A Decade of Ground Deformation in Kunming (China) Revealed by Multi-Temporal Synthetic Aperture Radar Interferometry (InSAR) Technique

Wu Zhu, Wen-Liang Li, Qin Zhang, Yi Yang, Yan Zhang, Wei Qu, Chisheng Wang2019 · DOI

TL;DR

This study uses a decade of satellite radar data (ALOS-1 and COSMO-SkyMed) to map ground deformation in Kunming, China, revealing significant subsidence. The findings highlight the combined impact of soft soil consolidation, building loads, and groundwater extraction as key drivers of this subsidence, offering crucial evidence for urban planning and disaster mitigation.

What Problem Does This Paper Address?

Rapid urbanization in Kunming has led to severe ground subsidence, posing a significant threat to urban infrastructure and safety. While previous studies have identified subsidence in the region, they often lacked the spatial resolution and temporal coverage to fully understand its evolution and causes. This paper aims to fill this gap by providing a comprehensive, decade-long (2007-2016) analysis of ground deformation in Kunming using multi-temporal InSAR, offering detailed insights into the spatial patterns and temporal variations of subsidence. This detailed understanding is crucial for effective disaster prevention and mitigation strategies in this rapidly developing urban area.

Methods

The core methodology employed is multi-temporal Interferometric Synthetic Aperture Radar (InSAR). This technique leverages multiple SAR images acquired over time to detect subtle ground surface movements. Specifically, the researchers used:

  • Interferogram Generation: Pairs of SAR images were created with strict temporal and spatial baseline constraints to minimize noise and maintain coherence. Different baseline criteria were applied for the L-band (ALOS-1) and X-band (COSMO-SkyMed) data due to their differing wavelengths and sensor characteristics.
  • Phase Unwrapping and Correction: After generating interferograms, topographic and orbital phases were removed using a high-resolution DEM and orbital data. Adaptive spectral filtering was applied to suppress noise. The minimum cost flow (MCF) method was used for phase unwrapping to retrieve absolute phase. Residual phase errors (orbital and atmospheric) were mitigated using quadratic polynomial fitting and spatial/temporal filtering.
  • Time Series Deformation Estimation: The processed interferograms were used to estimate the time series deformation.
  • Vertical Deformation Conversion: Line-of-sight (LOS) deformation derived from InSAR was converted into vertical deformation, assuming that horizontal deformation was negligible in Kunming, a common assumption for this region based on prior studies.
  • Correlation Analysis: The derived deformation data was correlated with geological factors (soft soil consolidation, sediment thickness), building density, and groundwater exploitation data to identify the primary causes of subsidence.

Datasets

The study utilized the following SAR datasets:

  • L-band ALOS-1: 20 images acquired between January 2007 and March 2011 (ascending orbit).
  • X-band COSMO-SkyMed (CSK): 40 images acquired between June 2011 and January 2016 (descending orbit).
  • C-band Sentinel-1A: 91 additional images were used for validation purposes (ascending orbit, January 2015 - February 2017).

Additionally, the study incorporated:

  • Leveling Data: 37 leveling points were used for validation.
  • Light Detection and Ranging (LIDAR) Digital Elevation Model (DEM): A 3m resolution DEM was used to remove topographic phase.
  • Geological Data: Information on quaternary sediment thickness, active faults, and borehole data.
  • Land Cover Data: Images from 1984 and 2016 to assess urbanization.
  • Groundwater Data: Information on groundwater exploitation regions and water level changes.

Key Insights and Ideas

  • Decadal Subsidence Revealed: The study provides the first decade-long (2007-2016) view of ground deformation in Kunming, revealing a clear pattern of subsidence.
  • L-band vs. X-band Differences: A subtle but notable difference in subsidence magnitude was observed between L-band (ALOS-1) and X-band (COSMO-SkyMed) data. This was attributed to differing acquisition times (and thus varying urbanization/extraction rates) and imaging geometries, rather than sensor limitations.
  • Multiple Subsidence Drivers: The research strongly suggests that Kunming's subsidence is not due to a single factor but a combination of:
  • Soft Soil Consolidation: The city's location in a lacustrine basin filled with compressible soft soils (clay, muddy clay, peaty soils) forms the fundamental basis for subsidence.
  • Building Load: The rapid increase in high-rise buildings and dense construction exerts significant pressure on these soft soils, accelerating consolidation.
  • Groundwater Extraction: Intensive groundwater withdrawal, particularly in areas like Guandu, directly contributes to the compaction of subsurface layers.
  • Spatial Heterogeneity: Subsidence is not uniform, with two distinct severe subsiding regions identified (Xishan and Guandu), indicating localized geological conditions and anthropogenic pressures.
  • Validation is Key: The comparison of InSAR results with ground-based leveling data demonstrated the reliability of the satellite-derived measurements, with low RMSE values.

Key Results

  • Annual Deformation Velocity: The study mapped annual deformation velocities, identifying two severe subsiding regions with a maximum subsidence rate of 35 mm/year (using ALOS-1 data).
  • Cumulative Subsidence: Over the 2007-2016 period, cumulative subsidence reached 210 mm in Guandu and 98 mm in Xishan.
  • Validation Accuracy: Root-mean-square error (RMSE) values for the comparison between InSAR and leveling data were approximately 4.5 mm for L-band and 3.7 mm for X-band, indicating high reliability.
  • Correlation Coefficients:
    • A negative correlation (around -0.65) was found between ground deformation and quaternary sediment thickness, suggesting thicker soft soil layers contribute to more subsidence.
  • A negative correlation (around -0.61) was observed between ground deformation and building density, indicating that areas with higher building density experience more subsidence.
  • Groundwater Level Decline: In Guandu, groundwater levels declined by up to 12.1 m between 2004 and 2012, correlating with subsidence trends.

Limitations and Future Work

  • Horizontal Deformation: While assumed negligible, the paper acknowledges that horizontal deformation components were not explicitly measured, which could provide a more complete picture of the 3D deformation.
  • Mechanism Clarity: The authors note that while the primary drivers are identified, a "clear description of the mechanism of land subsidence" is difficult without more continuous ground-based monitoring data.
  • Future Work: The researchers plan to collect sufficient in situ measurements in the future to further analyze the mechanism of ground deformation in Kunming. This will likely involve more detailed geological surveys and continuous ground-based deformation monitoring.

Summary updated

BibTeX
@misc{zhu2019decade,
  author = {Wu Zhu and Wen-Liang Li and Qin Zhang and Yi Yang and Yan Zhang and Wei Qu and Chisheng Wang},
  title = {{A Decade of Ground Deformation in Kunming (China) Revealed by Multi-Temporal Synthetic Aperture Radar Interferometry (InSAR) Technique}},
  year = {2019},
  doi = {10.3390/s19204425},
}
Figure