SAR Remote Sensing: From Theory to Deforestation Detection

What is SAR Remote Sensing?

SAR (Synthetic Aperture Radar) is a remote sensing technology that uses radar waves instead of visible light. Its biggest advantage: it works day and night, through clouds.

This is particularly important in Vietnam — where cloud cover is nearly year-round, making optical imagery (Landsat, Sentinel-2) often unusable.

The Two Satellites I Used

Sentinel-1 (ESA)

• C-band (5.4 GHz)
• Revisit cycle: 6 days
• Free and open data
• Good for rapid change monitoring

ALOS-2 (JAXA)

• L-band (1.2 GHz)
• Better canopy penetration → sensitive to forest structure
• Revisit cycle: 14 days
• Paid data (but available for research)

Methodology

1. Data acquisition — download SAR images from Copernicus Hub (Sentinel-1) and JAXA (ALOS-2)
2. Preprocessing — radiometric calibration, speckle filtering, geocoding (using SNAP)
3. Change analysis — multi-temporal comparison, change index calculation
4. Classification — identify deforested areas using thresholds and dual-band combination
5. Validation — compare with ground truth and optical imagery

Results

Combining both satellites (C-band + L-band) yielded significantly better results than using either alone:

• L-band (ALOS-2) is sensitive to major structural changes in forests
• C-band (Sentinel-1) adds information about surface-level changes

Results were published at the GIS-IDEAS 2023 conference.

Tools to Get Started

If you want to start with SAR remote sensing:

• SNAP (ESA) — free SAR image processing software
• QGIS — open-source GIS
• Python + snappy — processing automation
• Google Earth Engine — cloud processing (free for research)