Manyaka, Naledi2026-06-152026-06-152025https://hdl.handle.net/10566/24455Flooding is one of the most devastating natural disasters globally, causing extensive damage to infrastructure, agriculture, and human settlements. In many regions, recurrent flood events continue to hamper disaster management programmes, largely due to the limited availability of near real-time monitoring systems and comprehensive risk assessment frameworks. This study aimed at addressing these challenges by developing an integrated spatial explicit flood risk assessment and monitoring framework using Google Earth Engine (GEE) cloud computing platform and multi-source spatial data (Sentinel-1 SAR, Sentinel-2 and CHIRPS). The study aimed to assess flood dynamics in the study area through two main components: a detailed assessment of flood events and an inter-annual analysis covering the period 2020-2024. Firstly, a flood event assessment was conducted (February 2023 flood) using threshold-based change detection methods applied to Sentinel-1 Synthetic Aperture Radar (SAR) data to map the spatial extent of inundation. The analysis revealed that the Ehlanzeni District experienced the highest level of flooding, covering approximately 17000 hectares, followed by the Gert Sibande District, where inundation ranged between 4300 and 8500 hectares. The inter-annual analysis further looked at temporal patterns and variability in flood occurrence over many years in the entire Mpumalanga province, providing insights into flood-prone areas. Overall, the results showed that Nkangala district experienced lower flood susceptibility. The inter-annual analysis showed substantial variability in flood extent, ranging from 30800 ha (2024) to 488100 ha (2021). The findings of the study further showed that the year 2021 and 2023 experienced severe flooding due to Tropical Cyclones Eloise and Freddy, respectively. Cropland impacts peaked at 39363 ha in 2023, while built-up area flooding reached 80711 ha in 2021. Rainfall threshold analysis identified optimal 7-day accumulation windows for flood prediction, with regional thresholds varying from 45 mm (Lowveld) to 65 mm (Highveld), providing empirical triggers for early warning systems. The integration of SAR data with optical imagery and weather data within the GEE platform proved effective for operational flood monitoring in data-scarce regions. The methodology provides a scalable, cost-effective framework that can be applied to similar contexts across Southern Africa.enCloud computingdisaster risk reductionFlood risk mappingmachine learningmulti-source dataThesis