ScienceProblem: SpectralEnergyDistributions

PrimaryActor:

Research astronomer

---

ScienceGoal:

To combine all existing photometric (and spectroscopic?) data on a given object (and, by extension, list of objects, e.g. resulting from a database query) into a spectral energy distribution (SED).

---

DataSets:

Could be a wide range of photometric datasets: radio, infrared, optical, UV, X-ray, etc.

---

ProblemDescription:

One of the principal means of classifying an astronomical object, and of understanding its physical properties, is through its spectral energy distribution (SED), which is defined to be a distribution of energy by electromagnetic radiation expressed as a function of wavelength (or frequency).

For example, when estimating a photometric redshift for an object (as in the ScienceProblem GetGalaxyRedshift), one would seek to match observed photometric data with a model SED redshifted by various amounts, and, similarly, in MultiWavelengthAssociations, when seeking to associate a source detected in one band with one of a number of possible counterparts detected in another band, one factor that is considered is whether the flux ratio between the two bands implied by that putative association seems reasonable, given the range of SEDs of populations of objects likely to be detected in the two datasets.

There are a number of practical aspects to this problem. There is an interoperability issue, given the vast number of names that can be used to denote the same sort of magnitude in different catalogues. There is also a data characterisation and consistency issue, relating to how clear it is what sort of magnitude (total - how estimated?; aperture - what size?; isophotal - what isophote?) is listed in any given catalogue...and the same goes for data in other passbands. There is also a problem about how best to present data from both spectrophotometry and broad band photometry, given that the former will have a much higher spectral resolution than the latter.

More fundamentally, there is the question of how to associate entries in different databases that refer to the same astronomical object - see MultiWavelengthAssociations for a discussion of this point.

---

CurrentSolution:

For data that one does not have to hand, probably the best current solution is that provided by NED, which will plot what photometric data it knows of for a given object (specified by name or position). The NED system now offers an option for zooming in on part of the SED plot, which is useful for the NIR/optical/UV region, where the photometric data points are closely spaced in wavelength, but they do not include real spectrophotometric data.

This zooming feature only serves to emphasise the problem with combining different types of magnitude: if one looks at the NED SED plot of M31, for example, and zoom in on the optical region, the photometric datapoints are seen to span more than four orders of magnitude, clearly indicating that NED is plotting inconsistent data. NED does provide a note to accompany each photometric datapoint plotted, though, and, in addition to the literature reference for the data, the specification of the band (e.g. V(Johnson)) etc, there is an entry for Spatial mode, which is supposed to indicate what sort of magnitude was computed. In some cases this is helpful (e.g. Flux in fixed aperture, with the aperture size quotes in the Qualifiers field), but in others it is not (e.g. Multiple methods with a Qualifiers entry of Homogeneized from new and previously published data).

---

VOSolution:

It's not immediately clear what the VO solution to this ScienceProblem should be. It is clear that we'd want the VO to do more than just collate photometry data, as NED does now, and that the metadata provided with VO data should be rich enough that one can try to construct a more meaningful SED (e.g. using consistent aperture magnitudes), but it's not clear how to do this in detail: for example, how can the VO know what aperture corrections to apply?

One thing that the VO solution probably should contain is the ability to obtain upper limits from imaging data in bands in which the source is not detected. Again, there are complications about how to do that in practice, but that would seem to be a highly desirable feature.

The VO solution should also be able to integrate spectroscopic and broad-band photometric data, so that the user has access to as much information as exists in any given wavelength region.

Another desirable feature would be an easy way to manipulate the observational SED data points and model SED fits. For example, it might be nice to be able to perform an interactive version of the photometric redshift estimation portion of ScienceProblem GetGalaxyRedshift, and overplot the observational data with SED models that could be redshifted using some GUI. In a similar vein, it would be good to be able to do an interactive version of ScienceProblem GalaxySpectralAnalysis, and take higher-resolution spectroscopic data and overplot that with model spectra with different metallicities, or whatever.

---

Discussion:

Clearly need for more thought here, on what the scientific requirements really are, and prioritise development work.

---

KeyReferences:

---

---

GoodStyle: Please add comments below. This area should be used for refinement of the above document. If you want to ask questions or start a dialogue with the author, please use (or create) a topic in the Science Problems Forum. For other ScienceProblems, refer to the ScienceProblemList.
Author: Once the refinements here and comments in the forum die down, perhaps you could rewrite the problem, incorporating the comments and refinements.

---

-- BobMann - 12 Feb 2002