ScienceProblem: DeepFieldSurveys

PrimaryActor

Astronomer

---

ScienceGoal

To establish the properties of faint extragalactic sources in the Hubble Deep Field (North) at all available wavelengths

---

DataSets

Deep field observations by instruments including:

HST

MERLIN/VLA/EVN/WSRT

ISO

SCUBA

CHANDRA

---

ProblemDescription:

The Hubble Deep Field (HDF) is a 'blank' area of sky observed with unprecedented resolution and sensitivity by the HST, revealing about 3000 faint galaxies within a 3 arcmin-square region (also including flanking fields). Fields of up to 40' centred on the HDF have since been imaged at wavelengths from radio to x-ray. Much effort is involved in aligning the data sets, and searching for significant correlations between sub-sets of properties. For example, it turns out that there are an excess of radio sources (including those too faint to be catalogued) within the error boxes of selected optical sources in the HDF. The selection criteria were optical colours indicating starburst, and the occurence of radio sources was binned as a function of Mv. Other tests could include binning as a function of z for sources with known redshift, and statistical comparison with the occurence of radio sources near non-starburst optical sources and at comparable numbers of random locations. Correlation with other catalogues could then be performed.

---

CurrentSolution:

Very labour-intensive including visual inspection of maps and much time spent comparing large populations selected by various criteria from different catalogues and discovering there was no significant correlation. Some information e.g. redshifts may be only available in separate studies of individial objects.

---

VOSolution:^?

• Automatic registration and calibration:
  • Start with the most astrometrically accurate data set, use a suitable criterion (e.g. compactness, flux density within limits of good but unsaturated snr) to select sources in it. This is probably radio (defines the ICRF) if there is a reasonable source density.
  • Search the next most suitable data-set - e.g. optical HST - high source density, good relative position accuracy, good resolution, first-order transforms eg rotation are defined in the header and can be applied automatically by AG. The max radius for (e.g. radio-optical) matches can be used-supplied or derived automatically eg from the radius in which all radio sources have at least one optical counterpart.
  • Additional criteria can be used for matching, e.g. if tight faint spiral galaxies are unlikely to be radio-galaxies, omit them... (spirals are one of the few things where there are good? existing descriptive algorithms).
  • Use a simple error-weighted least squares minimisation routine (the simplest is aligning centroids) to find the offsets to apply to dataset 2 to align with dataset 1. This fit could be in more than 2 dimensions - I believe there are standard algorithms to fit for x, y offsets, rotation, stretching....? This could only be done to one optical plate (or equivalent) at a time but then applied to all plates (etc) in the same configuration.
  • A few of the brightest (or best-studied) sources may have better positions in table data (e.g. the IRAM position for the brightest SCUBA source in the HDF), or data at some wavelengths may only be available as tables not images, so AG needs to be able to compare with this - an additional user request, perhaps.
  • Return the aligned data sets together with a set of vectors, perhaps as functions of position in plate, for applying to all data from the same source.
  • If the second or subsequent data-set has non-linear errors and accuracy <0".1 is needed, then the user or AG may need to do something more, as the residuals could be unique ie functions of elevation, temperature etc. At this point only the user may have to interactively manipulate images or calculate and enter shift vectors for individual regions.
  • There may also be non-error offsets due to source structure, which the user would have to either sort out interactively or supply an algorithm for.
  • Repeat for other data sets! Probably by comparison with the most highly-populated (default AG choice) pre-aligned data-set, or by user choice (e.g. if X-ray emission is more likely to be aligned with radio).

• If necessary reprocess or convolve data sets to comparable resolutions or to extract other information, e.g.
  • In the case of interferometry (and possibly other) data it may be more economic to store non-image (eg visibility) data and map areas of interest on demand.
  • Zoom in on areas which become of interest e.g. re-map MERLIN-only data at high resolution from the MERLIN+VLA visibility data (the default maps available maximise sensitivity).
  • Form spectral index maps or measure flux densities and then calculate spectral indices from table data
  • Form optical or IR colour or colour-subtracted maps.

• Seach of all available published data.

• Tests for correlations (based on user-supplied criteria) across many catalogues.

• Searches of image (or other) data for uncatalogued sources which become significant if found to co-incide with detections at other wavelengths.

• Search for sources not detected in optical - dust-enshrouded starbursts?

---

Iteration Breakdown:

• Astrogrid.Itn04DeepFieldSurveys - science workflow for case in Iteration 04
• Astrogrid.Itn06DeepFieldSurveys - science workflow for case in Iteration 06
• Astrogrid.Itn08DeepFieldSurveys - science workflow for case in Iteration 08

---

KeyReferences:

Williams, Robert E., Blacker, Brett, Dickinson, Mark, Dixon, W. Van Dyke, Ferguson, Henry C., Fruchter, Andrew S., Giavalisco, Mauro, Gilliland, Ronald L., Heyer, Inge, Katsanis, Rocio, Levay, Zolt, Lucas, Ray A., McElroy, Douglas B., Petro, Larry, Postman, Marc, Adorf, Hans-Martin, & Hook, Richard 1996, Astronomical Journal v.112, p.1335 The Hubble Deep Field: Observations, Data Reduction, and Galaxy Photometry

Richards, E. A., Kellermann, K. I., Fomalont, E. B., Windhorst, R. A., & Partridge, R. B. 1998, The Astronomical Journal, Volume 116, Issue 3, pp. 1039-1054. Radio Emission from Galaxies in the Hubble Deep Field

Richards, E. A. 2000, The Astrophysical Journal, Volume 533, Issue 2, pp. 611-630. The Nature of Radio Emission from Distant Galaxies: The 1.4 GHZ Observations

Muxlow, T. W. B., Wilkinson, P. N., Richards, A. M. S., Kellermann, K. I., Richards, E. A., & Garrett, M. A. 1999, New Astronomy Reviews, Volume 43, Issue 8-10, p. 623-627. High resolution imaging of the Hubble Deep and Flanking Fields

Muxlow, T. W. B. et al., 2002, MNRAS, in prep.

Garrett, M. A.+ 2001, Astronomy and Astrophysics, v.366, p.L5-L8 (2001) AGN and starbursts at high redshift: High resolution EVN radio observations of the Hubble Deep Field

Garrett, M. A., de Bruyn, A. G., Giroletti, M., Baan, W. A., & Schilizzi, R. T. 2000, Astronomy and Astrophysics, v.361, p.L41-L44 (2000) WSRT observations of the Hubble Deep Field region

Aussel, H., Cesarsky, C. J., Elbaz, D., & Starck, J. L. 1999, Astronomy and Astrophysics, v.342, p.313-336 (1999) ISOCAM observations of the Hubble Deep Field reduced with the PRETI method

Hughes, D. H., Serjeant, S., Dunlop, J., Rowan-Robinson, M., Blain, A., Mann, R. G., Ivison, R., Peacock, J., Efstathiou, A., Gear, W., Oliver, S., Lawrence, A., Longair, M., Goldschmidt, P., & Jenness, T. 1998, Nature, 394, 241-247 (1998) High-redshift star formation in the Hubble Deep Field revealed by a submillimetre-wavelength survey.

Brandt, W. N.+ 2001, The Astronomical Journal, Volume 122, Issue 6, pp. 2810-2832. The Chandra Deep Field North Survey. V. 1 Ms Source Catalogs

---

---

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.

---

Generic idea of federating deep multi lambda data sets is a valid driver for VO's. The specific case of the HDF-N has probably been done.

It would be useful to run this as a test case - check results c.f. what is already published. But note - the main data sets here are external ones - so will it be possible to interface to these in the nearterm?

-- NicholasWalton - 17 Apr 2002

Most (?All) are in CDS - in fact for our analysis we have partial 'private' copies - we could probably work something out. As you say, the work has mostly been done e.g astrometric alignment, but it could be a useful test and also a preparation for GOODS.

-- AnitaRichards - 02 May 2002

I have fleshed this out after consulting Tom Muxlow

-- AnitaRichards - 16 May 2002

Related UseCases

See also CommonGroup

LibraryFunctions which will be needed include:

Astrometric alignment; Least squares fit; Express result as probability; Measure noise in data; Apply K-correction; Calculate colour; Compare reults of cosmological models using different parameters.

Specific use cases

AstrometryBootstrap

GetFluxOrUpperLimitAtPosition (including on- and off-source)

InstrumentFootprint (possibly needed to distinguish between non-detections in e.g. ISO due to faintness of object, v. those due to location outside field of view).

RedshiftDetermination

GalaxyMorphologyRecognition

---

UseCaseDesign:

DeepFieldSurveysSD sequence diagramme

---

-- AnitaRichards - 04 Feb 2002 - 07 Aug 2002