Spatial Mapping and Modeling Flood Risk Susceptibility in Global Urban Floating Slums Using Remote Sensing and Machine Learning

Earth is urbanizing fast, and overcrowded cities are giving rise to floating slums in high-risk flood zones, an urgent yet overlooked issue. This study aims to (1) Map the spatial expansion of floating slums near submerged or flood-adjacent built-up zones, (2) Assess environmental degradation of surrounding vegetation due to slum encroachment and recurrent flooding, (3) Model flood susceptibility using hydrological and land-use/land-cover variables to identify vulnerable slum areas, and (4) Quantify built-up footprints within flood-prone zones and estimate population exposure across risk categories in informal settlements globally. This study introduces a novel integration of high-resolution multispectral satellite imagery, terrain metrics, and socio-economic indicators within a dual-stage Random Forest classification framework, enabling precise mapping of urban floating slums and flood susceptibility across five global sites with over 90% model accuracy. The results show that (1) Built-up areas expanded significantly, with Makoko increasing from 73.84% in 2019 to 82.59% in 2025; Cái Răng from 34.54–41.63%; and Manacapuru from 2.14–3.04%. (2) Vegetation cover declined, particularly in Kampong Ayer (from 65.45–60.98%), Cái Răng (49.29–42.99%), and Manacapuru (49.94–47.77%). (3) Flood susceptibility mapping revealed diverse exposure levels, with a substantial portion of each site falling into high and very high-risk zones. (4) Population exposure in very high-risk zones is severe, including 63,450 in Makoko, 388,450 in Kampong Ayer, 99,600 in Tonle Sap, 2,406,800 in Manacapuru, and 3,282,000 in Cái Răng. Urban resilience is urgently needed to limit urban floating slums related issues in line with SDG 11’s goal for inclusive and safe cities.