Reading — Multi-Spectral Signatures

Reading — Multi-Spectral Signatures#

By the end of this lesson you should be able to:

  1. Place an aircraft’s signature across the visual, acoustic, and EO/IR bands, and say what each sensor actually sees.

  2. Fuse L22’s five IR components with the other bands into one multi-spectral picture.

  3. Apply the loudest-band rule — detection is a union across bands, so \(R_\text{det} = \max_b R_b\).

  4. Prioritize signature investment across the spectrum using the loudest-first budget from L26, and explain why this is the analytical core of Project 3.

Radar was never the only sensor#

L27 finished the RF story. Shaping bought the orders of magnitude, radar-absorbing material bought the stubborn last decibels, and emission control answered the one signature you choose to radiate. If radar were the only threat sensor, the B-21 would now be as quiet as physics allows. But the aircraft is still seen by the eye, still heard on the airflow, and — most stubbornly — still felt in the infrared. This lesson gathers every band that is left, then asks the question that actually decides survivability: when a defender can use all of them at once, which one gives you away?

The short answer, and the spine of the whole lesson, is that a modern integrated air-defense system (IADS) does not have to win with radar. It fuses radar-warning, IR-search-and-track (IRST), electro-optical (EO), and even acoustic cues, and it needs only one of them to cross a threshold. So the sensible way to think about a fused defender is not “how good is my radar cross-section,” but “which of my bands is loudest, and how close does that let them get.”

The visual band: the oldest sensor#

Before any receiver there is the human eye, and it still works. The unaided eye resolves an aircraft-sized target at a few kilometers in clear air — an angular-size limit, not a power one. Two amplifiers make it matter far past that:

  • Contrails. Engine water vapor freezes into a persistent ice-crystal trail in cold, humid air at altitude — a sky-wide arrow pointing straight at you, visible for tens of kilometers.

  • Glint. Specular flashes off the canopy, leading edges, and flat panels are brief but very bright.

The counters here are tactical and finish, not coatings. You avoid a contrail by choosing route and altitude, you kill glint with matte finishes, and you cut the contrast the eye lives on with low-reflectance paint. No radar-absorbing material on earth hides a condensation trail — a point that will come back hard when we talk about floors.

The acoustic band: loud down low#

Engines and airflow make noise, but sound is a slow, lossy messenger. It travels at roughly 340 m/s and fades fast, losing energy to geometric spreading and to atmospheric absorption that rises with frequency. For a high-altitude bomber the slant range to a ground sensor is long, so acoustic is a minor band — included for completeness and for the fusion argument, not because it threatens the B-21 at altitude. Down low it is a different story: low-altitude penetrators, helicopters, and small UAS (with their blade-passage tones) are often heard first, which is why acoustic gunfire and UAS-detection arrays exist at all.

Key Concept

The relevance of a band depends on the mission. Acoustic barely matters for a high-altitude penetrator but dominates for a helicopter on the deck. A signature argument is only meaningful against a stated altitude, aspect, and threat — there is no single “quietest” airframe.

The band that never left: EO/IR#

Everything from L22 is still glowing. The five IR signature components return unchanged as the persistent multi-band signature: three emitters — hot parts (~700–900 K, mid-wave), the plume (CO₂ near 4.3 µm, mid-wave), and aero-heated skin (long-wave) — and two reflectors — reflected solar (day) and reflected sky/earthshine. Mid-wave infrared (MWIR) carries the plume and hot parts tail-on; long-wave infrared (LWIR) carries the all-aspect skin.

The sensors that read these are the IRST and forward-looking infrared (FLIR) systems from L23, and the crucial property is that they read passively. This is the direct tie-back to L27’s emission control: passive IR needs no emission of its own to intercept, so there is nothing for your radar-warning receiver to hear and nothing your EMCON discipline can deny it. Thermal crossover (L22) hides the LWIR skin twice a day when it matches the background, but the MWIR plume does not care about the time of day.

Key Concept

EO/IR is the band that never leaves. Radar can be shaped and absorbed, and emitters can be silenced, but a hot engine glows on its own and a passive seeker collects that glow without radiating. Signature management does not get to “solve” IR the way it solves radar — it can only manage which IR component is loudest.

One aircraft, many truths — and the fusion rule#

Each sensor sees a different aircraft and crowns a different loudest part. Radar sees geometry, IR sees heat, the eye sees contrast, the ear hears engines. No single band is the whole story, and that is exactly why a fused defender is dangerous: beat one band and it simply leans on another.

Formalize that. If any band crosses its own detection threshold, the defender is cued — you are found. Detection is therefore a union over bands, and the effective detection range is not the average of the bands but the maximum:

\[ R_\text{det} = \max_b R_b . \]

The consequences are sharp. Reducing a band that is not the limiter changes \(R_\text{det}\) by nothing at all — the defender was already going to be cued by the loudest band, so your investment bought zero range. Cutting the loudest band does move \(R_\text{det}\) down, but only until it meets the next band, at which point \(R_\text{det}\) flattens because that next band has inherited the limit. Your worst band, not your best, decides how close you can fly.

Key Concept

\(R_\text{det} = \max_b R_b\) is the multi-spectral fusion rule: you are only as quiet as your loudest band. Money spent on a quiet band is wasted until that band becomes the limiter, and money spent past the point where the next band takes over is wasted weight, cost, and maintenance.

Four bands, four laws — the rule in numbers#

The fusion rule only bites once each band carries a range law, because different bands pay for signature reduction at very different rates. This lesson’s demo and its companion script L28_LoudestBand.m run one notional aircraft against four matched threats:

Band

Baseline signature

Baseline range

Range law

Radar

\(\sigma = 0.1\) m²

45 km

\(R = 45\,(\sigma/0.1)^{1/4}\)

IR

\(J = 500\) W/sr

30 km

\(R = 30\sqrt{J/500}\)

Visual

unit contrast \(C\)

12 km (floor 6 km)

\(R = \max(12\sqrt{C},\,6)\)

Acoustic

—

4 km

fixed (long slant range from altitude)

At baseline, then, \(R_\text{det} = \max(45, 30, 12, 4) = 45\) km and radar is the loudest band. The 6 km floor on the visual band is the demo’s: it has a contrast slider, and the floor is what stops paint and finish from buying the band away, because a contrail is not a coating problem. The companion script has no visual knob at all, so it simply stops at the 12 km baseline — which is why the two agree everywhere the script goes.

The laws matter because they convert decibels into range at different exchange rates. A reduction of \(d\) dB multiplies the signature by \(10^{-d/10}\), so a fourth-root band shrinks as \(10^{-d/40}\) while a square-root band shrinks as \(10^{-d/20}\). Radar therefore buys range slowly — L26’s fourth-root law again — and IR buys it twice as fast per decibel. Now run three cases:

  • Spend on the wrong band. Pour 10 dB into IR. Its range drops from 30 km to \(30 \times 10^{-10/20} = 9.5\) km — a real cut — and \(R_\text{det}\) stays at 45 km, exactly where it started. Radar was the limiter before and still is, so the investment bought no survivability at all.

  • Spend on the loudest band. Pour it into radar instead. \(R_\text{det}\) now falls along \(45 \times 10^{-d/40}\), but only until it meets IR at 30 km, which happens at \(d = 40\log_{10}(45/30) \approx 7\) dB. Past that the curve flattens: IR has inherited the limit, and further radar reduction changes \(R_\text{det}\) by nothing.

  • Spend on both. Only cutting radar and IR together drives \(R_\text{det}\) down to 12 km, where the visual band takes over. No amount of radar or IR treatment buys below that, because neither one touches what the eye sees.

That step-then-flatten shape is the whole lesson in one curve: each cut hands the lead to the next band, so \(R_\text{det}\) descends as a staircase rather than smoothly, and every flat stretch on it is money spent on a band that is no longer the limiter.

The budget, now across the spectrum#

This is the same discipline as L26, moved from within the RCS budget to across the whole spectrum. In L26 you swept the RCS contributors loudest-first; L27’s materials and discipline bought the last decibels of the loudest RF contributor. Now the “contributors” are entire bands. You cut the limiter, you expect the problem to move to the next band, you re-plan, and you repeat — and you stop when the marginal band would cost more than it saves. Some floors resist this entirely: the eyeball floor (a contrail or a glint) and, tail-on, the plume. Below those, no coating helps, because there is nothing to coat.

The same arithmetic generalizes straight into Project 3, where the “bands” are three threat systems rather than the demo’s four physical bands — a VHF surveillance radar, an X-band fire-control radar, and an IR MANPADS — so the rule reads \(R_\text{det} = \max(R_\text{VHF}, R_\text{X}, R_\text{IR})\). The laws are ones you already have: \(R \propto \sigma^{1/4}\) for each radar and \(R \propto \sqrt{J}\) for the seeker, which means the same decibel-to-range conversions carry over unchanged. Work a notional package through it. Take an aircraft the VHF radar sees at 120 km, the X-band radar at 60 km, and the MANPADS at 7 km, and buy a package that spends 16 dB on X-band RCS and 6 dB on stern intensity. Both of those are real cuts to real bands. The question the type-along below puts to you — and it is the whole trade — is whether either of them moved the range at which this aircraft is found, and if not, which decibel would have.

The trap this sets is worth naming, because it is the one real programs fall into: an aircraft can carry a world-class result in the band it worked hardest on and still be found at the range of a band nobody managed. Tail-on with an unsuppressed plume, the most heroic RCS program on earth buys no survivability at all — the next dollar belongs to whichever band is actually the limiter, not to whichever band the program is already good at. That is the question Project 3 puts to you in concrete numbers: against its threat mix, which band limits the B-21, and can the budget you are given move it?

Type-along

One aircraft faces three threats — a VHF surveillance radar at 120 km, an X-band fire-control radar at 60 km, and an IR MANPADS at 7 km. Two candidate packages spend decibels differently. Type this and see which one actually buys survivability:

% L28 -- you are only as quiet as your loudest band
R0   = [120 60 7];      % VHF, X-band, IR MANPADS baseline [km]
band = {'VHF','X-band','IR'};
den  = [40 40 20];      % R~sigma^(1/4) -> 40;  R~sqrt(J) -> 20
cutA = [0 16 6];        % package A: spend on X-band and IR
cutB = [8 16 6];        % package B: add 8 dB at VHF
RA = R0 .* 10.^(-cutA./den);
RB = R0 .* 10.^(-cutB./den);
[mA,kA] = max(RA);
[mB,kB] = max(RB);
fprintf('A  %.1f %.1f %.1f  -> %.1f km (%s)\n', RA, mA, band{kA});
fprintf('B  %.1f %.1f %.1f  -> %.1f km (%s)\n', RB, mB, band{kB});

Then answer these two:

  1. Package A spends 22 dB and buys two better numbers. Why does \(R_\text{det}\) not move?

  2. How many dB at VHF would hand the limit to X-band?

Solution

A  120.0 23.9 3.5  -> 120.0 km (VHF)
B  75.7 23.9 3.5  -> 75.7 km (VHF)

  1. Detection is a union, so \(R_\text{det}\) is the maximum over bands — and every one of those 22 dB went into a band that was never the limiter. X-band fell from 60 km to 23.9 km and the seeker from 7 km to 3.5 km, both real cuts, but VHF at 120 km was loudest before and is loudest after: two better numbers and zero survivability bought.

  2. Enough to drop VHF below X-band’s 23.9 km, which is \(d = 40\log_{10}(120/23.9) \approx 28\) dB. Package B’s 8 dB is the only spend in either package that moved \(R_\text{det}\) at all (120 km to 75.7 km), and VHF still limits. The handoff is computable — check whether you actually reached it before claiming credit for it.

Quick problem — Walk the Staircase

A notional penetrator against three threat sensors, and a program that buys its decibels in two steps.

  • Search radar: 80 km, \(R \propto \sigma^{1/4}\)

  • Fire-control radar: 34 km, \(R \propto \sigma^{1/4}\)

  • IRST: 50 km, \(R \propto \sqrt{J}\)

  • Step 1: RAM and edge treatment cut the RCS by 9 dB at both radars

  • Step 2: IR suppression cuts the intensity \(J\) by 3 dB

(a) \(R_\text{det}\) and the limiting band at baseline.
(b) After step 1: \(R_\text{det}\) and the limiting band.
(c) After step 2 as well: \(R_\text{det}\) and the limiting band.

Solution

(a) \(R_\text{det} = \max(80, 34, 50) =\) 80 km, and the search radar limits.

(b) A fourth-root band scales by \(10^{-9/40} = 0.596\), so the search radar falls to \(80 \times 0.596 = 47.7\) km and the fire-control radar to \(20.3\) km. The IRST is untouched at 50 km, so \(R_\text{det} =\) 50 km and the IRST now limits. The 9 dB paid for 32.3 km at the search radar but bought 30 km of \(R_\text{det}\).

(c) A square-root band scales by \(10^{-3/20} = 0.708\) — 3 dB is a factor of two in \(J\), so range falls by \(\sqrt{2}\) — and the IRST drops to \(50 \times 0.708 = 35.4\) km. Now \(R_\text{det} = \max(47.7, 20.3, 35.4) =\) 47.7 km, and the search radar limits again. The 3 dB paid for 14.6 km at the IRST but bought only 2.3 km.

So what: every cut hands the limit to the next band, so \(R_\text{det}\) walks down a staircase, not a slope — re-rank the bands after each step, before you spend the next decibel.

Wrap-Up#

Visual (contrails, glint, paint) and acoustic (mostly a low-altitude problem) round out the non-RF bands, but EO/IR is the persistent one: L22’s five components, read passively by IRST and FLIR with nothing for EMCON to deny. Fuse them and detection is a union, so \(R_\text{det} = \max_b R_b\) — you are only as quiet as your loudest band. Signature management is therefore a balanced, loudest-first budget across the whole spectrum, not one heroic band, and over-investing past the next limiter is wasted weight and money. That rule is not just a slogan: it is the analytical core of the Block 3 signature trade study. Next, L29 is the Project 3 work day — bring the loudest-band rule, because it is the project’s central argument.

Contents