Written and maintained by CASRAI Editorial Board
Last updated
The Copernicus Sentinel-1 satellite mission provides all-weather, day-and-night C-band synthetic aperture radar (SAR) observations. Quantitative analysis of Sentinel-1 radar imagery requires converting raw digital numbers into physically calibrated backscatter coefficients—specifically sigma nought (σ⁰) expressed in decibels (dB)—to support precise land cover classification, surface deformation tracking, and environmental modeling.
Definition in One Paragraph
Copernicus Sentinel-1 is an Earth observation constellation developed by the European Space Agency (ESA) for the European Union Copernicus Programme, carrying a C-band Synthetic Aperture Radar (C-SAR) operating at a central frequency of 5.405 GHz (wavelength ~5.55 cm). Orbiting at a nominal altitude of 693 km in a sun-synchronous polar orbit, the constellation (comprising Sentinel-1A and Sentinel-1C) delivers radar data across four operational modes: Interferometric Wide Swath (IW), Stripmap (SM), Extra Wide Swath (EW), and Wave (WV). In quantitative geospatial analysis, Sentinel-1 functions as a foundational radiometric and interferometric instrument, quantifying surface roughness, soil moisture, and crustal deformation regardless of cloud cover or solar illumination.
Formula and Radiometric Calibration Equations
To conduct quantitative analysis or compare radar scenes across dates, raw digital numbers (DN) from Sentinel-1 Level-1 products must be radiometrically calibrated using look-up tables (LUTs) embedded in the product metadata (Miranda & Meadows, 2015). For Level-1 Ground Range Detected (GRD) and Single Look Complex (SLC) products, the conversion of a pixel intensity at line i and sample j to sigma nought (σ⁰) is defined as:
$$sigma_{i,j}^0 = frac{DN_{i,j}^2 + k}{A_{sigma, i,j}^2}$$
Where:
- σ⁰ (sigma nought): The normalized radar backscatter coefficient (dimensionless radar cross-section per unit ground area, m²/m²).
- DN_{i,j}: Digital number pixel intensity recorded at sample j and line i.
- A_{σ, i,j}: Calibration factor from the
sigmaNoughtlook-up table interpolated at pixel coordinates. - k: Calibration offset constant, equal to 0 for all current Sentinel-1 IPF processing versions.
Linear sigma nought is standardly transformed into decibels (dB) via the logarithmic equation:
$$sigma_{text{dB}}^0 = 10 cdot log_{10}(sigma^0) = 10 cdot log_{10}left(frac{DN^2}{A_sigma^2}right) = 20 cdot log_{10}(DN) – 20 cdot log_{10}(A_sigma)$$
For interferometric SAR (InSAR) surface displacement measurements combining two co-registered SLC images (s₁ and s₂), complex interferometric coherence (γ) is calculated as:
$$gamma = frac{|langle s_1 cdot s_2^* rangle|}{sqrt{langle |s_1|^2 rangle cdot langle |s_2|^2 rangle}}$$
Where |…| denotes magnitude, * denotes complex conjugate, and ⟨…⟩ denotes spatial averaging. Values range from 0 (complete decorrelation) to 1 (perfect interferometric phase stability).
Worked Example: Calculating Calibrated Sigma Nought Backscatter
Consider a practical calibration calculation for two pixels from an IW-mode Sentinel-1 Level-1 GRD image with an interpolated calibration factor of A_σ = 380.0.
Example 1: Urban Infrastructure Target (High Backscatter)
A building reflection produces a digital number of DN = 425.
- Square the digital number: DN² = 425² = 180,625
- Square the calibration factor: A_σ² = 380.0² = 144,400
- Calculate linear sigma nought (σ⁰): σ⁰ = 180,625 / 144,400 = 1.250866
- Convert to decibels (dB): σ⁰_dB = 10 × log₁₀(1.250866) = 10 × 0.097210 = +0.97 dB
The positive value (+0.97 dB) indicates strong double-bounce scattering from vertical structures reflecting microwave pulses back to the sensor.
Example 2: Smooth Water Target (Low Backscatter)
A calm reservoir surface produces a digital number of DN = 75.
- Square the digital number: DN² = 75² = 5,625
- Square the calibration factor: A_σ² = 380.0² = 144,400
- Calculate linear sigma nought (σ⁰): σ⁰ = 5,625 / 144,400 = 0.038954
- Convert to decibels (dB): σ⁰_dB = 10 × log₁₀(0.038954) = 10 × (-1.40945) = -14.09 dB
The low backscatter (-14.09 dB) reflects specular reflection, where calm water acts as a mirror reflecting pulses away from the receiver.
Interpretation: Radar Cross-Section and Backscatter Reading Guidance
Interpreting C-band radar backscatter requires evaluating surface roughness, orientation, and moisture. Unlike optical sensors like Sentinel-2, Sentinel-1 measures microwave interaction with terrain geometry.
| Land Cover Category | Typical VV (dB) | Typical VH (dB) | Scattering Mechanism | Analytical Guidance |
|---|---|---|---|---|
| Calm Open Water | -26 to -18 dB | -32 to -24 dB | Specular reflection | Very dark appearance; wind creates capillary waves increasing backscatter. |
| Smooth Bare Soil | -18 to -12 dB | -26 to -20 dB | Surface scattering | Backscatter rises with moisture due to increased soil dielectric constant. |
| Agricultural Crops | -14 to -8 dB | -22 to -14 dB | Mixed surface/volume | VH backscatter increases as crop canopy volume and biomass expand. |
| Dense Forest Canopy | -12 to -7 dB | -17 to -11 dB | Volume scattering | Stable, high cross-polarization backscatter throughout seasonal cycles. |
| Urban Built-Up Areas | -2 to +6 dB | -12 to -4 dB | Double-bounce reflection | Bright returns from right angles between walls and streets. |
Assumptions and Common Mistakes
When incorporating Sentinel-1 SAR products into spatial data analysis, researchers must navigate specific physical and methodological constraints:
- Neglecting Topographic Distortions: On sloped terrain, foreshortening compresses illuminated areas while layover and shadow distort backscatter. Radiometric Terrain Correction (RTC) using a high-resolution DEM is necessary to normalize backscatter to gamma nought (γ⁰).
- Omitting Speckle Filtering: Coherent radar interference creates granular speckle noise. Spatial filtering (such as Refined Lee or Frost filters) should be applied prior to classification.
- Confusing Beta, Sigma, and Gamma Nought: Beta nought (β⁰) reflects slant-range radar brightness, sigma nought (σ⁰) reflects ground-projected area, and gamma nought (γ⁰) normalizes to the plane perpendicular to the line of sight (γ⁰ = σ⁰ / cos θᵢ). Mixing these creates systematic errors.
- Ignoring Polarization Sensitivities: VV polarization is sensitive to surface roughness and moisture, whereas VH polarization responds primarily to volume scattering from vegetation.
- Comparing Uncalibrated Digital Numbers: Analyzing raw integer DN values across different tracks or dates invalidates multi-temporal change detection algorithms.
How to Report It (APA Example)
When reporting Sentinel-1 radar calibration and backscatter metrics in academic manuscripts adhering to APA 7th edition guidelines, document sensor modes, processing levels, calibration sources, and filtering parameters:
“Synthetic aperture radar data were obtained from the Copernicus Sentinel-1 constellation in Interferometric Wide Swath (IW) dual-polarization (VV and VH) mode. Level-1 Ground Range Detected (GRD) products were radiometrically calibrated to normalized backscatter coefficients (σ⁰ in decibels) using official IPF calibration annotation vectors (European Space Agency, 2026; Miranda & Meadows, 2015). Radiometric terrain correction and speckle suppression (Refined Lee filter, 7 × 7 window) were applied through the Sentinel Application Platform (SNAP v10.0) referencing the Copernicus 30 m Digital Elevation Model. Wetland sample plots yielded a mean cross-polarized backscatter of M = -15.82 dB (SD = 1.34 dB), whereas surrounding agricultural fields exhibited M = -10.45 dB (SD = 1.12 dB).”
How to Calculate It in Excel, R, and Python
Programmatic radiometric calibration can be implemented in standard research computing environments.
Python (NumPy and Rasterio Workflow)
In Python geospatial pipelines utilizing Rasterio, array calibration is executed directly:
import numpy as np
def calibrate_sentinel1_backscatter(dn, a_sigma):
"""
Calculate Sentinel-1 sigma nought linear and decibel backscatter.
"""
dn = np.asarray(dn, dtype=float)
a_sigma = np.asarray(a_sigma, dtype=float)
sigma_nought = (dn ** 2) / (a_sigma ** 2)
sigma_nought_db = 10.0 * np.log10(np.maximum(sigma_nought, 1e-10))
return sigma_nought, sigma_nought_db
# Example calculation
dn_val, a_val = 425, 380.0
sig_lin, sig_db = calibrate_sentinel1_backscatter(dn_val, a_val)
print(f'Linear Sigma0: {sig_lin:.6f}, Backscatter: {sig_db:.2f} dB')
# Output: Linear Sigma0: 1.250866, Backscatter: 0.97 dBFor regional-scale analysis, researchers often employ cloud platforms such as Google Earth Engine, which ingests pre-calibrated Sentinel-1 GRD imagery adhering to Open Geospatial Consortium (OGC) standards.
R (Terra Package Workflow)
In R spatial pipelines, calibration is performed as follows:
# Sentinel-1 Radiometric Calibration in R
calibrate_s1_backscatter <- function(dn, a_sigma) {
sigma_lin <- (dn^2) / (a_sigma^2)
sigma_db <- 10 * log10(pmax(sigma_lin, 1e-10))
return(data.frame(linear = sigma_lin, db = round(sigma_db, 2)))
}
# Example calculation
result <- calibrate_s1_backscatter(dn = 425, a_sigma = 380.0)
print(result)
# Output: linear = 1.250866, db = 0.97Microsoft Excel Spreadsheet Formulas
To calculate backscatter with Digital Number (DN) in cell A2 and calibration factor (A_σ) in cell B2:
- Linear Sigma Nought (σ⁰) in cell C2:
=(A2^2)/(B2^2) - Sigma Nought in Decibels (dB) in cell D2:
=10*LOG10(C2) - Combined Single-Cell Formula:
=10*LOG10((A2^2)/(B2^2))
Frequently Asked Questions
What is Copernicus Sentinel-1?
Copernicus Sentinel-1 is the European Space Agency radar imaging constellation providing continuous, all-weather, day-and-night C-band synthetic aperture radar (SAR) data for global land and ocean observation.
How does Sentinel-1 differ from Sentinel-2?
Sentinel-1 is an active radar satellite emitting microwave pulses that penetrate clouds and rain day and night. Sentinel-2 is a passive optical satellite measuring reflected sunlight across visible and infrared bands, which cannot penetrate cloud cover or operate in darkness.
What is the difference between Level-1 SLC and Level-1 GRD products?
Level-1 Single Look Complex (SLC) products preserve both signal amplitude and phase in slant-range geometry, required for interferometry (InSAR). Level-1 Ground Range Detected (GRD) products provide detected amplitude projected onto an Earth ellipsoid, suitable for land-cover classification and backscatter mapping.
Why is radiometric calibration necessary for Sentinel-1 analysis?
Radiometric calibration converts arbitrary raw digital numbers into normalized radar cross-section values (sigma nought, σ⁰) using official look-up tables, enabling valid physical comparison across different dates, orbits, and sensors.
Which acquisition mode is primarily used for land monitoring?
The Interferometric Wide Swath (IW) mode is the primary operational mode over land, acquiring a 250 km swath at 5 m × 20 m spatial resolution using the TOPSAR technique.
Sources
- European Space Agency. (2026). Sentinel-1 SAR Technical Guide. ESA / Copernicus Programme. https://sentinels.copernicus.eu/web/sentinel/missions/sentinel-1
- Miranda, N., & Meadows, P. J. (2015). Radiometric Calibration of S-1 Level-1 Products Generated by the S-1 IPF (ESA Technical Note ESA-EOPG-CSCOP-TN-0002). European Space Agency. https://sentinels.copernicus.eu/documents/247904/685163/S1-Radiometric-Calibration-V1.0.pdf
- Copernicus Data Space Ecosystem. (2026). Sentinel-1 Product Specifications and Data Access Guide. European Commission and ESA. https://dataspace.copernicus.eu/
- Alaska Satellite Facility Distributed Active Archive Center. (2026). Sentinel-1 Data Processing Recipes and Radiometric Terrain Correction. University of Alaska Fairbanks. https://asf.alaska.edu/data-sets/sar-data-sets/sentinel-1/
