Multi-Temporal SAR Color Composites: See Change in a Single Image

The Concept

Standard SAR images are grayscale — a single date, a single band, displayed in shades of gray. Comparing changes requires toggling between dates, which is slow and makes it hard to see the spatial pattern of change.

Multi-temporal color composites solve this by encoding time as color:

• Red channel → SAR image from Date 1
• Green channel → SAR image from Date 2
• Blue channel → SAR image from Date 3

The result is a single color image where color indicates change and gray indicates stability.

Reading the Colors

Understanding what the colors mean is straightforward once you know the RGB assignment:

Color	Meaning
White/Gray	Bright in all three dates → stable, no change
Black	Dark in all three dates → stable (e.g., calm water)
Red	Bright only on Date 1 → surface changed after Date 1
Green	Bright only on Date 2 → temporary change on Date 2
Blue	Bright only on Date 3 → new feature appeared by Date 3
Yellow (Red + Green)	Bright on Dates 1 and 2, dark on Date 3 → disappeared by Date 3
Cyan (Green + Blue)	Bright on Dates 2 and 3 → appeared after Date 1
Magenta (Red + Blue)	Bright on Dates 1 and 3, not Date 2 → temporary disappearance on Date 2

Why This Works So Well for SAR

This technique is particularly powerful for SAR because:

1. SAR is weather-independent — You can reliably get images at regular intervals without cloud gaps
2. SAR is single-band — Each date naturally maps to one RGB channel (optical requires more complex band selection)
3. SAR backscatter changes are meaningful — Changes in surface roughness, moisture, or structure produce clear backscatter differences
4. Sentinel-1's 6-day repeat — Provides dense temporal coverage for selecting optimal date combinations

Best Applications

Flood Monitoring

Choose dates spanning a flood event:

• R: Pre-flood
• G: During flood
• B: Post-flood

Flooded areas (dark during flood, bright before and after) appear magenta. Permanently flooded areas stay black. Areas with residual flooding after the event appear red (bright pre-flood, dark during and after).

Agricultural Crop Monitoring

Choose dates at different growth stages:

• R: Planting season
• G: Peak growth
• B: Harvest

Different crops with different phenologies show different color patterns. A field that is bare→green→bare will appear green. A field that stays vegetated across all three dates appears gray.

Urban Development

Choose dates months or years apart:

• R: Earliest date
• G: Middle date
• B: Latest date

New construction appears cyan or blue (bright in later dates due to new hard surfaces). Demolished buildings appear red (bright only in the earliest date).

Ship Traffic Patterns

Ships appear as bright points in SAR. In a multi-temporal composite:

• A ship present on only one date appears in that date's color
• Shipping lanes where vessels are frequently present appear as colored streaks
• Ports and anchorages with persistent vessel presence appear gray/white

How to Create a Multi-Temporal Composite

Step 1: Select Three Dates

Choose dates that bracket your event or span your monitoring period. For event detection:

• Include a pre-event date, event date, and post-event date

For general monitoring:

• Space dates evenly (e.g., 3-month intervals for seasonal analysis)

Step 2: Ensure Consistent Processing

All three SAR scenes should have:

• Same orbit direction (ascending or descending)
• Same polarization (VV or VH)
• Same relative orbit number (same viewing geometry)

Mixing orbit directions or polarizations will introduce artificial color differences unrelated to surface change.

Step 3: Load and Assign Channels

In Off-Nadir Delta, load Sentinel-1 SAR imagery from your three chosen dates. Use the layer manager to view them as separate layers and toggle between them to visually confirm the changes you're looking for.

Step 4: Interpret Carefully

Remember:

• Color = change, not a specific surface type
• The same surface change can appear as different colors depending on your date assignment
• Very bright or very dark areas in all dates will appear white or black regardless of subtle changes

Tips for Better Composites

Use VH Polarization for Vegetation Change

VH (cross-polarization) is more sensitive to volume scattering from vegetation canopy. Deforestation and crop growth produce stronger signals in VH than VV.

Use VV Polarization for Water and Urban Change

VV (co-polarization) is more sensitive to surface scattering. Flood extent, ship detection, and urban change show better contrast in VV.

Match Orbit Directions

Always use scenes from the same orbit direction (ascending or descending). Mixing them changes the look angle, which changes the backscatter even if nothing on the ground changed — creating false color in your composite.

Consider Speckle

SAR speckle creates pixel-level noise that produces speckled color in composites. For cleaner results, apply speckle filtering before compositing, or use multi-look products.

The Bigger Picture

Multi-temporal composites are a visualization technique — they don't quantify change or produce statistics. But they're unmatched for exploratory analysis: scanning a large area to identify where changes occurred, what spatial patterns they form, and how they evolved over time.

Once you've identified interesting change areas in the composite, you can then apply quantitative methods like difference analysis to measure the change precisely.

Try It in Off-Nadir Delta

1. Open Sentinel-1 SAR viewer
2. Search for three dates over your area of interest
3. Add all three to the map as separate layers
4. Use the layer manager to toggle between dates and identify change areas
5. Compare with Sentinel-2 optical imagery to confirm the nature of detected changes