HII Region Analysis in Dwarf Irregular Galaxies:

TheSlog & HIIphot

Advisors: Chris Smith & Knut Olsen


We have CTIO 4-m telescope + Mosaic wide-field CCD camera images taken in UBVRI and Halpha, [SII], and [OIII] of  three Local Group dwarf Irregular galaxies: NGC 6822, WLM, and Phoenix, taken as part of the Survey project. As a step towards understanding the mode of star formation in these galaxies, an automated method is used to derive a complete sample of HII regions in at least one of the galaxies, which would be used to compare the properties of the HII regions (through luminosity functions and distribution of morphologies) with those in other Local Group galaxies.  In particular, these galaxies appear to be dominated by bubble-type HII regions; can this impression be quantified, and is there an effect on the luminosity functions?

This project is part of the larger survey project, the Local Group Survey.

O & B stars emit on the order of 1037 erg sec-1 of Halpha -- enough flux to excite hydrogen gas to doubly ionize, producing HII regions. Dwarf irregular galaxies are particularly rich in O & B stars. As a result, these galaxies are excellent candidates for surveying the morphology, structure, and significance of these HII regions.

HII regions within NGC6822 were first identified by Hubble in 1925. Since then, there has been extensive analysis of NGC6822's galactic HII structure. Currently (as of 1988) there have been a total of 157 detected HII regions.  Surveys of NGC6822's HII regions reveal extensive gaseous structure, characteristic of the stellar formation.

Hodge, et al. (1988,PASP) did a survey of HII regions in the dwarf irregular galaxy NGC682.Using a 2.1m telescope on Kitt Peak. Choosing a discernable limit of 2 x 1017 ergs cm-2 sec-1 and three aperature diameters for HII regions, they selected distinct star forming regions of NGC6822 to photometrically analyze (Fig. C), eventually producing an absolute Ha luminosity function and several diameter distributions. Hodge was limited by the telescope size, having to select regions of interest, and aperature approximation, which may lead to flux variation.

Using the CTIO 4m Blanco telescope with the 8K Mosaic imager. Unlike Hodge et al.,  we were able to sample the entire galaxy. We have one night of broadband UBVRI  images as well as narrowband Halpha, SII, and OIII images taken of the dwarf irregular galaxies NGC6822, WLM, and Phoenix. These mosaic images are approximately 2160x2160 arcseconds. With such coverage, we can perform thorough HII-region survey over each entire galaxy. In the case of NGC6822, we hope to detect additional regions, contributing to the known total. Unlike Hodge et al., the HII-region analysis will be done using an entirely automated program HIIphot (written by D. Thilker). Calibrating our NGC6822 flux counts with the accepted Hodge results,  we can then similarly analyze our two additional candidates, the WLM and Phoenix.
Figure A.                                    Figure B.

Fig. A & B above are HII emission subsection images of the galaxy NGC6822. They coorespond to the regions A and C in Hodge, et al. (1988b.). These regions are overlayed in on our H alpha emission image below (Fig C).

Figure C.

Reductions & SLOG

The process of reducing 8Kx8k Mosaic images is an extensive undertaking. Each mosaic image is composed of eight individual CCDs (known as extensions) (Fig. D). Counting each extension individually, we were initially confronted with 2270 separate images. Due to the gaps in the mosaic imager, observations of a single object are dithered. A total of four or five dithered frames are taken and eventually "stacked" to create a single, gap-free image (Fig. E). Piecing  together all extensions of each individual object and then stacking the dithered frames has taken ~60gigabytes and approximately 2 months. For a detailed description of mosaic reduction see the SLOG.

                                           Figure D.                                                                                                        Figure E.

Continuum Subtraction

After completing the reductions we are left with single images of the galaxies. In order to study the HII regions, it is necessary to subtract the stellar contiuum, or remove the stars that do not contribute to the Halpha emission. This is done using the R-band image and the Halpha image.The R band (5500 ~ 9300 Angstroms) spans wavelengths that includes the narrow Halpha band (centered at 6563 Angstroms).

For the first run through, the exposure times for the R-image and Halpha image were syncronized. The R-image was then subtracted from the Halpha .  This usually resulted in oversubtraction, leaving a negative image. To determine a good factor to scale the R-image and Halpha image for subtraction, the fluxes of a group of the same stars in both filters were measured and a ratio of R flux/Halpha flux was established. This number was then applied to the R-image. Subtracting again produced the residual HII.

NGC6822 residual HII regions.

Flux Calibration

In order to calibrate our work using the automated HIIphot, we first sought to "bootstrap" our photometric analysis of NGC6822's HII regions with the published fluxes of Hodge, et al (Hodge, et al. 1988, PASP 100, 935. 1989, PASP101, 32). The bootstrap method, as previously employed by Rand (1992, AJ,103, 815) and Thilker, et al. (2000, AJ, 120, 3070), consists of defining aperatures similar to those used by Hodge for specific HII regions and measuring the flux in counts, using the IRAF package Polyphot. These counts can then be converted to physical units, using the Hodge results. We find an inaccuracy of ~10%, employing this method for approximately twelve well-defined regions. This inaccuracy may, of course, be due to our inabilty to replicate Hodge's aperatures. It may also be the result of Hodge's assignment of three different aperature assumptions, resulting in flux variations. Polyphot's photometry may be to blame as well.


Written by Thilker (Thilker, et al. 2000, AJ, 120, 3070), HIIphot models HII regions using "footprints" and "seeds", where "seeds" are merely trimmed "footprints". Deinfing a "terminal brightness" for flux cutoff (in EM, where 1 EM = 2.22x10-12erg cm-2s-1arcsec-2 ), HIIphot maps the flux in HII regions, regrowing the galaxy using the "footprints" and "seeds", which appear much like map contours. This is an iterative process. Unfortunately we have run into the problem that it takes ten hours to analyze a 500x500 pix section and we have 8kx8k images.  The image below is from Thilker's (2000) paper on the automated task (Fig. F).

Figure F. M51 with various "seed" growth.

Figures G., H., I.

The images above (Fig. G.,H., I.) are three color images of NGC6822, WLM, and Phoenix., respectively  Comparing that with the continuum subtracted Halpha images (Fig. J., K., & L.), gaseous structures become visibly significant. The fluxes of these regions can be analyzed, leading to insight of characteristic stellar formation.


The bubble structure, for example, in the lower right hand corner of NGC6822 becomes extremly complex viewed in the subtracted line minus continuum Halpha band, 6563/75.


Back to the 2001 CTIO REU/PIA Program page...