Astronomy · What light can tell us

An image and a spectrum answer different questions

Organise one small synthetic dataset by position and wavelength, then learn what is preserved or lost when measurements are combined.

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

Back to the chapter

After reading, you can

Identify the position and wavelength coordinates of a measurement product.
Compute a selected-bin image sum and retain the corresponding spatially extracted spectra.
Read spectrum axes and distinguish a feature from a unique physical identification.

Position and wavelength are different ways to organise a signal

An astronomical image assigns recorded signal to positions in a field. Its two spatial coordinates usually correspond to sky directions, possibly through a stated image projection. They do not automatically give physical distances within the source. Angular scale, orientation, detector response and processing remain relevant.

A spectrum assigns recorded signal to wavelength or frequency intervals. It is collected or extracted from a stated spatial region and exposure. Its horizontal axis asks where the signal lies in spectral coordinates; its vertical axis might be counts, flux density or a normalised quantity. Read those labels before comparing plots.

An image and a spectrum are therefore complementary products, not rival claims about which telescope is more scientific. Imaging spectroscopy can retain a spectrum at many spatial positions. A dataset can contain position, wavelength and time dimensions together, even when a particular display shows only two.

Keep the full data before making a display

Use an original synthetic dataset with two spatial cells, A and B, and three wavelength bins centred at 450 nm, 550 nm and 650 nm. Each bin is 10 nm wide. The table stores relative received signal integrated within each bin, using one common scale for all six values. These are invented numbers, not photon counts or physically calibrated fluxes.

Cell A has values 6, 2 and 6; cell B has 2, 6 and 2. Summing only these three selected bins gives 14 for A and 10 for B. An image-like sum map would therefore show a larger selected-bin total in cell A.

That sum map loses the way the total is distributed among the bins. At 550 nm, B’s value is 6 while A’s is 2. An extracted three-bin spectrum for each cell preserves that contrast. “A has more signal in the selected sum” and “B has more in the middle bin” can both be correct.

The cells are indices of two imaginary spatial regions. Their letter labels do not specify angular separation or physical size. Likewise, the three sparse bins do not constitute a complete continuous spectrum over the gaps or an all-wavelength image.

Original synthetic data: relative received signal integrated in equal 10 nm-wide bins, all on one common scale. The total includes only the three listed bins.
Spatial cell	450 nm bin	550 nm bin	650 nm bin	Sum of the three bins
A	6	2	6	14
B	2	6	2	10

One dataset, two different views

Scroll sideways to see the full diagram. Its explanation is below.

This original synthetic example has two spatial cells, A and B, and three equal-width 10 nm wavelength bins centered at 450, 550 and 650 nm. Cell A has relative received signals 6, 2 and 6 in those bins, so its three-bin sum is 14. Cell B has signals 2, 6 and 2, so its three-bin sum is 10. The sum map shows these two values at labelled cell positions; the boxes identify cells and have no physical sky scale. The spectrum view shows the same six signals against wavelength. Filled circles mark A and hollow diamonds mark B; faint vertical strips span the actual bin edges 445–455, 545–555 and 645–655 nm. The strips indicate bin width, not measurement uncertainty. No lines connect the unmeasured gaps. Both wavelength and signal axes start at zero; the signal axis runs to 8 in relative units per 10 nm bin. The map is the sum of only these three selected bins, not an all-wavelength image. Spatial and wavelength information can coexist in an imaging-spectroscopy dataset. The invented patterns identify no element, absorption line, celestial object or real instrument. Blank spectral gaps are unmeasured, not zero signal.

A spectral shape is evidence to interpret

A continuum is a broad spectral distribution. Emission features add signal over part of a spectrum; absorption features reduce the received signal relative to an appropriate surrounding or model continuum. Physical radiation processes and material along the path can create these patterns.

Laboratory measurements and physical models relate patterns of spectral features to atoms, ions, molecules and conditions. Real line strengths also depend on temperature, excitation, ionisation, radiative transfer, overlapping features and instrument response. Weak or absent visible features do not by themselves prove low or zero abundance of a chemical element.

The three-bin dip for cell A is deliberately not labelled as the fingerprint of a real species. With so little sampling, many distributions could fit it. The example teaches information organisation and comparison; it cannot identify a material, measure a velocity or establish a source temperature.

A justified identification seeks a consistent set of features under the relevant source and calibration conditions. Foreground gas, atmosphere and processing may also contribute. Keep the measured pattern and its physical explanation distinct, as in the chapter’s first lesson.

More plotted points are not automatically more detail

Spatial resolution concerns distinguishing nearby directions in the image; spectral resolution concerns distinguishing nearby wavelengths in a spectrum. Each depends on the instrument and observation. Adding file pixels or interpolated plotting points cannot recover information that the measurement blurred together or never sampled.

A spectrum extracted from a large aperture may combine several spatial structures. A narrow spatial selection can isolate one region but reduce the available signal. Changing the extraction region is therefore a change in the measurement being summarised, not merely a cosmetic plot adjustment.

Similarly, summing spectral bins can improve a particular total-signal estimate while discarding differences between bins. Whether that is useful depends on the question. Preserve the underlying data and document the sum or extraction so later readers know what was combined.

Read units, normalisation and colour mapping

A bin-integrated value is not the same quantity as a density per nanometre. A flux density describes how a quantity is distributed per unit spectral coordinate, whereas a bin value covers an interval. Changing from a wavelength axis to a frequency axis can require changing the vertical density definition as well; relabelling the horizontal axis alone is not generally sufficient.

For a first comparison, use like with like: the same units, wavelength convention, calibration, exposure treatment and normalisation. A continuum-normalised spectrum has been divided by an estimated continuum; its vertical scale no longer states absolute received flux. Keep that operation in the caption.

A colour image may combine selected bands using assigned display channels. Read which bands were used and which stretches or colour assignments were applied. A red patch on a screen can be a useful encoded comparison without being a direct unaided-eye colour or a unique temperature measurement.

The Shared Foundations detector/spectrometer and spectral-diagnostics records provide concise definitions. Here, the practice is to choose the view that answers the question while recognising which dimensions and calibration details the display leaves out.

Worked example

Find two valid rankings in the same data

Using the synthetic table, which cell has the greater three-bin sum, and which has the greater 550 nm-bin signal?

Sum A: 6 + 2 + 6 = 14. Sum B: 2 + 6 + 2 = 10.
Compare the middle-bin entries: A = 2 and B = 6.
The spatially selected spectral information explains why the rankings differ.

A has the larger selected-bin total; B has the larger 550 nm-bin signal. Neither statement identifies a real chemical species or total source luminosity.

Try the reasoning

If only A’s total 14 is retained, can its original 6, 2, 6 distribution be uniquely reconstructed?
Show the answer and reasoning
No.
Many sets of three numbers can sum to 14. The sum discards the spectral distribution unless the separate bins are saved.
A spectrum’s vertical axis says “continuum-normalised signal”. Does it directly state absolute flux in W/m²?
Show the answer and reasoning
No.
Normalisation changes the quantity being displayed. Absolute flux needs appropriate calibration and its own stated units.
Can one instrument or dataset retain an image and a spectrum at many positions?
Show the answer and reasoning
Yes.
Imaging spectroscopy combines spatial and spectral information. The display categories are not mutually exclusive instrument types.
A three-bin plot is smoothly interpolated into 300 display points. Has it acquired 300 independent wavelength measurements?
Show the answer and reasoning
No.
Interpolation changes the display, not the original sampling or instrumental resolution.

What to carry forward

Images organise spatial information; spectra organise wavelength information, and a dataset can retain both.
A selected-bin sum and a spectrum answer different questions about the same measurements.
Read units, normalisation, extraction and resolution before inferring a physical mechanism.

References

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