Strehl Ratio: Measuring How Sharp a Real Telescope Image Is

What the Strehl Ratio Actually Measures

A telescope collects a flat wavefront from a distant star and, if perfect, focuses it into an Airy pattern: a bright central disk surrounded by faint concentric rings. The peak brightness of that central disk is the theoretical maximum — no optical system of the same aperture can do better, because diffraction sets a hard limit.

The Strehl ratio S is defined as:

• S = (peak intensity of the real point spread function) / (peak intensity of the ideal, aberration-free PSF)

Both are measured for the same aperture and wavelength. If aberrations distort the incoming wavefront, the focused light no longer stacks coherently at one point; some of it scatters into the wings. The central peak drops, and S falls below 1. Because energy is conserved, a lower Strehl means more light leaks out of the diffraction core into a broad halo — the core stays the same width (still set by λ/D) but grows dimmer. That is why Strehl is really a measure of contrast and concentration, not resolution per se.

The Mechanism: From Wavefront Error to Strehl

The exact Strehl ratio is computed from the aberrated wavefront phase Φ over the pupil. It is the squared modulus of the pupil-averaged complex phase factor:

• S = | ⟨ exp(iΦ) ⟩ |², averaging exp(i·phase) over the whole aperture.

Perfect optics give Φ = 0 everywhere, so the average is 1 and S = 1. Any phase scatter pulls the average toward zero. Expanding this for small aberrations gives the workhorse Maréchal approximation (André Maréchal, 1947):

• S ≈ exp(−σ²), where σ is the root-mean-square wavefront error in radians.
• To first order, S ≈ 1 − σ².

Here σ = 2π·(RMS optical-path error)/λ. The relation is remarkably accurate for σ up to about 1–2 radians (S above ~0.1). Because uncorrelated error sources multiply, the total Strehl is the product of the Strehls from each contributor — atmosphere, deformable-mirror fitting, telescope optics, defocus — which is why AO error budgets add variances (σ²) in quadrature.

Key Numbers and a Worked Example

Consider the classic Rayleigh quarter-wave criterion: an optic with a peak-to-valley wavefront error of λ/4. For a smooth defocus-like aberration, that corresponds to an RMS error of about λ/14, or σ = 2π/14 ≈ 0.449 rad. Plugging in:

• S ≈ exp(−0.449²) = exp(−0.201) ≈ 0.80.

This is precisely why S = 0.80 is the universally quoted 'diffraction-limited' threshold — it is the Strehl equivalent of the historic λ/4 rule. Some worked benchmarks at λ = 550 nm:

• RMS error λ/20 (28 nm) → σ = 0.314 → S ≈ 0.91
• RMS error λ/14 (39 nm) → σ = 0.449 → S ≈ 0.80
• RMS error λ/8 (69 nm) → σ = 0.785 → S ≈ 0.54
• RMS error λ/4 (138 nm) → σ = 1.571 → S ≈ 0.08

The steep exponential dependence is the punchline: halving your wavefront error roughly quadruples the fraction of light in the core near the threshold, which is why the last few nanometers of polishing matter enormously for high-contrast imaging.

How Strehl Is Measured and Where It Appears

In practice, observers measure Strehl directly from an image of an unresolved star: fit or read off the PSF peak, compare it to a synthetic diffraction-limited PSF of the same aperture (accounting for the central obstruction and pixel sampling), and normalize by total flux. Wavefront sensors — Shack–Hartmann or pyramid sensors — provide an independent estimate by reconstructing σ and applying the Maréchal formula.

Strehl is the primary figure of merit for:

• Adaptive optics — Keck, VLT/SPHERE, Gemini/GPI, and the future ELTs report Strehl to grade correction, star by star.
• Space telescopes — Hubble after its 1993 COSTAR fix and JWST both operate near S ≈ 0.8–0.9 in the near-infrared.
• Coronagraphy and exoplanet imaging, where residual halo light from a low Strehl directly buries faint planets.
• Astrometry and photometry, since a stable, high Strehl means a sharp, repeatable PSF.

Because turbulence flickers, ground-based Strehl is a fluctuating quantity — quoted as a mean with substantial variance over seconds.

Strehl Versus Its Cousins: FWHM, Encircled Energy, and Seeing

Strehl is often confused with related metrics, but each captures something different:

• FWHM (full width at half maximum) measures the width of the PSF core — i.e. angular resolution. A partially corrected AO image can keep a diffraction-limited FWHM (sharp core) yet have a low Strehl (dim core, bright halo). Width and peak are decoupled.
• Encircled energy gives the fraction of light within a chosen radius — more relevant for aperture photometry than for peak contrast.
• Seeing (the Fried parameter r₀ and the ~0.5–1.5 arcsec seeing disk) describes the atmosphere before correction; Strehl describes the corrected image.

The link between them runs through r₀: an uncorrected large telescope has σ ∝ (D/r₀)^(5/6), driving S to near zero. Adaptive optics shrinks the residual σ, lifting S back up. Strehl uniquely combines resolution and contrast into one scalar, which is why it, rather than FWHM alone, is the headline number for high-performance imaging. It is also a coherent-optics analog of the modulation transfer function integrated over spatial frequencies.

Significance, Limits, and Famous Cases

The most famous Strehl story is Hubble's spherical aberration: a mirror ground to the wrong figure (about 2.2 microns of error at the edge) gave the launched telescope a Strehl of roughly 0.15 instead of the intended ~0.8 — light spread into a wide halo that crippled faint-object work until the corrective COSTAR/WFPC2 optics were installed in 1993, restoring S ≈ 0.85.

Where Strehl gets debated:

• Breakdown at low S: below S ≈ 0.1 the Maréchal formula overestimates; the exact integral or an 'extended Maréchal' form is needed. Some authors add a Gaussian-statistics correction, S ≈ exp(−σ²) for the coherent part plus a scattered pedestal.
• Measurement ambiguity: reported Strehls depend on pixel sampling, how the reference PSF is built, and noise — the paper title "Is that really your Strehl ratio?" captures a real reproducibility problem.
• Amplitude (scintillation) errors: a fuller expression multiplies a phase term by an amplitude term, S ≈ exp(−σ_φ²)·exp(−σ_χ²).

For the Extremely Large Telescopes (39 m ELT, TMT, GMT), pushing near-infrared Strehl above 0.7 across wide fields via multi-conjugate AO is a central design driver — and the ultimate test of how sharp a real telescope image can be made from the ground.

Strehl ratio across telescope regimes and the wavefront error each implies

System / regime	Typical Strehl ratio	RMS wavefront error	Image character
Perfect / space telescope (JWST, HST)	0.80–0.95+	< λ/14 (~0.07 waves)	Diffraction-limited, clean Airy core
Marginal 'diffraction-limited' cutoff	0.80	λ/14 RMS (~λ/4 P-V)	Rayleigh quarter-wave criterion met
Good NIR adaptive optics (K-band)	0.40–0.70	λ/8 to λ/6	Sharp core + faint AO halo
Excellent NIR AO, bright star	0.80–0.90	< λ/12	Near-space-quality on the ground
Visible-light AO (700 nm)	0.05–0.20	~λ/4 to λ/3	Partial correction, strong halo
Uncorrected atmospheric seeing	0.001–0.01	many waves	Seeing-limited blob (~1 arcsec)