Astronomy · Seeing faint and small details

Atmospheric seeing and adaptive optics

Explain how atmospheric wavefront disturbances broaden a ground-based image, then distinguish adaptive correction from transmission and background control.

Continue in the approved chapter order. Necessary vocabulary, quantities and assumptions are explained in the lesson.

Back to the chapter

After reading, you can

Describe atmospheric seeing as an image-quality effect.
Explain the measurement-and-correction loop of adaptive optics and its limits.
Compare source extraction under a declared broader-image model without treating it as a universal atmosphere.

A point source can arrive through a changing optical path

Air with spatially and temporally varying refractive properties changes the optical path across an incoming wavefront. Different parts of a telescope aperture then receive a disturbed version of the ideal wave. The resulting image can move and change shape; a longer exposure can average those changes into a broadened pattern.

Astronomical seeing describes image degradation associated with atmospheric turbulence. It depends on conditions, viewing direction, wavelength and the measurement convention. A seeing value needs that context rather than being treated as one permanent property of an observatory.

Seeing, transparency and sky brightness concern different effects. A night can transmit a useful fraction of the source light while delivering blurred images. A sharp image can still sit on a bright background. The following lesson examines transmission and added sky signal separately.

Diffraction remains part of the optical response. Atmospheric disturbance adds another stage to the delivered image; a clear sky does not remove the finite pupil’s wave-optical behaviour.

Adaptive optics acts on a measured disturbance

Adaptive optics uses information about an incoming wavefront to command an optical correction, commonly through a deformable mirror. A feedback system repeatedly measures and updates the correction as atmospheric conditions change.

The measurement can use a suitable reference source or an artificial guide beacon, depending on the design. A reference direction samples a path through the atmosphere. A target in another direction need not experience exactly the same disturbance, so a correction can vary in quality across the field.

Finite update speed leaves errors when the disturbance changes faster than the correction loop can follow. Finite actuator spacing limits the spatial detail a correcting surface can reproduce. Measurement noise and the geometry of atmospheric layers and guide paths also matter.

Adaptive correction therefore has a specified wavelength, field and performance context. It is not a promise of perfect images everywhere on the sky. It also does not restore radiation absorbed by the atmosphere or remove every sky and instrument background contribution.

Observing above most of Earth’s atmosphere changes this turbulence problem, while diffraction, instrument response and detector sampling still require treatment. The complete observing system determines the final useful image.

A broader image can require a larger extraction region

Use a disclosed synthetic model: circular two-dimensional Gaussian source images with FWHM 1 or 2 arcseconds. FWHM is the full width through the centre at half the peak intensity. A Gaussian is a teaching approximation here, not the exact atmospheric or diffraction response.

Choose a circular extraction radius equal to each image’s FWHM. For this particular model, both regions enclose the same source fraction 15/16. Hold total source strength fixed and choose its scale so that each extraction contains 100 expected source electrons.

Doubling the extraction radius from 1 to 2 arcseconds multiplies its angular area by four. At a fixed uniform sky contribution per angular area, an expected 900 background electrons in the smaller extraction become 3600 in the larger one.

Under the independent source-plus-background Poisson model with a known background mean and other noise neglected, SNR = S/√(S + B). The two cases give about 3.16 and 1.64. Keeping the same captured source fraction did not keep the background uncertainty the same.

A fixed small aperture would instead capture a different source fraction as the image broadened. Real extraction methods can fit a response model or choose another region. The appropriate comparison includes those choices and their calibration, rather than assuming every broader image has this exact numerical cost.

Original Gaussian-image comparison. Radius equals FWHM in each case; this gives the same 15/16 source fraction. Background is uniform per angular area.
Image FWHM	Extraction radius	Expected source	Expected sky	Restricted model SNR
1 arcsecond	1 arcsecond	100 electrons	900 electrons	3.16
2 arcseconds	2 arcseconds	100 electrons	3600 electrons	1.64

Image quality belongs with the observation

An observing record should identify the image-quality measure, wavelength range, exposure and relevant conditions. Record whether an adaptive system was used and how its response was characterised. A nominal diffraction scale alone does not describe every delivered image.

A correction can improve separation or extraction performance while leaving background or calibration as the limiting factor. The useful question is which change supports the measurement, with the residual response and uncertainty retained.

The next lesson follows source attenuation and sky glow into the data record. Together these effects explain why “good conditions” needs a description of more than whether clouds are visible.

Worked examples

Follow the extraction geometry

In the synthetic Gaussian comparison, why does the sky contribution rise from 900 to 3600 electrons when extraction radius doubles?

The circular angular area is proportional to radius squared.
Doubling the radius multiplies area by four.
At the same uniform background and exposure, expected sky electrons also multiply by four.

The extraction covers four times the angular area. The source fraction was separately held equal by the declared Gaussian construction.

Calculate the two conditional ratios

With S =100 in both cases, compare B =900 and B =3600 under S/√(S + B).

For the smaller extraction,100/√1000≈3.16.
For the larger extraction,100/√3700≈1.64.
These values apply to the stipulated independent-noise model.

The broader-image extraction has a lower model SNR. It is not a universal ratio for every atmosphere or extraction method.

Try the reasoning

Does good transmission alone establish sharp ground-based images?
Show the answer and reasoning
No.
Atmospheric turbulence can degrade image quality while substantial source light still passes through.
Can a finite-speed correction loop perfectly follow arbitrarily fast wavefront changes?
Show the answer and reasoning
No.
Update delay and measurement limits leave residual error.
Must a correction measured in one guide direction be equally accurate in every other direction?
Show the answer and reasoning
No.
Different directions can sample different atmospheric paths and disturbances.
Does adaptive optics automatically remove sky background or restore absorbed source photons?
Show the answer and reasoning
No.
Wavefront correction, transmission and background are different parts of the observation.

What to carry forward

Atmospheric seeing changes the delivered image through a disturbed wavefront.
Adaptive optics is a bounded measurement-and-correction process with time, spatial and field limits.
Image quality can change extraction and background costs even when total source strength is unchanged.

References

Teaching resources and research supporting this lesson’s explanations and examples.

European Southern Observatory3 sources