Contents
  1. Types of surveillance radar
    1. Primary Surveillance Radar (PSR)
    2. Secondary Surveillance Radar (SSR)
    3. Certified accuracy limits
  2. How PSR works — and where errors come from
    1. Range measurement
    2. Azimuth measurement
    3. The combined error ellipse
  3. Terma Scanter 4002
    1. Worked example: error at 55.66 km
  4. How SSR works — and its fundamental range limit
    1. The interrogation–reply cycle
    2. SSR azimuth accuracy
    3. Mode C altitude — Gillham code
Types of Surveillance Radar

The two radar systems used to allege infringements — and their certified accuracy limits

UK air traffic control relies on two fundamentally different radar technologies to detect aircraft. Both have well-documented accuracy limits set out in CAP 670 and the EUROCONTROL ATM Surveillance System Performance Specification. Understanding which type of radar was used — and what its certified error tolerances are — is essential to evaluating any infringement allegation.

Primary Surveillance Radar (PSR)

PSR detects aircraft by transmitting pulses of radio energy and listening for reflections. It requires no cooperation from the aircraft — no transponder, no electrical power on board. It measures range (distance) and azimuth (bearing), both with inherent errors that grow with distance. PSR provides no altitude information.

Secondary Surveillance Radar (SSR)

SSR works on a completely different principle. The ground station transmits an interrogation signal on 1030 MHz; the aircraft's transponder actively replies on 1090 MHz. This provides identification (Mode A squawk), altitude (Mode C), and in modern systems enhanced data (Mode S). SSR range measurement is subject to a hard accuracy floor of 150 m imposed by the ICAO-permitted transponder delay tolerance of ±0.5 µs if if the radar was perfect - which it isn't.(ICAO Annex 10, Vol. IV, §3.1.1.7.5.1).

Certified accuracy limits

CAP 670 SUR02.47–49 references the EUROCONTROL ATM Surveillance System Performance Specification (ESASSP) for the performance thresholds that UK ATC radar must meet:

Separation standardAverage error per flightSingle-plot worst caseReference
5 NM (en-route) ≤ 550 m 926 m (0.5 NM) ESASSP 5N_C-R4, R5
3 NM (terminal) ≤ 330 m 555 m (0.3 NM) ESASSP 3N_C-R4, R5

The average error figures are just that — averages. Individual plots regularly exceed them, and a three consecutive plots can be up to 926 m (en-route) or 555 m (terminal) from the aircraft's true position. In addition, the airspace boundary displayed on the controller's screen may itself be up to 450 m (0.25 NM) from its true surveyed position (CAP 670 SUR11.131). Radar error and map error can act in the same direction — producing a combined worst-case of approximately 1.4 km of permitted error, all within certified specification.

Full error budget

The following table brings together every documented error source from CAP 670 and the EUROCONTROL specification. Every value shown is within normal, compliant operation — these are not faults.

Error SourceMax Permitted ErrorReferenceNotes
Average position error per flight — 5 NM en-route ≤ 550 m ESASSP 5N_C-R4 Individual plots regularly exceed this average
Single-plot worst case — 5 NM en-route 926 m (0.5 NM) ESASSP 5N_C-R5 Up to three consecutive plots can be this far out and the radar is still within specification
Average position error per flight — 3 NM terminal ≤ 330 m ESASSP 3N_C-R4 Individual plots regularly exceed this average
Single-plot worst case — 3 NM terminal 555 m (0.3 NM) ESASSP 3N_C-R5 Up to three consecutive plots can be this far out and the radar is still within specification
Display map feature accuracy (en-route / approach) ≤ 450 m (0.25 NM) CAP 670 SUR11.131 Permitted error of displayed boundary vs surveyed position on the controller's screen
Display map acceptance threshold at commissioning ≤ 900 m (0.5 NM) CAP 670 SUR12.85 New-system flight trial acceptance criterion
PSR north alignment (recommended tolerance) ≤ 0.1° CAP 670 SUR04.9 Recommendation only — not an absolute requirement
Combined worst case (radar error + map error, same direction) ~1.4 km ESASSP + SUR11 combined Single-plot error (926 m) + map error (450 m) acting in the same direction
Key Point

An alleged infringement smaller than the radar's certified error is not proven by radar alone

If the alleged penetration into controlled airspace is 29 m, 50 m, or even 200 m, a single radar plot can be up to 926 m from the aircraft's true position (en-route) or 555 m (terminal) — many times larger than the allegation. The radar simply cannot distinguish between an aircraft just inside and just outside the boundary at these margins. CAP 1404's first evaluation question — "Can the ICG confirm an infringement actually occurred?" — should be answered "No" whenever the alleged penetration falls within the radar's error margin.

Primary Surveillance Radar

How Primary Radar (PSR) works — and where the errors come from

Primary Surveillance Radar detects aircraft by transmitting pulses of radio energy and listening for reflections. It requires no cooperation from the aircraft — no transponder, no electrical power on board. The radar measures two things: the range (distance) and the azimuth (bearing) of the target. Both measurements have inherent errors.

Range measurement

The radar transmits a pulse and times how long the echo takes to return. Since radio waves travel at the speed of light (c ≈ 299,792,458 m/s), the range is calculated as:

R = c × t / 2

where t is the round-trip time. The division by 2 accounts for the pulse travelling out and back. Errors in range measurement arise from:

Understanding Bias vs. Sigma

Two distinct types of error

Bias is the residual systematic error that remains after calibration — it cannot be removed. Think of a rifle sight that drifts slightly with temperature: every shot in a session lands in the same wrong place, but the offset changes between sessions. During any given flight, the bias is constant — so every radar plot of your aircraft is shifted in the same direction by the same unknown amount, up to the specified limit.

Sigma (σ) is the random scatter around the biased point. Even with perfect calibration (zero bias), individual measurements will scatter around the true position. At 1σ, approximately 68% of measurements fall within the stated tolerance. At 3σ (99.7% confidence), the scatter is three times larger. For safety-critical evidence, the 3σ bound is the appropriate measure — because the question is not where the aircraft probably was, but whether the radar can reliably prove it was inside the boundary.

Azimuth measurement

The radar antenna rotates continuously. The azimuth (bearing) of a target is determined by the antenna's pointing direction at the moment the echo is received. The antenna beam has a finite width — typically around 1° at the half-power points — so the system must estimate which part of the beam the target lies in. Errors in azimuth arise from:

Critically, azimuth error translates to a cross-range positional error that grows linearly with distance. A small angular error at long range produces a large positional displacement:

Cross-range error = R × sin(θ) ≈ R × θ (for small θ in radians)

At 55.66 km range, an azimuth sigma of 0.15° translates to a 1σ cross-range error of approximately 146 m. At 3σ (99.7% confidence), this becomes 437 m. This is the fundamental reason why radar positional accuracy degrades dramatically with distance.

The combined error ellipse

Because range and azimuth errors are independent and have different magnitudes, the true position uncertainty is not a circle — it is an ellipse, elongated in the cross-range (azimuth) direction. The semi-axes of the 3σ error ellipse are:

Along range:    a = 3 × σrange
Cross-range:  a = 3 × R × σazimuth

The total position error combines the bias offset with the 3σ noise ellipse. The conservative scalar bound is:

E(R) ≈ bpos(R) + 3 × σpos(R)

where bpos = √(br² + (R × bθ)²)   and   σpos = √(σr² + (R × σθ)²)
Real-World Example

Terma Scanter 4002: the radar Southampton Airport uses

The Terma Scanter 4002 is a Primary Surveillance Radar used by NATS at multiple UK sites. Following a Subject Access Request and Information Commissioner appeal, NATS confirmed this was the radar used to detect an alleged infringement. Its published specifications allow us to calculate the actual positional error at any given range.

ParameterSpecification
Frequency band9000–9200 MHz (X-band)
Transmitter6 kW, fault tolerant solid state (GaN), 8 modules
Instrumented rangeUp to 60 NM
Minimum detection range0.15 NM
Antenna18-foot, Cosec² elevation pattern, circular polarisation
Rotation rate12–20 RPM (default 15 RPM)
Accuracy — range< 25 m bias (< 60 m sigma)
Accuracy — azimuth< 0.1° bias (< 0.15° sigma)
Resolution — range< 36 m (measured at 10 dB SNR)
Resolution — azimuth< 1°
InterfaceEthernet UDP/TCP IP, ASTERIX format

Source: Terma Scanter 4002 product brochure

Worked example: error at 55.66 km (the infringement range)

In the victim's case, the aircraft was 55.66 km from the radar head. The alleged infringement was 29 m inside the Solent CTA boundary. Applying the Scanter 4002's published specifications:

Numerical Results at R = 55.66 km

The radar's error is 20 times larger than the alleged infringement

Bias offset (systematic error):

  • Range bias: 25 m
  • Cross-range bias: R × bθ = 55,660 × 0.1° × π/180 = 97 m
  • Combined bias magnitude: bpos = √(25² + 97²) = 100 m

3σ noise (99.7% confidence random scatter):

  • Range (3σ): 3 × 60 m = 180 m
  • Cross-range (3σ): 3 × R × σθ = 3 × 55,660 × 0.15° × π/180 = 437 m
  • Combined σpos = √(60² + (55,660 × 0.15° × π/180)²) = 158 m

Conservative bound (bias + 3σ):

  • E = 100 + 3 × 158 = ≈ 573 m

The alleged infringement was 29 m. The radar's conservative error bound at this range is 573 m — nearly 20 times larger. A Monte Carlo simulation confirms that the majority of radar returns for an aircraft 30 m outside the boundary would appear to be inside controlled airspace. The radar simply cannot distinguish between "30 m inside" and "30 m outside" at this distance.

What the CAA Claimed

"The radar is accurate to a few centimetres"

During the investigation of this case, the CAA official handling the infringement reportedly claimed the radar was accurate "to a few centimetres." The Scanter 4002's own manufacturer specifies range accuracy of 25 m bias with 60 m sigma — and that is range only, before azimuth error is added. The claim of centimetre-level accuracy is off by a factor of approximately 1,000 to 10,000. The CAA official was in possession of the radar make and model at the time of making this claim.

Secondary Surveillance Radar

How Secondary Radar (SSR) works — and its fundamental range limit

Secondary Surveillance Radar works on a fundamentally different principle to primary radar. Instead of detecting reflected energy, SSR relies on the aircraft's transponder to actively reply to an interrogation signal. This provides additional information — identification (Mode A) and altitude (Mode C/S) — but introduces its own error sources.

The SSR interrogation–reply cycle

The ground-based SSR interrogator transmits a coded pulse on 1030 MHz. The aircraft's transponder receives this interrogation, processes it, and after a fixed delay transmits a reply on 1090 MHz. The ground station measures the total elapsed time from interrogation to reply, subtracts the nominal transponder delay, and calculates range from the remainder.

The 3 µs Transponder Delay — A Hard Accuracy Limit

SSR range cannot be more accurate than 150 m — even with a perfect radar

ICAO Annex 10, Volume IV, Section 3.1.1.7.5.1 specifies that the aircraft transponder must reply to a valid interrogation with a delay of 3.0 µs ± 0.5 µs. This tolerance is a fundamental property of every Mode A/C/S transponder in service worldwide.

The ground station assumes a nominal delay of exactly 3.0 µs and subtracts it from the measured round-trip time. But if the transponder's actual delay is anywhere within the permitted ±0.5 µs tolerance, the calculated range will be wrong by up to:

  • Total timing uncertainty: 1.0 µs (from -0.5 to +0.5 µs)
  • Range uncertainty: c × 1.0 µs / 2 = 299,792,458 × 0.000001 / 2 ≈ 150 m

This is a hard floor on SSR range accuracy that cannot be overcome by any improvement to the ground radar. No matter how precise the interrogator's timing circuits, no matter how advanced the signal processing, the transponder delay uncertainty of ±0.5 µs translates to a 150 m range uncertainty window. A perfectly engineered, perfectly calibrated SSR ground station — with zero error of its own — would still be unable to determine an aircraft's range to better than 150 m.

ICAO Annex 10, Volume IV — Surveillance and Collision Avoidance Systems, Section 3.1.1.7.5.1

SSR azimuth accuracy

SSR determines azimuth by the same principle as PSR — the antenna's pointing direction when the reply is received. The same azimuth errors apply: bias from antenna misalignment and sigma from beam interpolation. At long range, the cross-range error from azimuth uncertainty dominates, just as with PSR. The combined position uncertainty from transponder delay tolerance plus azimuth error at range makes SSR no more accurate than PSR for lateral position.

Mode C altitude — the Gillham code error chain

Mode C altitude data passes through multiple stages, each adding error:

  1. Your altimeter — measures static pressure; subject to instrument error and QNH variation
  2. Your encoder / transponder — converts the altimeter reading to a Gillham code; ICAO Annex 10 Vol. IV permits a tolerance of ±125 ft (±38.1 m)
  3. SSR ground system — decodes the code and displays altitude in 100 ft increments; quantisation error up to ±100 ft
  4. QNH correction — applied by the display system; any error in the stored QNH value propagates directly into the displayed altitude for all aircraft
Error SourceMagnitudeReference
Gillham encoding tolerance ±125 ft (±38.1 m) ICAO Annex 10, Vol IV
Mode C display quantisation (100 ft steps) Up to ±100 ft SSR system design
Altimeter instrument error (certified aircraft) ±30–75 ft typical EASA Part 21 / airworthiness standards
QNH setting / atmospheric variation Variable METAR / AIP meteorological data
ADS-B ground domain altitude resolution (minimum) Must not degrade below 100 ft CAP 670 SUR07.38
Combined plausible error up to 300 ft All sources combined

The controller sees a derived figure, not the aircraft's actual altitude. A displayed value of FL055 (5,500 ft) could represent any actual altitude between approximately 5,200 ft and 5,800 ft under normal, compliant operation. CAP 670 SUR07.31 states that only barometric altitude shall be displayed to controllers for separation purposes — if GPS-derived geometric altitude has been used in an allegation, this is procedurally incorrect. [CAP 670 SUR07.29–31]

Key Point

Neither PSR nor SSR can prove a marginal infringement

For PSR, the error ellipse at typical ranges runs to hundreds of metres. For SSR, even a perfect ground station faces a 150 m hard limit from transponder delay tolerance alone. When the CAA alleges an infringement of 29 m, 50 m, or even 200 m, neither surveillance technology is capable of confirming the allegation to the required standard. The CAA's first question under CAP 1404 — "Can the ICG confirm an infringement actually occurred?" — should be answered "No" whenever the alleged penetration falls within the radar's error margin.