Astronomy · Astronomy across the spectrum

Combining views at different wavelengths

Compare matched multiwavelength measurements through their calibration, spatial resolution and observing time, using an explicit example where equal totals hide different structures.

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 metadata needed to compare measurements across bands.
Explain how equal unresolved totals can hide different spatial contributions.
Use assigned colours and historical multiwavelength examples without inferring a unique source property from a display.

An object can have several wavelength-dependent views

Different wavelength ranges can reveal different emitting components and propagation effects in the same astronomical field. A useful multiwavelength comparison brings those measurements into a shared question: which structures are present, which vary, and which physical models can account for the combined evidence?

First establish what each dataset measures. A calibrated flux integrated across a band differs from a spectral density per frequency, a raw count rate or a displayed image brightness. A comparison can require the band limits, unit, calibration, exposure and sensitivity of each observation.

Next establish where and when the observations apply. Coordinate registration aligns the same sky locations; field of view sets the area available to compare. Observing epochs matter when an object changes. A carefully aligned image can still describe a different time from another image.

Spatial resolution adds a separate constraint. One instrument may separate features that another blends together. Resampling a coarse image onto smaller pixels changes its sampling grid but does not recover spatial information absent from the original measurement. Registration and resolving power solve different problems.

Equal field totals do not fix the spatial contributions

The figure uses original invented data for one matched field and observing epoch. It contains two labelled cells, A and B, and two disjoint observing bands: band 1 from 450 to 460 nm and band 2 from 2.0 to 2.1 µm. No physical angular scale or real astronomical object is assigned.

All values are calibrated fluxes integrated within the stated band, in a common unit of 10⁻¹⁵ W/m². They are not raw detector counts or spectral-density samples per hertz. The band widths differ, so equal integrated values would not imply equal spectral densities or photon arrival rates.

In band 1, A contributes 9 units and B contributes 1. In band 2, A contributes 1 and B contributes 9. Each field total is 10 units, yet the stronger cell differs between bands. An instrument that leaves the two cells unresolved reports the same total in both bands without providing that allocation.

The equality of the two unresolved totals therefore does not identify which component emitted which share. Across the two selected disjoint bands, each cell also sums to 10 units; that selected sum is still not an all-wavelength source luminosity. The figure illustrates lost allocation information, not a measured blur model or a temperature determination.

Original synthetic matched-field data. Values are band-integrated fluxes in units of 10⁻¹⁵ W/m²; totals alone do not allocate flux to cells.
Spatial selection	Band 1: 450–460 nm	Band 2: 2.0–2.1 µm
Cell A	9	1
Cell B	1	9
Unresolved A+B field	10	10

Two bands, one field

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

Original synthetic cell diagram for two disjoint observing bands. Band 1 integrates calibrated flux from 450 to 460 nanometres; band 2 integrates from 2.0 to 2.1 micrometres. All cell values and totals use the same unit, 1e-15 watts per square metre. The field and observing epoch are the same. Cell A is on the left and cell B on the right in both panels; no physical sky scale is shown. In band 1, A has flux 9 and B has flux 1, so the field total is 10. In band 2, A has flux 1 and B has flux 9, also totaling 10. Solid borders identify A and dashed borders identify B, in addition to their printed labels. A separate panel represents an instrument that leaves A and B unresolved: it reports total flux 10 in each band, without assigning those totals to either cell. Equal unresolved totals do not determine the different spatial contributions: A is stronger in band 1 and B in band 2. Values are band-integrated fluxes, not detector counts or samples of density per hertz. Band widths differ, so these equal integrated totals do not imply equal spectral density or equal photon arrival rates. The cells are invented data containers, not real astronomical images or a measured point-spread-function model. No object, material, temperature or composition is identified.

Let the measurement’s resolution limit the attribution

When a fine-resolution observation separates two structures and a coarse one blends them, assigning the coarse-band emission to one structure requires additional information or an explicit model. Merely overlaying the images does not supply the missing spatial separation.

A comparison can instead state what is supported at a common angular scale, or use a documented response model to test possible contributions. In either case, retain the resolution limitation. Larger pixels drawn on a page are not necessarily a larger telescope beam, and an image’s visual sharpness is not a complete calibration.

Different sensitivity limits also affect apparent differences. A blank patch may be a non-detection with an upper limit rather than zero flux. Compare uncertainty and selection criteria before declaring that a structure exists only in one band.

If the observations were taken at different times, variability can mimic a wavelength difference. A simultaneous or coordinated campaign reduces some ambiguity, but “coordinated” does not mean every datum was recorded at the same instant. Read the observing-time metadata at the level required by the scientific question.

Read a real campaign with its date and purpose

Chandra’s 2021 M87 release discusses observations associated with a coordinated 2017 campaign using ground and space facilities across the spectrum. It provides a real example of complementary wavelength coverage and differing angular resolution. This lesson uses it qualitatively, without adopting an exact facility count or a current observational record claim.

The campaign’s images and spectra are measurements from different instruments, not interchangeable photographs. Their observing times, angular scales and physical interpretations need the accompanying documentation. The existence of a multiwavelength display does not by itself establish that every feature is spatially resolved or that one mechanism explains all bands.

This example supports the observing strategy: combine evidence that constrains different parts of a source model. The invented two-cell diagram above remains a separate teaching case, with no claim that its numbers reproduce M87 or any real observing campaign.

Assigned colours help locate differences, then need a legend

Invisible-band measurements are often mapped into visible display channels. The legend identifies which band contributes to each colour; scaling and contrast settings affect how strongly each component appears. ESA’s filter-fusion explanation illustrates why those mappings are part of the presentation.

A red-coded region is not automatically a cool object, a particular chemical substance or the same physical layer as an overlapping blue-coded region. Those interpretations need calibrated measurements and a source model. A display can suggest a question without uniquely answering it.

The next original chapter examines spectra more closely. Spectral features can add information about temperature, material and motion, but they retain the same obligations to wavelength calibration, model choice and uncertainty. Combining views is most useful when the evidence from each view remains identifiable.

Worked examples

Recover what a total can actually say

An unresolved instrument reports 10 flux units in each of the two synthetic bands. Can it choose between the stated allocation A = 9,1 and B = 1,9, and an allocation of 5,5 to each cell in each band?

For the stated allocation, each band sums to 9 + 1 = 10.
For equal cell contributions, each band sums to 5 + 5 = 10.
Both allocations satisfy the same two totals; the unresolved measurement supplies no unique cell assignment.

No. The totals constrain the sums but do not select the spatial allocation. Additional resolved data or a tested model is required.

Keep sampling separate from information

A coarse-band image is resampled onto the fine image’s grid. It now has the same number of pixels. Has it gained the fine image’s ability to separate A and B?

Resampling constructs values on a new grid from the available measurement.
It does not add independent information that the original instrument failed to resolve.
Any proposed separation must be supported by additional data or a stated inference model.

No. Matching grids can help alignment and display, but it does not by itself create missing resolving power.

Try the reasoning

Are the synthetic totals of 10 values per hertz, photon rates, or integrated fluxes within the stated bands?
Show the answer and reasoning
Integrated fluxes within the stated bands.
The common unit is 10⁻¹⁵ W/m². The band limits and density distinction are part of the example.
Why check observing dates before assigning a brightness difference solely to wavelength?
Show the answer and reasoning
The object may have varied between observations.
An epoch difference can imitate or alter a spectral comparison.
Should an undetected patch in one band automatically be entered as exact zero flux?
Show the answer and reasoning
No.
It may provide an upper limit depending on the sensitivity and detection criterion.
Does a red display channel uniquely identify temperature or composition?
Show the answer and reasoning
No.
A visible colour may encode a selected invisible band and display scaling. Physical interpretation needs more evidence.

What to carry forward

Compare units, bands, coordinates, resolution, sensitivity and epoch before interpreting a multiwavelength difference.
Equal unresolved totals can hide different spatial contributions.
Assigned colours organise measurements; calibrated evidence and physical models establish their meaning.

References

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