Identifying where and when urban renewal occurs: a continuous change detection-based framework using two decades’ worth of Landsat data

Urban renewal plays a central role in enhancing city liveability by rebuilding outdated structures into productive and vibrant spaces. While satellite remote sensing enables physical characterization of urban environments, identifying the precise location and timing of renewal remains challenging. Here we developed a 30 m city-scale urban renewal mapping framework with the use of dense Landsat time-series information. By leveraging the Continuous Change Detection and Classification (CCDC) algorithm, we designed a decision tree model to identify pixels experiencing urban renewal and utilized the temporal contextual knowledge to estimate temporal metrics, including start time (ST), end time (ET), and duration (DUR). Experimental results in Beijing city confirmed the feasibility of the framework, achieving spatial and temporal accuracies of 82.36% and 71.39~86.60%, respectively. Our mapping results revealed that the total area of urban renewal within the study area reached 340[Formula: see text]55 km2 from 1999 to 2019, distributed unevenly along the urban-rural gradient. We also identified the dominance of quick demolition and reconstruction implementation accomplished within five years. The framework provides a new paradigm for continuously monitoring city development from the perspective of urban renewal, thus supporting the improvement of urban land planning and management.