SAR Subsurface Radar Machine Learning Mapper

SAR Radar Calibration PHYSICS ENGINE

Microwave Frequency Band

Polarization Mode HH

Incidence Angle (θ) 35°

Subsurface Sand Dielectrics

Volumetric Soil Moisture (mv) 0.5%

Target Sand Depth 4.5m

Archaeological Presets

2D SAR Spatial Plan View (120m x 80m)

Subsurface Attenuation & Radar Profile Slices (δz Depth)

ML Feature Extraction CNN PIPELINE

1. Lee Speckle Filter Reduces multiplicative radar noise

2. Dielectric Permittivity Deconv Calculates subsurface contrast ratio

3. U-Net Architecture Classifier Reconstructs linear ancient road geometry

Calculated Radar Metrics

Penetration (δz)

3.82 m

Atten. Rate (α)

1.18 dB/m

Dielectric Constant

2.85 ε'

Signal SNR

14.2 dB

Archaeological Survey Telemetry

Selected Band: L-band (1.25 GHz)

Max Skin Depth: 3.82 m

Detected Structures: 3 Ancient Features

Classification Confidence: 94.8%