SExtractor is a source extraction software created by Bertin & Arnouts (1996). SEx- tractor is able to separate blended objects, perform photometry in the extracted sources and classify the sources. The software has six steps: estimating the sky background, thresholding (locating the sources), deblending, filtering of the detections, photometry and star/galaxy separation.
3.6. Galaxy Catalogues 95
Table 3.13: Pointings used to create flat fields for ADF-S images 01/02/2007 to 09/02/2007.
Pointing Number Date Programme N3 N4 S7 S11 L15 L24
1100013-001 08/02/2007 MP-AGNUL 01 - - - - - 1300006-001 09/02/2007 MP-CLEVL 04 04 12 12 12 12 1300008-001 09/02/2007 MP-CLEVL 04 04 12 12 12 12 1300137-001 07/02/2007 MP-CLEVL - 05 - 30 - 30 1300137-002 07/02/2007 MP-CLEVL - 05 - - - 30 1300137-003 08/02/2007 MP-CLEVL - 05 - 30 - 30 1300158-001 08/02/2007 MP-CLEVL - 05 - - - 30 1400421-001 08/02/2007 MP-ISMGN 04 04 12 12 - - 1500705-001 04/02/2007 MP-SOSOS 01 - - - - - 1700039-001 04/02/2007 MP-AGBGA - - - - 12 12 1700040-001 03/02/2007 MP-AGBGA - - - - 12 12 1800014-001 05/02/2007 MP-VEGAD - - - - 12 12 1800015-001 05/02/2007 MP-VEGAD - - - - 12 12 1800065-001 07/02/2007 MP-VEGAD 04 04 12 - - - 1800172-001 03/02/2007 MP-VEGAD 04 04 12 - - - 1800183-001 03/02/2007 MP-VEGAD - - - - 12 12 2110508-001 04/02/2007 LS-LSNEP 03 02 09 06 09 06 2110529-001 06/02/2007 LS-LSNEP 03 03 09 09 09 09 2110562-001 09/02/2007 LS-LSNEP 03 03 09 09 09 09 2110607-001 08/02/2007 LS-LSNEP 03 02 09 06 09 06 2110664-001 08/02/2007 LS-LSNEP 03 02 09 06 09 09 2110732-001 01/02/2007 LS-LSNEP 03 02 09 06 09 06 2110813-001 02/02/2007 LS-LSNEP 03 02 09 06 09 06 2110905-001 02/02/2007 LS-LSNEP 03 02 09 06 09 06 3230001-001 02/02/2007 OT-ISAS-Z1LBG 04 04 - - 12 12 3230001-002 08/02/2007 OT-ISAS-Z1LBG 03 04 - - 09 12 3230001-003 08/02/2007 OT-ISAS-Z1LBG 04 04 - - 12 12 3230002-001 05/02/2007 OT-ISAS-Z1LBG 04 04 12 12 12 12 3230002-002 05/02/2007 OT-ISAS-Z1LBG 04 04 12 12 12 12 3230002-003 05/02/2007 OT-ISAS-Z1LBG 04 04 12 12 12 12 4040001-004 01/02/2007 OT-ESA-IRLBG 05 - 30 30 - 4040001-005 01/02/2007 OT-ESA-IRLBG 05 - - - 30 - 4230006-001 03/02/2007 OT-ESA-EROMU 03 04 09 - 09 12 5121024-001 05/02/2007 DT-DTIRC 03 02 - - 09 06 5122024-001 06/02/2007 DT-DTIRC - - - - 09 06
Figure 3.5: Image showing the ghost near the top left of the image.
Table 3.14: ADF-S discarded frames. Filter Pointing Number Frame Number Reason
N4 3200002_001 F006033902_N002 Many hot pixels S7 3200004_001 F006038908_S004 Flux error L15 3200003_001 F006036697_L003 Cosmic ray L15 3200003_001 F006036697_L004 Cosmic ray L15 3200004_001 F006038908_L004 Flux error L24 3200003_001 F006036693_L004 Many hot pixels L24 3200003_001 F006036699_L002 Cosmic ray L24 3200003_001 F006036699_L003 Cosmic ray L24 3200003_001 F006036699_L004 Cosmic ray
L24 3200006_001 F006034162_L002 Artificial stripe pattern L24 3200006_001 F006034162_L003 Artificial stripe pattern L24 3200006_001 F006034162_L004 Artificial stripe pattern
3.6. Galaxy Catalogues 97
(a) ADF-S N3 (b) ADF-S N4
(c) ADF-S S7 (d) ADF-S S11
(e) ADF-S L15 (f) ADF-S L24
The first step SExtractor performs on an image passed through it, is to estimate the sky background. The sky background is a background map of what the image would look like if there were no sources. This step is separating the flux of each individual pixel into background flux and flux from the source. Firstly, sigma clipping is used to test for crowded/uncrowded areas. For uncrowded fields the interim sky background is given by the mean of the clipped histogram, and for crowded fields the interim map is given the mode of the histogram. Then a median filter is applied to the interim map; this removes over dense regions caused by bright stars.
The second step is detecting the sources by using a thresholding method. This flags pixels above a certain threshold as possible sources and extracts a specific number of pixels from around the source. Bertin & Arnouts (1996) have chosen a thresholding method because it works well at finding low surface brightness sources.
Step three performs the deblending of merged objects. This step separates two (or more) sources which have merged together to appear as one source. Bertin & Arnouts (1996) use the theory that each source will have its own flux peak within the area of the total flux. A model of the light distribution is created for each set of pixels extracted in the detection step. The model is created by applying 3D thresholds to each set of pix- els, from the peak flux to the lowest flux. An algorithm is then used to test for blended sources. Figure 3.7 shows a 2D representation of how the algorithm works. The al- gorithm begins at the peak flux and descends down the branch. At a flux trough the algorithm will flag the flux on the other side of the trough as a source, if the integrated pixel intensity of the possible second source is greater than a specific percentage of the total luminosity of the extracted pixels. The algorithm will then continue down to smaller fluxes, creating the ‘tree like structure’ in Figure 3.7
The fourth step filters the detections, checking the faint sources for false detections. This step subtracts the contribution to the mean surface brightness of the object in question. If, after subtraction, the flux of the object is below the flux threshold, then the object is classed as a false detection, and is not flagged as a source.
3.6. Galaxy Catalogues 99
Figure 3.7: A 2D representation of how deblending is performed in SExtractor. The outline is the flux profile. The thick lines, which represent a tree-like structure, show the path the algorithm takes. The shaded areas represent the luminosity above the horizontal branch (Bertin & Arnouts, 1996).
different methods for finding the photometry. The three methods are: corrected isopho- tal magnitudes, adaptive aperture magnitudes and an estimate of the ‘total’ magnitude of the source. For the work of this thesis the adaptive aperture magnitude setting was used, see Section 3.6.1.3 for a description of the procedure.
The sixth and final step in SExtractor performs a star-galaxy separation. This step clas- sifies sources as well as separating galaxies and stars. SExtractor uses a neural network to distinguish between galaxies and stars. It had been shown that SExtractor does not correctly class faint stars, but classifies them as galaxies; as SExtractor is unable to distinguish between a point source of a faint star and a point source of a galaxy. The galaxies in ELAIS-N1 and ADF-S are very rarely extended sources, they are nearly all point sources, and hence morphologically look like stars. For this reason, for the work of this thesis, the inbuilt star/galaxy separation in SExtractor was not used, and stellar subtraction was performed independently.
3.6.1.2 How Source Extraction Was Performed
To create the source catalogue for the IRAC Dark Field, ELAIS-N1 and ADF-S mo- saicked images, SExtractor was used (Bertin & Arnouts, 1996). Section 3.6.1.1 gives a summary of how SExtractor performs source extraction. The SExtractor settings used for the source extraction are shown in Table 3.15. These settings are based on those used in Murata et al. (2013) for the North Ecliptic Pole (NEP) survey. The main dif- ferences from the NEP settings and the ones used in the work of this thesis, are that for the NIR and MIR-L detector images the minimum area = 5 and the threshold = 3, and for the MIR-S detector the minimum area = 5 and the threshold = 3.5. The settings were optimised to have the least number of false source detections (see Section 3.7.3) and the greatest number of true source detections.
In the ADF-S NIR and MIR-S mosaicked images, the brightest stars had diffraction spikes, and SExtractor extracted these spikes as sources. Due to this the brightest stars were masked.
3.6.1.3 Photometry
As stated in Section 3.6.1.1, this thesis uses the adaptive aperture magnitude method (as described below) in SExtractor to calculate the photometry of the detected sources. Bertin & Arnouts (1996) chose to use an adaptive aperture to correct for the fact that normal aperture photometry does not work so well in crowded fields and because using an adaptive aperture allows the source extraction to run more quickly.
The adaptive aperture photometry method was chosen because it works better for cal- culating the photometry of non-stellar sources than the isophotal method.
Adaptive aperture photometry uses an algorithm to calculate second order moment (with mean standard deviation σiso) of the profile of the extracted source. 6σiso are
used to calculate the elliptical aperture, where the ellipticity of the aperture is given by ε, and the position angle is given by θ . 6σisois approximately equal to 2 isophotal
3.6. Galaxy Catalogues 101
Table 3.15: SExtractor parameters.
Parameter value DETECT_TYPE CCD DETECT_MINAREA 5 DETECT_THRESH 3 ANALYSIS_THRESH 1.5 FILTER Y FILTER_NAME gauss_2.5_5x5.conv DEBLEND_NTHRESH 32 DEBLEND_MINCONT 0 CLEAN N CLEAN_PARA 1.0 MASK_TYPE CORRECT PHOT_APERTURES 8 PHOT_AUTOPARAMS 2.5, 3.5 SATUR_LEVEL 50000.0 MAG_ZEROPOINT 0.0 MAG_GAMMA 4.0 GAIN 1.0 PIXEL_SCALE 0 SEEING_FWHM 1.2 BACK_SIZE 64 BACK_FILTERSIZE 3 BACKPHOTO_TYPE GLOBAL BACKPHOTO_THICK 24 WEIGHT_GAIN Y WEIGHT_IMAGE YES WEIGHT_TYPE MAP_RMS
Table 3.16: Table showing the flux conversion and aperture corrections for the filters used in the IRAC Dark Field, ELAIS-N1 and ADF-S, for ADU to micro-Janskey.
Filter Flux Conversion Aperture Correction
N3 0.4394 0.873 N4 IRC03 0.2584 0.871 N4 IRC05 0.1753 0.871 S7 1.022 0.918 S11 0.7732 0.902 L15 1.691 0.852 L18W 1.146 0.793 L24 0.04892 0.685
parameters εkr1and kr1/ε give the semimajor and minor axes of the elliptical area that
the flux of the source is contained within. ε is the ellipticity of the aperture and r1 is
the first moment as calculated by Equation 3.13. Bertin & Arnouts (1996) recommend a value of k = 2, as this is when systematic and random errors are at a minimum.
r1= ∑ rI(r) ∑ I(r)
(3.13)
The flux of the extracted sources were in analogue to digital units (ADU). To cal- culate the galaxy flux from ADU to Janskys, the flux conversion from Tanabé et al. (2008) was used. The Jansky (Jy) is a non-standard SI unit, used to measure flux. 1 Jy = 10−26Wm−2Hz−1. An aperture correction is required to correct for galaxy flux not included within the aperture radius. After this, the aperture correction from Ari- matsu et al. (2011) was applied. Table 3.16 shows the flux conversion and aperture correction for each of the filters in the IRAC Dark Field, ELAIS-N1 and ADF-S.