Enhancing wetland vulnerability assessment through big data and innovative cloud computing techniques

Abstract

Wetlands are among the most productive ecosystems globally, providing critical ecological services, yet they are increasingly threatened by land use and land cover change, particularly in rapidly urbanising semi-arid regions. This study assessed the spatial and temporal vulnerability of wetlands within the Cape Town Metropolitan Area using an integrated remote sensing, machine learning, and multi-criteria decision analysis framework implemented in a cloud-computing environment. Multi-decadal Landsat (1993–2023) and Sentinel-2 imagery, together with climate data, were processed in the Google Earth Engine platform to delineate wetland extent and land use dynamics using Random Forest classification. Biophysical indicators of vegetation condition, surface water availability, hydrological setting, and climatic variability were combined with anthropogenic pressure indicators related to urbanisation, infrastructure, and agricultural proximity to derive a composite wetland vulnerability index.