Long-term monitoring of coastal aquaculture ponds (APs) is essential for sustainable food production and environmental management. Intensive aquaculture has reshaped coastal landscapes and threatened ecosystem services. Here, we developed a Spectral-Temporal Filtering–Random Forest (STF-RF) framework to map APs along China’s coast from 1985 to 2024 using Landsat time-series imagery on Google Earth Engine. Ten representative bays and shoals were analyzed to capture regional heterogeneity. The STF-RF method achieved >90% classification accuracy, effectively distinguishing APs from spectrally similar land covers. Temporal trends reflected policy-driven dynamics: rapid expansion during the 1980s–2000s was followed by localized contraction and stabilization under ecological management. Spatiotemporal evolution, quantified via Standard Deviational Ellipse (SDE) analysis, showed initial landward expansion into sheltered areas followed by seaward extension, with northern regions exhibiting greater spatial dispersion and southern regions forming clustered distributions. This four-decade dataset provides a unique perspective on aquaculture’s ecological footprint and informs sustainable planning, ecological risk mitigation, and cleaner production. The STF-RF + SDE framework is transferable to other developing coasts, offering broad applicability for supporting global Sustainable Development Goals.
Wang et al. (Sun,) studied this question.