Urban Renewal Mapping: A Case Study in Beijing from 2000 to 2020
Hao Ni, Le Yu, Peng Gong, Xuecao Li, Jiyao Zhao
Abstract
Understanding the distribution and land history of old urban areas (OUAs) and renewed urban areas (RUAs) has become the key point of urban management. However, it is hard to acquire adequate information for lack of pertinent detection methods. Here, we established a complete mapping framework on Google Earth Engine (GEE) platform to identify OUAs and RUAs and detect the temporal information of urban renewal, which was implemented in Beijing during 2000–2020. We used Landsat imagery and LandTrendr algorithm to fit the spectral trajectories of 14 bands/indices with specific segment attributes as the feature inputs for Random Forest classification. We produced the maps of OUAs and RUAs with an overall accuracy of 95.36%. On this basis, we further utilized LandTrendr to detect the start year, end year, and duration of urban renewal with the accuracies within the ±5-year difference of 85.52%, 80.97%, and 74.53%, respectively. These maps all present informative spatiotemporal patterns. Furthermore, the urban renewal process is likely to be influenced by major national or international events. The study answers the issues about urban renewal from multiple angles and provides scientific support for future urban planning.