WISE Extended Sources
WISE Extended Sources

The Bottom Line

Extended sources, in the form of fuzzy galaxies and Galactic nebulosity, will be present in every WISE image frame, thus complicating the primary mission goal of detecting, characterizing and cataloging point sources. The basic data reduction must include some level of extended source characterization to properly handle point source photometry and quality assessment of the resulting catalogue. A secondary issue is the scientific potential of the extended sources themselves. WISE will resolve hundreds of thousands of galaxies and its images will contain a rich assortment of Galactic emission, both high surface brightness and diffuse varieties. A relatively small/modest effort to incorporate extended sources into the data reduction pipeline would greatly enrichen the WISE source catalogue.

The main issues to consider:

My own sense is that we should dedicate at least one FTE/yr throughout the mission to deal with extended sources. That is a bare minimum (but not unlike 2MASS!). A more optimal approach would be to have a scientist and data analyst assigned to extended sources -- from 1.5 to 2 FTE/yr. This document outlines the impact that extended sources have on the WISE mission. Read on ...

I. Introduction

The WISE mission will map the whole sky, capturing every planet, asteroid, star, nebula and galaxy that falls within its sensitivity limits. However, due to the limited resources that have been mostly dedicated to hardware development/construction, the extended source component of the mission is, for the most part, undeveloped and unrealized. Resolved galaxies, nebulae and other fuzzy objects will not be measured or catalogued as part of the primary mission data products. I will argue in this document that the WISE P.I. and science team should consider placing basic extended source processing within the lien list of tasks that we desire to carry out. Even with a modest set of resources, the resulting large return in data quality (and science potential) should justify the expense.


II. Definition of Extended Sources

There are three basic kinds of resolved or extended sources: discrete, diffuse and a combination of the two. Examples include:

There is some overlap between these rough categories. PNs may be more diffuse than discrete; some compact HII regions may be more discrete than diffuse; some galaxies are so diffuse that they are like whispy clouds in space. Zodical emission is so diffuse that it represents a smooth (but not necessarily uniform) background 'light' that will be in every WISE image. Observations that cross the plane of the Milky Way will potentially have every kind of extended source, thus presenting a major challenge for data reduction and source characterization.


III. Example Infrared Images of Extended Sources


Tadpole Galaxy (UGC10214) as seen from optical to infrared.

Sombrero Galaxy (M104) as seen w/ IRAC (3 to 8 um).

The Galaxy-Rich Coma Cluster as seen w/ IRAC (3 to 8um).

Resolved galaxies in a SWIRE 3.6um image of the Lockman Hole

Supernova Remnant Cass A as seen with HST, Chandra and Spitzer

Ring Nebula (M57) as seen with IRAC (3-8 um)

LMC as seen with IRAC (3-8 um) and MIPS 24um

Tarantula Nebula in the LMC as seen with IRAC (3-8 um)

Double Helix Nebula as seen with MIPS (24 um)

Embedded Stellar Cluster as seen with IRAC (3-8 um)

GLIMPSE 3-8um view of the Milky Way (glon=284, glat = -1)

MIPSGAL 24um view of the Milky Way (glon = 318, glat = -0.5)


IV. Fuzzies Complicate Point Source Extraction

Point sources are sometimes superimposed on extended sources and their flux measurement can be biased by the surrounding/nearby extended emission. Conversely, point-like sources can be spuriously detected on large, extended objects. The figure to the right shows the nearby galaxies M83 & M51 with 2MASS point source detections overlayed. Many of these sources are pieces of the galaxy (e.g., SF/HII regions) or noise bumps that are enhanced by the underlying galaxy light. Without even minimal characterization of the extended source, we have no way of reliably flagging "point" sources that might be contaminated or modified by the underlying emission. At the very least we will need to flag sources that are in close proximity to 2MASS galaxies.


V. Angular Resolution of WISE

Since the WISE PSF beam is 6 - 12 arcsec in the 3 to 24um window that it observes, the characteristic resolving element thus ranges between ~20 to 40 arcsecs (assuming > 3*beam ). Based on the 2MASS Extended Source Catalog (XSC), we can expect tens of thousands of discrete sources (galaxies) to be resolved with WISE, at least at the short wavelengths. Below is a table of 2MASS XSC sources that have sizes greater than the WISE resolving element. A plot showing the XSC size distribution is given here.

At the WISE short wavelengths, some ~200,000 galaxies will be resolved, thus requiring a more careful photometric extraction -- note that a straight PSF-fitting or aperture-correction measurement of a resolved source will result in a systematic underestimate of the true flux. At longer wavelengths, some 25,000 sources may be resolved by WISE, although this estimate is more uncertain since the K-band size metric is sensitive to stars whereas the 12/24um is sensitive to the ISM (which has a different spatial distribution from the evolved stellar population). If anything, WISE will resolve more sources than this K-band estimate because of the disk-extended nature of the hot dust component (e.g., see the image of the Sombrero Galaxy, above). Resolved sources must be treated more carefully to avoid systematic biases in the source catalog fluxes.

In the plane of the Milky Way, complex sources (e.g., HII regions, YSOs) will dominate the population of extended sources. Diffuse emission arising from ISM dust-heating by hot stars will be everywhere, creating a very difficult background to measure and characterize (e.g., see the MIPSGAL image of a typical field in the Milky Way). How do we flux calibrate this diffuse emission? Image artifact mitigation will be an important exercise to create scientifically valid images that can be used to study the diffuse and extended components of WISE observations that cross the Plane.


VI. Basic Measurements of Discrete Sources

At the very least, the data reduction pipeline must have a way to measure the flux of discrete extended sources. Circular apertures that capture a significant fraction of the source light is the basic measurement. Consistent measurements from band-to-band ensure reliable color metrics. A corresponding size metric is another basic measurement. The size of the source determines how large the aperture measurements should be. To these ends, the local background must be measured to a high accuracy in order to remove residual light from the source fluxes.
To summarize the basic measurements that are needed for extended sources:

  • detection
  • local background subtraction
  • Photometry (circular apertures)
  • Extent/Size metric
  • Quality status

  • NGC6703 with IRAC 3.6um. A set of nested circular apertures are centered on the nucleus. The local background is measured with an circular annulus, just outside the faintest isophotes.

    NGC6703 integrated flux curve of growth for the four IRAC bands.

    These measurements are relaively straight-forward, requiring only a minimal effort to develop and apply to the data reduction pipeline. Complications arise when the source is contaminated by foreground stars or nebulosity that is within the effective region (bounded by the local background measurements). To deal with non-trivial conditions, the next step in development complexity is required.


    VII. Advanced Measurements of Discrete Sources

    Reliable fluxes, surface brightnesses and sizes require background, artifact and contaminating source subtraction to allow clean characterization and extraction. Total fluxes, in particular, require very careful image cleaning and subsequent characterization. The usual procedure is to identify nearby stars and remove them by either masking them or subtracting their flux (using a PSF or some other simple model). The object source is modeled (e.g., using an ellipsoid) and stars are removed interatively; see the illustration below.


    Left panel: disk galaxy as seen in the K-band; several foreground stars contaminate the disk. Middle panel: galaxy is model and subtracted from image, leaving stars. Right panel: stars are subtracted, and the cleaned galaxy remains.

    With images cleaned of artifacts and stars, the background may be measured and the source characterized using circular and elliptical apertures. The procedure follows roughly:

    These procedures are iterative and CPU-intensive, not to mention a challenge to develop into a robust pipeline. For example, the 2MASS Extended Source Processor (Jarrett et al 2000) development was years in the making, including during actual mission operations. For WISE we do not have the luxury of polishing and honing procedures. On the other hand we can leverage much of this knowledge and coding to build something that will operate on extended sources in a relatively short span of development. It may not be possible to deblend galaxies, but it should be possible to carry out basic source subtraction and elliptical aperture photometry.

    VIII. Diffuse and Complex Sources

    What happens when WISE crosses over the Plane, filling its focal plane with every imaginable cosmic source -- diffuse emission, clumpy nebulosity from the disturbed and shocked ISM, discrete fuzzy sources and of course, untold numbers of stars? Take for example the NGC1333 star formation region in Perseus (see image, right). It is full of young stellar objects, many of which are resolved with extended envelopes of hot dust. The background is filled with a complex morphology of gas and dust, and many foreground stars. It is just about impossible to do a robust job of measuring the extended emission as discrete sources (after all, what is discrete and what is continuous with such complex objects).
    But what we can do is create images that are artifact-mitigated and have a flux calibration that is meaningful for extended emission. The goal of the WISE data processing should be to create scientifically valid images that can be used for not only point sources but also surface brightness measurements. Resources will need to be dedicated to artifact mitigation, including those arising from the optics (glints, ghosts, diffraction spikes, off-axis scattering) and those arising from the electronics and detector peculiarities (e.g., mux-bleed, persistence, jail-bars). And finally, flux calibration will require that we understand the scattering properties of our arrays -- a chilling reminder is the insanely complex properties that the IRAC silicon detectors have, thus creating a very difficult flux calibration exercise. We need to spend some resources on understanding the diffuse light, including the zodiacal emission, that WISE will capture.


    IX. Further Considerations


    X. Conclusion

    The carefully planned WISE data processing pipeline, constrained by budget, time and personal limitations, is streamlined to create images and catalogs that are optimized for point sources only. Both data products, however, will be subject to the unique properties of extended sources. On the one hand, point sources that are in close proximity to extended emission (galaxies, nebulae) will have extracted properties that are modified (e.g., contaminated) by the underlying emission. On the other hand, the images will contain both uniform (e.g., zodiacal light) and clumpy extended emission (from resolved galaxies, the Milky Way) that will have properties that demand a separate calibration from point sources. That is to say, for the surface brightness of any given pixel to be meaningful, the extended emission must be characterized and understood well enough to be flux calibrated. Moreover, the science return from basic measurements of extended sources (most importantly, resolved galaxies) will be a boon to the astronomical community. This is a good time for the WISE P.I. and and the science team to consider how WISE should deal with extended sources. My personal opinion is that we should dedicate at least FTE during development of the data processing to measure extended sources, and another 1.5 FTEs to the analysis of extended sources in the raw and subsequent release data products.


    Addendum

    Experiment to test completeness:

    An interesting test would be to take a SWIRE 24um image, convolve and degrade to WISE sensitivity/resolution, mark the 2MASS point and extended sources, then visually examine the image(s) to see if any 24um extended sources are clearly resolved but not identified by 2MASS. This will indicate the necessity (or otherwise) of separately identifying extended sources in WISE images (as opposed to just using a prior list). A similar experiment may be carried out with GLIMPSE/MIPSGAL images, covering the Plane where diffuse/nebular emission will be maximized and differentiated from the 2MASS K-band.

    Science Cases:

    Resolved galaxies will be relatively nearby, similar to the 2MASS XSC volume, stretching to z ~ 0.1. Although the WISE resolution is poor compared to 2MASS, the increase in sensitivity to bulge light (3-5um) and to disk light (12-24um) will compensate to some degree. The science cases for resolved galaxies are thus confined to the local universe, and like 2MASS, the whole-sky survey means that WISE can be used as an effective boundary condition to the high-z studies of galaxy evolution that are now coming to fruition with Spitzer. But unlike 2MASS, WISE will be sensitive to both high-mass galaxies (early types) and to star-forming galaxies (late types), thus complementing/completing the NIR surveys. More science cases are outlined in studies created during the NGSS proposal phase, including nearby galaxies, clusters and normal galaxies (relevant to both resolved and unresolved sources), and large scale structure.


    Last update - [26 June 2007]
    T.H. Jarrett (IPAC/SSC)