Multisource urbanization monitoring

Published · Long-term urban land-cover and expansion mapping from optical and SAR time series with machine learning.

Published

Research question. How can optical and synthetic-aperture-radar observations be combined to map urban land cover and expansion reliably — through clouds, shadows, and seasons — over decades?

Scale. Large-scale and multi-city analyses, from intra-year urban renewal to multi-decadal expansion.

Data. Optical (Landsat, Sentinel-2, WorldView), SAR (Sentinel-1, TerraSAR-X, PALSAR), and LiDAR observations.

Methods. Deep learning and machine learning for data fusion — including sparse representation, stacked autoencoders, U-Net variants, and fuzzy-set approaches — applied to impervious-surface mapping, road extraction, and urban change detection.

Key contributions. This line of work established multisource fusion approaches that map annual impervious-surface dynamics at large scale (Lin et al., 2020), extract road networks from combined optical–SAR data (Lin et al., 2021), improve urban mapping under cloud and shadow contamination (Lin et al., 2019), and detect intra-year urban renewal in metropolitan cities (Lin & Zhang, 2022). The resulting long-term urbanization records support analyses of environmental consequences, from heat exposure to coastal ecosystem change.

Stylized urban expansion motif with city blocks and radar scan lines

Illustrative motif — replace with a figure from the study (assets/img/projects/).

References

2022

  1. Intra-Year Urban Renewal in Metropolitan Cities Using Time Series SAR and Optical Data
    Yinyi Lin and Hongsheng Zhang
    In IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium, Jul 2022

2021

  1. JAG
    Leveraging optical and SAR data with a UU-Net for large-scale road extraction
    Yinyi Lin, Luoma Wan, Hongsheng Zhang, Shan Wei, Peifeng Ma, and 2 more authors
    International Journal of Applied Earth Observation and Geoinformation, Dec 2021

2020

  1. RSE
    Incorporating synthetic aperture radar and optical images to investigate the annual dynamics of anthropogenic impervious surface at large scale
    Yinyi Lin, Hongsheng Zhang, Hui Lin, Paolo Ettore Gamba, and Xiaoping Liu
    Remote Sensing of Environment, Jun 2020

2019

  1. Improving Impervious Surface Extraction With Shadow-Based Sparse Representation From Optical, SAR, and LiDAR Data
    Yinyi Lin, Hongsheng Zhang, Gang Li, Ting Wang, Luoma Wan, and 1 more author
    IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Jul 2019