Imagine trying to measure a city sinking by fractions of a millimeter from an orbit 700 kilometers above the Earth. Traditional satellite imagery captures changes in the landscape's appearance, but it cannot easily resolve sub-centimeter shifts in elevation over time.
Synthetic Aperture Radar (SAR) changes this paradigm by treating the radar signal not just as an image, but as a wave with a measurable phase. By comparing the phase of two radar images taken over the same area at different times—a technique known as Differential InSAR (DInSAR)—we can map ground displacement. However, a single pair of images is highly vulnerable to noise: atmospheric water vapor slows down the radar wave, creating artifacts, and changes in vegetation or soil moisture scramble the signal entirely.
To overcome these limitations, modern geodetic research relies on Time-Series InSAR. By stacking dozens of radar images collected over months or years, algorithms can track the phase history of specific points on the ground, effectively filtering out transient atmospheric noise to reveal long-term deformation trends.
The Core Mechanism: Phase Differencing
When a satellite emits a radar pulse, the wave travels to the ground, bounces off a target, and returns to the sensor. The returning wave's phase depends directly on the distance it traveled. If the ground moves even a fraction of a wavelength between two satellite passes, the phase of the returning wave shifts.
By subtracting the phase of the first image from the second, we create an interferogram. This raw phase difference ($\Delta \phi$) contains several components:
$$\Delta \phi = \phi_{\text{deformation}} + \phi_{\text{topography}} + \phi_{\text{atmosphere}} + \phi_{\text{noise}}$$
To isolate $\phi_{\text{deformation}}$, researchers use a Digital Elevation Model (DEM) to simulate and subtract the topographic phase ($\phi_{\text{topography}}$). The real challenge lies in separating the true ground movement from the atmospheric phase delay ($\phi_{\text{atmosphere}}$) caused by weather variations between acquisitions. Time-series analysis solves this by exploiting the fact that atmospheric noise is random over time, whereas ground deformation is typically continuous or seasonal.
Persistent Scatterer Interferometry (PS-InSAR)
In urban environments, the ground is covered by engineered structures like concrete buildings, bridges, and metallic fixtures. These objects reflect radar waves exceptionally well and remain physically unchanged over years.
In Combined PS-InSAR Technology and High-Resolution Optical Remote Sensing for Identifying Illegal Underground Mining in the Suburb of Yangquan City, Shanxi Province, China, Yuanping Xia et al. demonstrate how these stable structures act as "Persistent Scatterers" (PS). Because a PS pixel maintains its structural and reflective properties over a long time series, its phase history is incredibly clean.
By analyzing pixels with low amplitude dispersion across a large stack of images, PS-InSAR algorithms can isolate and remove atmospheric noise. This allows researchers to track millimeter-level movements of individual buildings. For instance, Bahaa Mohamadi et al. leverage this high-precision tracking in Towards a PS-InSAR Based Prediction Model for Building Collapse: Spatiotemporal Patterns of Vertical Surface Motion in Collapsed Building Areas - Case Study of Alexandria, Egypt to identify structural deformation patterns that act as precursors to catastrophic building collapses.
PS-InSAR Strategy:
[Image Stack] ---> Identify Pixels with Stable Reflection (Buildings/Rock)
---> Filter Atmospheric Noise Across Time
---> Millimeter-Scale Displacement Map
Small Baseline Subset (SBAS-InSAR)
While PS-InSAR excels in cities, it struggles in natural landscapes. Vegetation grows, soil moisture changes, and open fields decorrelate quickly, meaning they lose phase coherence between distant satellite passes.
To map non-urban areas, the Small Baseline Subset (SBAS) method takes a different approach. Instead of linking every image back to a single master image, SBAS only pairs images that have small spatial and temporal separations (baselines). This minimizes decorrelation, allowing the algorithm to track "Distributed Scatterers"—pixels that contain shared, natural features like gravel plains or sparse vegetation.
As highlighted by Yahui Guo et al. in Ground Deformation Analysis Using InSAR and Backpropagation Prediction with Influencing Factors in Erhai Region, China, this short-baseline strategy creates a highly redundant network of interferograms. Because these pairs form isolated subsets in time, the algorithm uses Singular Value Decomposition (SVD) to mathematically link the subsets together. This allows for the retrieval of a continuous, long-term deformation velocity field across entire regions, a mechanism deployed by Huanghao Yu et al. in Surface Subsidence over a Coastal City Using SBAS-InSAR with Sentinel-1A Data: A Case of Nansha District, China to chart complex subsidence driven by soft coastal soils and rapid urbanization.
Resolving Multi-Dimensional Motion
A fundamental constraint of SAR is that the satellite only measures displacement along its Line-of-Sight (LOS)—the diagonal line connecting the radar sensor to the ground target. If a hillside moves purely horizontally, a single LOS measurement might misinterpret it as vertical sinking or miss it entirely depending on the viewing angle.
To resolve true multi-dimensional motion, researchers combine data from different orbital trajectories:
- Ascending Orbits: The satellite travels from south to north, looking eastward.
- Descending Orbits: The satellite travels from north to south, looking westward.
By combining the LOS displacement vectors from both ascending and descending passes over the same area, the true vertical and east-west horizontal components of ground motion can be geometrically decoupled.
Zhiyong Wang utilizes this dual-orbit combination in Resolving and Analyzing Landfast Ice Deformation by InSAR Technology Combined with Sentinel-1A Ascending and Descending Orbits Data to separate the vertical flexing of sea ice from its horizontal drift. Similarly, Qi Chen et al. apply this multi-orbit decomposition in Accessing the Time-Series Two-Dimensional Displacements around a Reservoir Using Multi-Orbit SAR Datasets: A Case Study of Xiluodu Hydropower Station to demonstrate that ground deformation near massive reservoirs is often a complex, two-dimensional response to water weight and hydrological pressure rather than simple vertical subsidence.
Ascending LOS (Looking East) \ / Descending LOS (Looking West)
\ /
v v
[Deformation Target]
|
Decomposed into: [Vertical] + [East-West Horizontal]
From Reactive Mapping to Proactive Early Warning
The ultimate value of time-series InSAR lies in shifting hazard management from a reactive posture to a proactive one. Traditional geological risk mapping relies heavily on historical inventories—recording where landslides or collapses have occurred in the past.
However, geomorphic failures are preceded by subtle, accelerating strain. In Evaluation of Landslide Susceptibility of the Ya'an-Linzhi Section of the Sichuan-Tibet Railway based on Deep Learning, Wang Shibao et al. demonstrate that integrating active InSAR deformation data into predictive models allows infrastructure managers to transition from static spatial susceptibility maps to dynamic early warning systems.
By capturing the real-time "breathing" and creep of unstable slopes before they fail catastrophically, advanced time-series interferometry provides the empirical measurements required to forecast giant landslides and protect critical infrastructure corridors, as detailed by Hengxing Lan et al. in Remote Sensing Precursors Analysis for Giant Landslides.