Extraction of Urban Built-Up Areas and Urbanization Analysis in Changsha Using Multi-Source Remote Sensing Image Fusion

With the advancement of China’s new urbanization strategy, dynamic monitoring of urban built-up areas has become crucial for optimizing territorial spatial patterns and promoting sustainable development. To address the limitations of single-source remote sensing data in spectral integrity, spatial resolution, and human activity characterization, this study proposes a multi-source remote sensing fusion framework for urban built-up area extraction and spatiotemporal analysis, using Changsha as a case study. Utilizing Landsat multispectral/panchromatic and NPP-VIIRS nighttime light data (2013-2022), we developed a multi-resolution fusion model after radiometric calibration, atmospheric correction, and temporal normalization of nighttime light data: First, a three-level Haar wavelet decomposition with parametric layer-wise fusion generated 15m-resolution Landsat fused imagery. This was then combined with resampled nighttime light data to create novel color nighttime light (CNTL) images preserving both land cover features and human activity intensity. Comparative analysis of NPP-VIIRS change detection, Landsat SVM classification, and our improved SVM model integrating spectral-nighttime light features confirmed the superiority of the latter (overall accuracy >92%, Kappa ≥0.86). Multidimensional metrics (expansion index, centroid shift, and standard deviation ellipse) revealed a “core-clustering/peripheral-diffusion” spatiotemporal pattern in Changsha: Dominant central urban expansion during 2013-2016 transitioned to suburban sprawl post-2016, with emerging polycentric development. The study demonstrates that multi-source remote sensing fusion enhances urban monitoring accuracy, providing data and methodological references for spatial optimization and sustainable urban planning in central China’s city clusters.