Scholarlink

<p>Although the filling of lakes for urban construction has provided short-term land security for some rapidly expanding cities, the long-term geological hazards resulting from such activities, especially continuous ground subsidence, pose a serious threat to infrastructure safety and sustainable urban development. Therefore, accurate monitoring and prediction of land subsidence is essential for the effective management and disaster mitigation in urban lake-infilled areas. This study investigates land subsidence in Shahu Lake, Wuhan, a typical urban lake reclamation area. Lake boundaries and reclamation extents were extracted using Modified Normalized Difference Water Index (MNDWI) from Landsat imagery, and deformation was quantified by Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) time-series analysis using Sentinel-1&#xa0;A imagery. Deformation was decomposed into trend, periodic, and stochastic components and predicted using polynomial fitting, a Convolutional Long Short-term Memory (ConvLSTM) network, and residual analysis, respectively. From 1990 to 2015, Shahu Lake contracted from 9.56&#xa0;km² to 2.48&#xa0;km²; while reclamation for urban expansion was concentrated in the northeastern sector, the remaining margins experienced gradual encroachment. Early reclamation was dominated by construction land use, whereas the later stage shifted toward ecological development. Subsidence was widespread across reclaimed areas but showed spatial heterogeneity, controlled by soil consolidation and static–dynamic loading. Significant subsidence occurred in newly reclaimed and high-load development zones, with deformation rates from − 26.49 to 2.53&#xa0;mm/a and a maximum cumulative subsidence of 264.11&#xa0;mm. The proposed hybrid prediction framework captures complex deformation and indicates a deceleration trend toward 2025, while hotspots persist in high-load areas driven by consolidation (reaching annual rates up to 6.920&#xa0;mm) and near-lake margins exhibit pronounced seasonal instability. This study develops a subsidence prediction framework that achieves high-precision forecasting through deformation decomposition and model integration, providing a basis for urban sustainability and infrastructure safety.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Land Subsidence Simulation and Prediction in Typical Urban Lake-infilled Area Using Polynomial and ConvLSTM Models: A Case Study of Shahu Lake, Wuhan, China

  • 期刊论文

  • 发表日期：

    2026年4月15日

查看全文
  • Yang Zhang,
  • Kai Lan,
  • Siyu Liu,
  • ChenYang Huang,
  • Hui Deng

摘要

Although the filling of lakes for urban construction has provided short-term land security for some rapidly expanding cities, the long-term geological hazards resulting from such activities, especially continuous ground subsidence, pose a serious threat to infrastructure safety and sustainable urban development. Therefore, accurate monitoring and prediction of land subsidence is essential for the effective management and disaster mitigation in urban lake-infilled areas. This study investigates land subsidence in Shahu Lake, Wuhan, a typical urban lake reclamation area. Lake boundaries and reclamation extents were extracted using Modified Normalized Difference Water Index (MNDWI) from Landsat imagery, and deformation was quantified by Small Baseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) time-series analysis using Sentinel-1 A imagery. Deformation was decomposed into trend, periodic, and stochastic components and predicted using polynomial fitting, a Convolutional Long Short-term Memory (ConvLSTM) network, and residual analysis, respectively. From 1990 to 2015, Shahu Lake contracted from 9.56 km² to 2.48 km²; while reclamation for urban expansion was concentrated in the northeastern sector, the remaining margins experienced gradual encroachment. Early reclamation was dominated by construction land use, whereas the later stage shifted toward ecological development. Subsidence was widespread across reclaimed areas but showed spatial heterogeneity, controlled by soil consolidation and static–dynamic loading. Significant subsidence occurred in newly reclaimed and high-load development zones, with deformation rates from − 26.49 to 2.53 mm/a and a maximum cumulative subsidence of 264.11 mm. The proposed hybrid prediction framework captures complex deformation and indicates a deceleration trend toward 2025, while hotspots persist in high-load areas driven by consolidation (reaching annual rates up to 6.920 mm) and near-lake margins exhibit pronounced seasonal instability. This study develops a subsidence prediction framework that achieves high-precision forecasting through deformation decomposition and model integration, providing a basis for urban sustainability and infrastructure safety.