owned weather stations calibrated and filtered using WMO regulations to collect air temperature, relative humidity, and wind speed to calculate Thom's thermal discomfort index. We investigated the spatial extent of thermal discomfort regulation effects around natural LCZs under low and high wind speed regimes. Using eXtreme Gradient Boosting models, we assessed how far the influence of forest (LCZ 11), shrubland (LCZ 13), and low vegetation (LCZ 14) extends into adjacent urban areas. Under low wind conditions, LCZ 11 had the longest influence (up to 300 m), which contracted to 100 m under high wind speeds. In contrast, LCZ 13's influence remained limited to the first 100 m regardless of wind regime. LCZ 14 was the most ventilated, showing the capacity to extend its influence up to 900 m in the presence of wind, although wind effects dominated beyond the first 100 m. In summary, this dissertation improves our understanding of how land cover and morphology influence urban microclimates across different spatial scales and seasonal contexts. It provides empirical guidance on which vegetation types to prioritize in different cities and highlights the limits of generic, one-size-fits-all approaches to UHI mitigation. The findings support a locally grounded, evidence-based foundation for urban planning that balances environmental performance with social needs, especially in resource-limited contexts like South Africa.