Monitoring floodplain wetland dynamics along uMfolozi and uMsunduzi confluences using remotely sensed data

Abstract

This study investigates the spatiotemporal dynamics of floodplain and vulnerability of wetlands to land use and land cover change (LULC) in KwaZulu-Natal, South Africa, focusing on the uMngeni Catchment and uMsunduzi floodplains. The investigation was structured around two primary objectives: (1) to retrospectively analyse the spatiotemporal dynamics by mapping and quantifying changes in wetland extent and LULC from 2000 to 2024, establishing a baseline of historical trends, and (2) to prospectively model and predict future LULC changes up to 2060 Multi-temporal Landsat 7, 8, and 9 datasets were processed on the Google Earth Engine (GEE), using the Random Forest (RF) algorithm. LULC mapping was achieved with overall accuracies ranging from 84.21% to 91.23%, with kappa values between 0.78 and 0.84, confirming robust results. To model future change, the Modules for Land Use Change Evaluation (MOLUSCE) was applied, integrating Cellular Automata (CA) and Markov Chain analysis with environmental drivers such as elevation, slope, distance to rivers, and proximity to roads. The model was calibrated with historical maps and validated using a simulated 2020 map, which achieved a predictive kappa of 0.79. The results showed substantial wetland loss in both systems. Within the uMngeni Catchment, wetlands declined from 17,300 ha in 2000 to 11,500 ha in 2024, representing a 34% decrease, while wetlands in the uMsunduzi floodplains declined from 2,977.9 ha to 1,873.9 ha, representing a 37% loss over the same period. Built up areas increased by 38% during the same period, while agriculture peaked in 2010 before declining slightly by 2024.With respect to the second objective, MOLUSCE projections indicate that wetlands may decline by a further 12.4% by 2060, while urban areas are expected to increase by 41%. Agricultural land shows a fluctuating trajectory, declining gradually after its 2010 peak. Forest and grassland areas remained relatively stable in both past and projected. Analysis confirms the acute vulnerability of wetland ecosystems to anthropogenic pressures, posing a significant threat to the ecosystem services they sustain. These results highlight an imperative for pre-emptive action, championing interventions like wetland restoration, the strict enforcement of riparian buffer zones, and the adoption of sustainable land-use strategies. Ultimately, by integrating historical trends with future projections of land-use change, this study generates spatially explicit evidence base designed to guide strategic environmental policy and land-use planning.,