Exploring the drivers and dynamics of urban waters: A case study of Wuhan from 1980 to 2060

• Novel methodological approach combining LEAS, CA, and Markov Chains for predicting urban water changes. • Quantitative analysis of the mechanisms driving water surface dynamics through RF data mining. • Patch-generating analysis detecting waters’ area fluctuating increasing in Wuhan from 1980 to 2020. Urban waters’ changes are critical indicators of environmental and urbanization dynamics. This study systematically explored the changes of waters in Wuhan from 1980 to 2020 and predicted the trends of waters in 2040 and 2060 using eleven divers including digital elevation models (DEM), precipitation, and land use/cover. Methods such as transition matrix, Fragistats, the Land Expansion Analysis Strategy (LEAS), CA models, and Markov Chain were employed. The results revealed fluctuating increases in waters’ area over the past four decades, from approximately 126,549 ha in 1980 to 164,978 ha in 2020, influenced by lake evolution and exhibiting transitions of 1000–3000 ha between waters and cropland or settlements. Further analysis indicated that DEM and precipitation were primary driving factors, contributing 41 % to the total changes, with buildings (contributing 8.8 %) being one of the most significant socio-economic influences. The Cellular Automata based on multi-type random patch seeds (CARS) modeling predicted that by 2060, the spatial structure of waters in Wuhan would remain similar to that of 2020, albeit with changes around lakes and ponds. These findings provide essential insights into the mechanisms and future trends of waters’ changes, with significant implications for urban planning and water resource management.