Zhuojin Pipe Fitting Co., LtdComparative Analysis of Two-Pass and Three-Pass Differential Interferometry Using 1:250,000 Terrain Data
Literature Overview
The paper by Liu Jiahang, Shan Xinjian, Qin Xuwen, and Zhang Guifang, published in 2003 in Remote Sensing Information (Vol. 25, Issue 1, pp. 4-6), compares two-pass and three-pass differential interferometric processing using 1:250,000 scale terrain data. The study uses the Ms 7.6 Mani earthquake in Tibet (November 8, 1997) as a case study, employing two pre-event and one post-event ERS SAR images. Funded by the Seismology Joint Fund (102096), the research evaluates whether lower-resolution terrain data can serve as external DEM input for two-pass differential interferometry.
Core Technical Content
The study evaluates whether 1:250,000 scale topographic data can serve as an external Digital Elevation Model (DEM) for two-pass differential interferometry. The three-pass approach uses the first two images to generate a DEM and the third image for deformation measurement, while the two-pass approach uses external terrain data as the reference DEM. The comparison reveals that both methods produce consistent co-seismic deformation field distributions overall, with good agreement near the fault zone, though differences increase with distance from the fault due to residual terrain effects.
Technical Analysis
Processing Methodologies Comparison
| Aspect | Two-Pass Differential | Three-Pass Differential |
|---|---|---|
| Number of SAR images required | 2 (1 pre + 1 post) | 3 (2 pre + 1 post) |
| DEM source | External 1:250,000 terrain data | Generated from first two images |
| Processing complexity | Lower | Higher |
| Temporal decorrelation sensitivity | Lower (shorter span) | Higher (longer span needed) |
| Terrain accuracy dependency | High (external DEM quality critical) | Self-calibrating |
| Data acquisition cost | Lower | Higher |
| Processing time | Shorter | Longer |
| Atmospheric error correlation | Partially correlated | Better temporal decorrelation of atmospheric effects |
Key Findings from Comparative Analysis
The comparison reveals several important patterns:
- Both methods produce consistent co-seismic deformation field distributions in overall fringe pattern structure
- Near the fault zone (within approximately 10 km perpendicular distance), both methods show good quantitative agreement
- With increasing perpendicular distance from the fault, residual terrain errors dominate fringe formation in the two-pass method
- Differences between the two methods increase progressively with distance from the fault
- The 1:250,000 terrain data proves adequate for near-fault deformation measurement under current ERS satellite resolution
Error Sources in Differential Interferometry
| Error Source | Magnitude | Effect on Deformation Measurement | Mitigation Strategy |
|---|---|---|---|
| Residual terrain error | Dominates far from fault | Spurious fringes mimicking deformation | High-resolution DEM or three-pass method |
| Atmospheric delay | Variable (1-10 cm equivalent) | Wavelength-scale phase errors | Temporal correlation, multi-pass processing |
| Orbit error | Systematic phase ramp | Geometric distortion | Precision orbit determination (DORIS) |
| Temporal decorrelation | Location-dependent | Loss of coherence, noisy fringes | Short time span between passes |
| Topographic data accuracy | Scale-dependent (±10-50 m at 1:250,000) | Terrain residual phase | Careful DEM registration and interpolation |
| Baseline decorrelation | Geometry-dependent | Reduced coherence at large perpendicular baselines | Baseline selection criteria |
Performance Assessment Criteria
| Criterion | Two-Pass Method | Three-Pass Method | Assessment |
|---|---|---|---|
| Near-fault accuracy | ±2-3 cm | ±1-2 cm | Both acceptable for tectonic analysis |
| Far-field reliability | Decreases with distance | More consistent | Three-pass preferred for regional studies |
| Processing efficiency | High | Moderate | Two-pass advantageous for time-critical applications |
| Data requirement | Minimal | More extensive | Two-pass preferred when data limited |
| Terrain dependence | Significant | Self-correcting | Three-pass preferred in rough terrain |
Engineering Practice Implications
For engineers and scientists working in InSAR-based deformation monitoring, this study provides practical guidance on DEM selection and processing method choice. The conclusion that 1:250,000 terrain data is sufficient for two-pass differential interferometry under current ERS satellite resolution significantly reduces data acquisition costs and processing complexity, particularly important for developing regions where high-resolution DEM data may be unavailable or prohibitively expensive.
The finding that differences between methods increase with distance from the deformation source has important implications for interpretation methodology. Engineers should focus on near-fault regions where both methods agree for quantitative analysis, and exercise caution when interpreting far-field measurements obtained with lower-resolution terrain data.
Practical Decision Framework
| Application Scenario | Recommended Method | Justification |
|---|---|---|
| Near-fault deformation (within 10 km) | Two-pass with 1:250,000 DEM | Adequate accuracy, lower cost |
| Regional deformation field (>10 km) | Three-pass | Better terrain error suppression |
| Time-critical monitoring | Two-pass | Faster processing, fewer images needed |
| High-precision tectonic analysis | Three-pass or two-pass with high-res DEM | Maximum accuracy required |
| Limited data availability | Two-pass | Fewer images required |
Study Insights and Reflections
This paper exemplifies the practical engineering approach of comparing alternative methods to determine when simplified approaches are acceptable. The systematic comparison methodology—analyzing agreement patterns as a function of distance from the fault—provides a clear framework for understanding error propagation in interferometric measurements. The study demonstrates that methodological choices should be driven by quantitative error analysis rather than assumptions about data quality requirements.
The work demonstrates that engineering decisions about data quality requirements should be based on quantitative comparison rather than assumptions. The conclusion that lower-resolution terrain data is adequate for specific applications represents a significant practical contribution that can guide resource allocation in monitoring programs. This is particularly relevant for earthquake early warning and post-seismic monitoring applications where rapid processing with available data is essential.
The paper also illustrates an important principle in remote sensing: the adequacy of a data source depends on the specific application requirements and the spatial scale of the phenomenon being measured. What may be insufficient for one application can be entirely adequate for another, and this must be determined through systematic evaluation rather than blanket assumptions about data quality.