Processing Workflow

Workflow Breakdown: From Raw Radar Data to Millimeter-Level Deformation Products

Each SAR satellite generates over 5TB of raw data daily. This guide details the standardized InSAR processing workflow aligned with international benchmarks like ESA's Sentinel-1 mission and NASA-ISRO NISAR system, revealing the technical core of professional-grade processing.

Data Acquisition & Preprocessing: Establishing Interferometric Foundations

Satellite Data Selection Strategy
  1. Band Matching

    C-band (Sentinel-1): Optimal for short-term deformation monitoring.
    L-band (ALOS-2): Superior vegetation penetration; X-band (TerraSAR-X): 0.25m resolution.

  2. Spatiotemporal Baseline Optimization

    Dijkstra algorithm selects optimal interferometric pairs with vertical baseline <300m (C-band) and compliant temporal intervals.

  3. Global Data Source Integration

    ESA Copernicus: Automated Sentinel-1 SLC downloads
    ASF DAAC: ALOS/PALSAR-2 data stream integration
    Commercial APIs: On-demand tasking via ICEYE/Capella Space


Precision Orbit Correction
  1. POE Orbit Refinement

    ESA's Precise Orbit Ephemerides (accuracy <5 cm) eliminate satellite positioning errors.

  2. Baseline Refinement Model

    SVD-computed relative orbit parameters control spatial baseline errors within 1%.

  3. Doppler Centroid Correction

    Compensates azimuth spectrum shift for sliding spotlight mode data.


Radiometric Calibration & Noise Suppression
  1. Absolute Radiometric Calibration

    Converts DN values to σ0 backscatter coefficients using corner reflectors or Amazon rainforest targets.

  2. Multi-looking

    4:1 (range:azimuth) ratio balances resolution and noise for enhanced SNR.

  3. Adaptive Filtering

    Goldstein-Werner filter with slope-adjusted strength (optimal 32×32px window).

Core Interferometric Processing: Decoding Phase Information

Interferogram Generation & Flat-Earth Phase Removal
  1. Complex Data Registration

    Cross-correlation method achieves 0.001-pixel subpixel accuracy.

  2. Flat-Earth Phase Simulation

    Calculates theoretical phase using orbital parameters and DEM data (e.g., SRTM 30m).

  3. Residual Orbit Correction

    Polynomial models remove long-wave phase gradients (residual <1 rad).


Phase Unwrapping: From Wrapped to Absolute Deformation
  1. Minimum Cost Flow Algorithm

    Triangulated network for coherence >0.3 areas (error propagation <5%).

  2. Multi-scale Strategy

    Coarse-scale: Low-res deformation trend surface
    Fine-scale: Branch-cut method for high-coherence details
    AI-enhanced: U-Net model detects phase jumps (3× efficiency gain)

  3. Atmospheric Delay Correction

    MERRA-2 meteorological data builds APS models
    Spatiotemporal filtering (20km cutoff wavelength)
    GNSS fusion improves accuracy to ±1.5mm

Deformation Modeling & Product Generation

Time Series Inversion

1. SBAS Algorithm: Redundant network (15 interferograms/pixel) with SVD solution
2. PS-InSAR: Selects targets (amplitude dispersion <0.25) using phase double-difference model


Geocoding & Validation

Converts to WGS84/UTM projections for ground validation:

1. GNSS cross-validation (R²>0.95)
2. Leveling routes (RMSE ±2.3mm)
3. Monte Carlo error ellipses


Engineering-Grade Outputs

1. Standard Formats

-- GeoTIFF: Deformation rate (mm/yr)
-- CSV: Time series (UTC millisecond precision)
-- KMZ: Google Earth overlays

2. API Services (Beta)

-- RESTful API for threshold alerts
-- Python/Matlab SDKs


"Industrialized InSAR processing is redefining surface monitoring boundaries." As professional global providers, we deliver millimeter accuracy with 100% data traceability for reliable deformation intelligence.
Perv
What is InSAR?
Next
Data Security