Module 03 of 11 · Academy overview

What is the radar cross section?

The Radar Cross Section (RCS), denoted σ, is a measure of an object's ability to scatter electromagnetic energy back towards the radar transmitter. It has the unit of area (m²), but it is not a geometric area — it is an effective scattering area that depends on wavelength, polarization, aspect angle and materials.

Formal definition:

σ = limR→∞ 4πR² · |Es|² / |Ei|²

Here R is the distance to the target, Es is the scattered electric field and Ei is the incident electric field. The limit R→∞ ensures that we measure in the far field (plane wavefront).

In practice, RCS is often given in dBsm (decibels relative to one square meter):

σdBsm = 10 · log10(σ / 1 m²)

Example: σ = 0.01 m² → σdBsm = 10 · log10(0.01) = -20 dBsm

Rayleigh, Mie and optical regions

An object's scattering behavior depends fundamentally on the ratio of its characteristic dimension a to the wavelength λ. Three regions are distinguished:

Rayleigh region

a << λ

The object is much smaller than the wavelength. RCS increases sharply with frequency (σ ∝ f4). Raindrops, insects at low frequencies.

Mie region (resonance)

a ≈ λ

Object size and wavelength are comparable. Strong RCS oscillations from constructive/destructive interference. Hard to predict.

Optical region

a >> λ

The object is much larger than the wavelength. For smooth surfaces, RCS is then nearly frequency-independent and can be described by geometric optics. For rough surfaces this no longer holds unconditionally. Aircraft, vehicles.

Practical example (77 GHz radar, λ ≈ 3.9 mm):

A DJI Mavic Mini (wingspan ~25 cm) has a/λ ≈ 64 → optical region.
A raindrop (d ≈ 2 mm) has a/λ ≈ 0.5 → Mie region.
A dust particle (d ≈ 50 μm) has a/λ ≈ 0.013 → Rayleigh region.

Limitation: the „optical region“ is not „simple“ at 77 GHz

The classification by a/λ only says something about the overall size of the target. What also matters for scattering behavior is how smooth the surface is relative to the wavelength. At λ ≈ 3.9 mm, screws, edges, cables, sensor domes and material seams of a drone are already on the order of the wavelength itself.

Such a target therefore behaves less like a single smooth reflector and more like a collection of many scattering centers, whose contributions add constructively or destructively depending on viewing angle. The result is a strongly fluctuating, aspect-dependent RCS — even where pure size considerations would suggest a calm, frequency-independent backscatter. This is one of the reasons a single average RCS value only describes the detectability of small flying objects to a limited extent.

RCS of simple geometric bodies

For a few idealized shapes, RCS can be calculated analytically. These serve as references and calibration standards.

Shape RCS formula Description
Sphere (radius a) σ = πa² Frequency-independent in the optical region. Isotropic backscatter.
Flat plate (area A) σ = 4πA² / λ² Only at normal incidence. Extremely direction-dependent, very high RCS at normal incidence.
Dihedral (edge length L) σ = 8πL4 / λ² Two perpendicular surfaces. Retroreflector in 2D. Double reflection.
Trihedral (edge length L) σ = 12πL4 / λ² Triple reflector (corner reflector), here in square form. For the common triangular trihedral, σ = 4πL4/(3λ²).
Cylinder (radius a, length L) σ = 2πaL² / λ At broadside illumination. Model for masts, pipes.

Why is the flat plate so important?

A 1×1 m metal plate at 77 GHz (λ = 3.9 mm) has an RCS of σ = 4π(1)² / (0.0039)² ≈ 826,000 m² (59 dBsm)! This value, however, only holds at the specular maximum: the main lobe of such a reflector is extremely narrow at about λ/D ≈ 0.22°, and outside it RCS drops by orders of magnitude. Smooth surfaces therefore produce very strong but tightly directed echoes — typical building clutter otherwise also arises from surface roughness, diffuse scattering and multiple reflections at edges.

RCS explorer: object & frequency

Choose an object and change frequency or size to see how RCS changes. The polar plot shows the RCS pattern over aspect angle (0°–360°).

The frequency changes this canonical scattering model.

Calculated RCS:

σ = 0.01 m²

= -20.0 dBsm

Region: optical (a/λ >> 1)

Fluctuating targets: Swerling I–IV

Real targets do not have a constant RCS. Motion, vibration and changing aspect angle make RCS fluctuate statistically. Peter Swerling defined four models in 1960 that describe different fluctuation characteristics.

Model PDF Fluctuation Typical target
Swerling I Rayleigh (exponential) Scan-to-scan (slow) Many equally sized scatterers, decorrelated between scans
Swerling II Rayleigh (exponential) Pulse-to-pulse (fast) Like I, but fast decorrelation (propeller, jet)
Swerling III Chi-squared (4 DOF) Scan-to-scan (slow) One dominant + many small scatterers
Swerling IV Chi-squared (4 DOF) Pulse-to-pulse (fast) Like III, but fast decorrelation

Swerling I / II (Rayleigh)

The RCS PDF is exponentially distributed:

p(σ) = (1/σavg) · exp(-σ/σavg)

The probability density is highest at small RCS values. Strong fading effects.

Swerling III / IV (chi-squared)

The PDF is a chi-squared distribution with 4 degrees of freedom:

p(σ) = (4σ/σavg²) · exp(-2σ/σavg)

A dominant scatterer stabilizes the RCS. Less extreme fading effects.

What does RCS depend on?

Wavelength / frequency

In the Rayleigh region: σ ∝ λ-4 (rises sharply with f).
In the optical region: σ ≈ const (frequency-independent).
In the Mie region: oscillating, hard to predict.

Polarization

Co-polarization (HH, VV): typical monostatic RCS.
Cross-polarization (HV, VH): usually significantly smaller.
The polarization matrix contains the complete scatterer information.

Aspect angle

RCS varies strongly with viewing angle. Frontal vs. side view can differ by 20+ dB. RCS diagrams are therefore always angle-dependent.

Material & geometry

Metal: near-perfect reflection (high RCS).
Carbon fiber (CFRP): electrically conductive and thus a good reflector; backscatter, however, depends strongly on laminate layup, fiber orientation and polarization. GFRP and unreinforced plastics, by contrast, are radar-transparent.
RAM (Radar Absorbing Material): absorbs EM energy, very low RCS (stealth).

RCS reference table

Rough order-of-magnitude reference values at 77 GHz, averaged over aspect angle. These are not measured data and are not suitable as a design basis: a target's instantaneous RCS fluctuates by 20 dB or more over aspect angle and time, which is why a single average value only describes detectability to a very limited extent.

Category Object RCS (m²) RCS (dBsm) Swerling model
Drones DJI Phantom 40.01-20II (propeller)
DJI Mavic Mini0.003-25II
DJI Inspire 20.03-15II
Fixed-wing UAV (1 m)0.05-13I
Racing quad (5")0.005-23II
Birds Pigeon0.001-30I
Gull0.005-23I
Eagle / buzzard0.02-17III
Swallow0.0003-35I
Goose / stork0.01-20I
Other Human (standing)1.00I
Human (walking)0.5–2.0-3 to +3II
Car10–10010–20I
Bicycle + rider2–53–7II

RCS in the radar equation

RCS is a central quantity in the radar equation, which links received power to system parameters.

Monostatic radar equation:

Pr = (Pt · G² · λ² · σ) / ((4π)³ · R4)

Pr = received power, Pt = transmit power, G = antenna gain,
λ = wavelength, σ = RCS, R = range to target

Maximum detection range:

Rmax = ((Pt · G² · λ² · σ) / ((4π)³ · Pr,min))1/4

Important: the R4 dependence means that doubling the range requires 16× the transmit power or 16× the RCS. This is one of the reasons small drones (σ ≈ 0.01 m²) are demanding targets. A concrete range does not follow from this alone, however: it only emerges from the full link budget with transmit power, antenna gain, noise figure, required Pd/Pfa and the integration gain of coherent processing, which the single-measurement form of the equation above does not include.

Scope and limitations

This module explains a method that is actually used in the Ozense processing chain. It describes the method and its limitations – not the measured performance of the system. The visualizations shown are simulated in the browser and are not a field performance proof.

How these methods work together at Ozense: Technology. What is actually measured today: Validation and development status.

Ozense · last reviewed on 13 August 2026