Synthetic Aperture Radar (SAR)

Radar backscatter (σ⁰) measures the fraction of transmitted energy returned to the sensor. Influenced by surface roughness, dielectric constant (moisture), and incidence angle. C-band (5.4 GHz, Sentinel-1) penetrates vegetation canopy partially, while L-band (ALOS-2) achieves deeper penetration.

SAR systems transmit and receive in different polarisation configurations: VV (vertical-vertical), VH (vertical-horizontal), HH, and HV. Cross-pol (VH/HV) is more sensitive to vegetation volume scattering, while co-pol (VV/HH) responds to surface and double-bounce scattering.

By comparing phase differences between two SAR acquisitions, InSAR measures surface displacement at millimetre precision. Applications include subsidence monitoring, volcanic deformation, earthquake co-seismic slip, and glacier flow velocity.

Coherence quantifies the phase stability between SAR image pairs. High coherence indicates stable surfaces (urban, rock). Low coherence (temporal decorrelation) indicates change — vegetation growth, flooding, or human activity.

SAR images exhibit granular noise (speckle) caused by coherent imaging. Multi-looking and adaptive filters (Lee, Frost, Gamma-MAP) reduce speckle while preserving edges and structural information.

Decomposition theorems (Freeman-Durden, Cloude-Pottier, Yamaguchi) separate the scattering mechanisms into surface, double-bounce, and volume components for land-cover classification.