SAR Backscatter Time Series: Detecting Change with All-Weather Radar

The All-Weather Advantage of SAR Time Series

Optical satellite monitoring has a fundamental constraint: clouds block the view. In tropical forests — where deforestation is most rapid — cloud cover can persist for weeks or months, leaving gaps exactly where continuous observation is most needed.

Synthetic Aperture Radar (SAR) transmits its own microwave pulses and measures what reflects back, making it completely independent of sunlight and cloud cover. A Sentinel-1 SAR time series provides observations every 6–12 days regardless of weather, giving you the most temporally complete picture of surface change.

How SAR Backscatter Works

SAR transmits microwaves in the C-Band (5.405 GHz, ~5.6 cm wavelength) and records the amplitude of the signal that returns to the sensor. This return — called backscatter — depends on:

• Surface roughness relative to the wavelength
• Dielectric properties (water content affects electrical response)
• Geometry of the scattering surface
• Vegetation structure (volume vs. surface scattering)

Unlike optical sensors that measure reflected sunlight, SAR literally measures how the land surface interacts with a microwave pulse — which is why it responds differently to water, bare soil, crops, forests, and buildings.

The Key SAR Indices for Monitoring

VV Intensity

VV (vertical transmit, vertical receive) backscatter is most sensitive to:

• Large flat surfaces — Roads, runways, calm water
• Urban structures — Buildings create strong double-bounce returns
• Surface roughness — Rough soil or crop residue scatter more than smooth bare soil
• Moisture — Wet soil has higher dielectric constant → stronger backscatter

Monitoring applications:

• Flood mapping (water reflects away → sudden VV drop)
• Construction tracking (new buildings → VV increase)
• Soil moisture estimation

VH Intensity

VH (vertical transmit, horizontal receive) measures cross-polarized backscatter, which requires volume scattering — the signal must bounce multiple times within a volume (like a forest canopy) before returning with changed polarization.

VH is sensitive to:

• Vegetation volume — Dense canopies produce strong VH
• Surface roughness at smaller scales
• Penetration depth into vegetation

Monitoring applications:

• Deforestation detection (VH drops as canopy is removed)
• Crop biomass estimation (VH grows as crops develop)
• Forest degradation before complete clearing

RVI — Radar Vegetation Index

Radar Vegetation Index isolates the vegetation signal from combined polarization measurements:

RVI = (4 × VH) / (VV + VH)

RVI ranges from 0 (bare surface) to 1 (dense random-volume vegetation). It is less sensitive to surface moisture than VV alone and provides a cleaner vegetation signal.

Key properties:

• Increases as vegetation density grows
• Decreases with canopy loss, drought stress, or harvest
• More robust to soil moisture variation than raw VH

RFDI — Radar Forest Degradation Index

Radar Forest Degradation Index is designed specifically for forest monitoring:

RFDI = (VV − VH) / (VV + VH)

In undisturbed forest, VH is relatively high (volume scattering dominates) → RFDI is low. When forest is degraded or selectively logged, VH decreases relative to VV → RFDI increases.

RFDI is particularly sensitive to partial degradation — the kind of damage from selective logging or illegal logging that does not show as complete clearing but still represents significant ecological impact.

CR — Cross Ratio

Cross Ratio is simply:

CR = VH / VV

High CR indicates volume scattering dominance (vegetation). Low CR indicates surface scattering dominance (water, bare soil). CR is useful for:

• Distinguishing flooded vegetation from open water (flooded vegetation has higher CR than open water)
• Monitoring crop development stages

Characteristic SAR Time Series Patterns

Open Water and Flooding

Water bodies have extremely low SAR backscatter — calm water acts like a mirror and reflects the radar pulse away from the sensor (specular reflection). This makes floods highly visible in SAR data.

Flood event signature:

• Sudden drop in VV intensity by 5–10 dB
• Often visible within one Sentinel-1 overpass (~6 days)
• CR increases for flooded vegetation (double-bounce through standing water)

Dense Forest

Undisturbed forest shows relatively stable, high VH and moderate VV:

• VH is high due to volume scattering from the canopy
• Year-to-year variation is low (unlike optical NDVI which has seasonal cycles)
• Sudden drops signal clearing events

Agricultural Fields

Crops show strong seasonal patterns in SAR:

1. After harvest/tillage: low VV and VH (smooth bare soil)
2. Early growth: gradual VH increase
3. Peak canopy: high VH, moderate VV
4. Senescence/harvest: rapid drop in both

The exact timing and magnitude differ by crop type, making multi-year SAR time series useful for crop type mapping.

Urban Environments

Urban areas show high and stable VV backscatter from buildings (double-bounce between vertical walls and the ground). New construction appears as a step increase; demolition as a step decrease.

Comparing SAR and Optical Monitoring

Aspect	SAR (Sentinel-1)	Optical (Sentinel-2)
Cloud penetration	Yes	No
Night observation	Yes	No
Vegetation sensitivity	Volume structure	Photosynthesis (color)
Flood detection	Excellent	Limited (clouds during floods)
Crop type discrimination	Moderate	Good
Seasonal cycle visibility	Moderate	Strong
Fire/burn detection	Limited	Excellent (NBR)

SAR and optical are complementary, not competitive. The most powerful monitoring systems use both.

Setting Up SAR Monitoring

For a comprehensive SAR monitoring workflow:

1. Select VV + VH + RVI for a vegetation zone — this gives you both raw backscatter and a cleaner vegetation signal
2. Add RFDI for forest areas where you suspect gradual degradation
3. Use SAR alongside NDVI — when Sentinel-2 has cloud gaps, SAR fills in
4. Set a start date 12+ months back to capture at least one full seasonal cycle for agricultural areas

Interpreting Anomalies in SAR Time Series

Unlike optical NDVI, SAR anomalies require understanding the direction of change:

Change	VV	VH	RVI	Likely Cause
Drop	Low	Low	Low	Clearing, harvest, flood
Rise	High	—	—	Construction, soil wetting
VH drop, VV stable	—	Low	Low	Canopy loss (selective logging)
Both drop sharply	Low	Low	Low	Major flood or complete clearing
VH rise over season	—	High	High	Crop growing season

Handling SAR Noise

SAR images contain inherent speckle noise — a grainy texture caused by coherent interference of radar waves from many small scatterers. This noise creates scatter around the true time series trend.

Time series analysis helps because:

• Multiple observations let you see the trend through the noise
• Averaging over a polygon area reduces single-pixel noise
• Statistical anomaly detection distinguishes real events from noise fluctuations

Summary

SAR backscatter time series provides an all-weather, day-and-night monitoring capability that optical sensors cannot match. VV intensity detects surface changes including flooding and construction; VH and RVI track vegetation volume and growth; RFDI isolates forest degradation. Understanding the expected "signature" of different land cover types in SAR time series is the key to correctly interpreting anomalies — combining SAR with optical NDVI gives the most complete and cloud-resilient monitoring system available from open satellite data.