Notes on the reduction and analysis of Hubble Space Telescope images of NGC 1808 by Max Mutchler 1. Introduction --------------- Hubble Space Telescope images of the starburst galaxy NGC 1808 were taken with WFPC2 on 14 August 1997. This was observing program 6872 by Jim Flood -- the last of 13 HST amateur programs. rootnames filter exposure time --------- ------------- ------------- u42l0101 F814W I 140 sec u42l0102 F814W I 140 sec u42l0103 F658N [NII] 700 sec u42l0104 F658N [NII] 700 sec u42l0105 F675W R 120 sec u42l0106 F675W R 120 sec Garth Illingworth took a snapshot image of NGC 1808 in 1994. The pointing is poor because no guide stars were used (gyros only). Apparently there was a large slew just before these images, so a gyro drift of about 35 arcseconds placed the nucleus on the edge of the WF4 chip, with much of the nuclear region vignetted. Garth took two 80 second exposures with the F606W filter. But perhaps this data could be used to extend our color baseline, and help us distinguish age/colors: R-I vs V-R The R features must be old or obscured; the V features must be old but have low extinction. Luis Colina has a UV (F218W) exposure of NGC 1808 in his SNAPshot program 6358, but it hasn't been scheduled yet. If he gets this data, we could try to get U-I colors/ages (although extinction must be handled carefully). If we could have had one more orbit, we could get another emission image and get line/gas excitation...see Osterbrock plot. Note that since the recessional velocity of NGC 1808 is 989 km/sec, the H-alpha emission at 656 nm is redshifted to 658 nm, hence we chose the F658N [NII] filter: redshift = v/c = 989 / 300,000 = 0.0033 delta lambda = 0.0033 * 656 = 2.2 nm redshifted lambda = 656 + 2.2 = 658.2 nm The telescope pointing was a particular concern which turned out well. The nucleus of the galaxy (i.e. the starburst region) is on the higher-resolution PC1 chip. The U3 ORIENT angle is 245 degrees (V3 position angle is 65 degrees), which places the line of HII regions in the galactic bar on the WF2 and WF4 chips. Basic NGC 1808 information: ----------------------------- ------------------------------------ right ascension (RA) 05h 07m 42.0s * declination (DEC) -37d 30' 51.0" * inclination angle 57 degrees position angel of major axis 135 degrees (bar is 155 degrees) recessional velocity 989 +/- 10 km/s distance (v/Ho) 13.2 Mpc = 43 million lightyears ** ----------------------------- ------------------------------------ * These were the coordinates used to point the telescope (i.e. the coordinates of the center of the PC1 chip). The central star cluster seen in the PC1 image is clearly offset from the center of the chip -- I will measure more accurate coordinates from this image.... ** Distance assuming the Hubble contant Ho = 75 km/sec/Mpc. This implies that the full mosaicked image is about 25,000 lightyears, and the PC1 image is about 5700 lightyears on each side. Each WF pixel is 100 milli-arcseconds or about 16 lightyears wide, and each PC pixel is 46 milli-arcseconds or about 7 lightyears wide. So the central star cluster is over 100 lightyears across, while the other star clusters are 10-50 lightyears across. 2. Calibration, reduction, and analysis --------------------------------------- The FITS files were retrieved from the HST archive and converted to GEIS format (using the strfits task) for further processing with IRAF/STSDAS. There was no need to recalibrate the images, because the pipeline-calibrated images were processed with the best reference files. In particular, the best DARK reference files were used. Two images were taken with each filter, so cosmic rays were removed by combining the two images with the crrej task. The sky background was removed from the combined images using gsky. The removal of hot pixels is the trickiest part of the reduction. There are a few different procedures, each with shortcomings. At present, the multi-phase approach (#3) was used: 1. Run warmpix (need hot pixel list from WWW and the DQF (*.c1h) files. This was difficult to get right....unsatisfactory results. 2. Drizzle the images using the DQF (*.c1f) as the bad pixel mask (after converting it using invert). This could also remove the small shifts mentioned below. 3. Multi-phase approach: run cosmicrays to get most hot pixels, then use imedit to manually remove additional hot pixels and bad columns, blinking with DQF (*.c1h) to verify that only flagged pixels were fixed. One problem with many of these tasks is that they only work on single-group images. When I'm done with these reductions, I'd like to create a mosaicked version of the clean images. However, simply using gcopy to get the single-group images into one multi-group image causes problems for the wmosaic task. Apparently, some critical header information gets lost in the process, and wmosaic fails. To produce a clean press-release mosaicked image, I manually fixed the worst hot pixels in the mosaicked images, and Zolt Levay did some additional cosmetic cleanup. Continuum subtraction --------------------- There is a small sub-pixel shift between the images (most noticeable in the PC1 images, since the pixels are "smaller"). Olivia Lupie and Matt Lallo have verified that the pointing is OK, so it looks like the shifts are due to the "filter wedge effect", i.e. because each filter has a slightly different optical thickness. Before we can subtract the continuum R image from the H-alpha image, the images must be perfectly registered. However, it is impossible to accurately measure the shifts between the images since they don't share enough common features. For example, it's impossible to tell if positional differences between the point-like star clusters are due to the filter wedge effect, or due to nebulosity seen in one image but not the other. I am seeking advice from others who have produced continuum-subtracted images to resolve theis problem. Once the shifts are accurately known, the imshift task can be used to align the images. Michael Dahlem suggests using pradprof to measure the shifts accurately. To check that the continuum was properly subtracted, pick a few areas in the continuum subtracted image where there is R flux but no H-alpha flux, and make sure the pixel value is zero there. To create a continuum-subtracted H-alpha emission image, the R image was subtracted from the redshifted H-alpha [NII] image in the following manner: A BB/stellar/flat spectrum was chosen, and synphot was used to estimate the counts-per-sec in both the R (F675W) and and redshifted H-alpha (F658N) filters (857 and 24, respectively). The R/NII flux ratio for WFPC2 is then 857/24=35.7. Sky backgrounds and DARK frames were already subtracted from the images. Divide R by it's exposure time (240 seconds) to create a Rcps image. Then didvide the Rcps image by 35.7, the flux ratio. Then multiply the image by the NII exp time to estimate the counts you would expect in the NII image (based on the R image). Subtract this image from the NII image to get a continuum subtracted NII emission image. Michael Dahlem and Sylvia Baggett (PASP Dec 97) made a smoothed copy of Illingworth's F606W image and subtracted it from the original (unsharp masking) to enhance the filamentary edges. We should try the same with these images, and see if we can trace the filaments down to their origin in the circumnuclear region. We can make 2-color maps of the nucleus to see color gradients, sort out features by color. The initial contour plots seem to show that the nucleus is brighter in H-alpha. Object positions ---------------- Get "best" coordinates for the center of the galaxy. Contour plots of the core seems to show two peaks -- and their positions vary in each filter. Could a dust lane be bisecting the nucleus? Coordinates of various objects in the galaxy (nuclear hotspots, HII regions in the bar, etc) were measured using daofind. Compare with Tacconi (p. 920). Measure position angle of galaxy and bar (line of HII regions), compare to previous estimates. Evidence of interaction with NGC 1792. Plot positions of HII regions and fit line to get PA. Grazing encounter with NGC 1792 (Mihos 92). Photometry ---------- Tricky due to all the obscuration/extinction. See Krabbe's extinction map. Photometry of objects was done with daophot? Compare results with Tacconi (p. 921). Broad band photometry (2-color diagrams). Get F606W colors from Illingworth's SNAP too? Overlay radio map, isophotes..... Size of starbursts? Lehnert p. 551 What is resolution? 1 PC pixel = x kpc 1 WF pixel = y kpc 4 Strength of starbursts? L / (v / sin i) See Lehnert & Heckman (Dec 93 paper 3). Color/age of starbursts? Lehnert p. 551 Estimate hotspot masses (red giants). Tacconi-Garman Parallel NIC and STIS data probably useless (adjacent fields). Bar morphology similar to NGC 7479. There is evidence of an inner bar, where gas falls inward to feed the starburst. Wish list --------- NICMOS K-band observations would reveal more star clusters, and would probably show more correlation with the radio sources. NIC1 camera is 11x11", NIC2 camera is 20x20" (PC1 is 35x35"). OIII or NII emission images would show line/gas excitation...Osterbrock plot. Publication ----------- The wmosaic task was used to create a mosaicked set of I, H-alpha, and R images, which Zolt Levay made into a false color-composite image. We presented preliminary results at the AAS meeting in Washington DC on 9 January 1998. USA Weekend magazine (a Sunday newspaper insert) will do an article about this project in their 22 March 1998 edition. The article will also be available via their website: http://www.usaweekend.com/ We are still undecided if this work will result in a refereed journal article, or if a more technical but popular science article will be written, while we seek additional data. References and related notes ---------------------------- Morgan, 1958 - Noted a group of optical hotspots in the central 1kpc region. Sersic, Pastoriza, 1965 - Included NGC 1808 in their catalog of galaxies with peculiar and complex nuclei. Burbidge & Burbidge, 1968 Arp & Bertola, 1970 Osmer, Smith, and Weedman, 1974 Veron-Cetty, Veron, 1985 Phillips, 1993 - Optical spectroscopy shows intense emission lines are produced in the central few kpc, and indicate patchy extinction toward the nuclear region. - Emission traces the location of giant HII regions which are ionized by young massive stars produced by a starburst. - Calibrate our H-alpha data with their line fluxes? Larson & Tinsley, 1978 van den Berg, 1978 - Peculiar optical morphology (similar to M82) and starburst activity may have been induced by tidal interaction with nearby NGC 1792, whic also appears to have an asymmetric morphology Laustsen, 1987 - Dust filaments protrude from the nucleus to the halo Haynes et al, 1975 Large et al, 1981 Ekers et al 1989 - NGC 1808 is a prominent radio source, and has been detected at frequencies from 0.03 to 14.8 GHz Sakia et al, 1990 - Mapped the inner kpc region using the VLA at 1.5, 5, and 15 GHz. These high-resolution (1" at 5GHz) measurements revealed an intense nuclear source surrounded by several compact radio sources within the central 1kpc. There is little correspondence between the compact radio sources and the optical hotspots. - Created A-E naming scheme for hotspots. Hotspot D is the nucleus. Collins - Radio maps at 0.3" resolution Dahlem, 1990 - The radio emission is produced by supernova remnants Krabbe, Sternberg, and Genzel, 1994, ApJ 425, 72 "Near IR spectral imaging of NGC 1808: Probing the starburst" - Br-gamma emmission knots do not coincide with the locations of optical hotspots. The Br-gamma hotspots trace the actual star clusters and their associated HII regions and SNR (and are better correlated with Sakia's compact radio sources), while the optical hotspots are mainly directions of low extinction. - There is no IR evidence for an embedded AGN - The starburst clusters are the dominant sources of the thermal far-IR dust emission - Distance = 10.9 Mpc, 1" = 50 pc = 163 ly (for Ho=75) - Define NGC 1808 as "morphologically peculiar spiral" (Sbc pec) galaxy - Produced an extinction map of the circumnuclear region. Visual extinctions range from ~3 to ~5 - Optical images are likely severley distorted by patchy foreground extinction toward the circumnuclear region. IR images are probably only slightly affected by foreground extinction. - Much of the foregound extinction may be caused by the remarkable dust filaments, such that the optical hotspots trace regions of low foreground extinction. - The nuclear starburst must be at least 50 million years old, and can be no older than 100 million years old. Star formation has been rapid and continuous. Without an influx of fresh molecular gas into the circumnuclear region, the star forming activity can only be maintained at this rate for another 6 to 20 million years. - There is little evidence for very hot and massive star formation Dahlem, Aalto, Klein, Booth, Mebold, Wielebinski, Lesch, 1990, Astron. Astrophys. 240, 237 - Radio continuum and CO line observations, disk-halo interactions - A rotating molecular ring of about 1.4kpc (4500ly) diameter. - See Ikeuchi (1988) and Norman & Ikeuchi (1989) for disk-halo review. - Filaments driven by supernovae in the starburst Dahlem, Baggett, PASP Dec 1997 - unsharp masking of Illingworth's SNAPshots of NGC 1808: subtracting a smoothed image sharpens the edges of the dusty filamentary structure Koribalski, Dettmar, Mebold, Wielebinski, 1996, Astron. Astrophys. 315, 71 "Gas streaming along the bar in NGC 1808" - H-alpha and HI data - an unusual bar at PA=155d, about 20d offset from the major axis, with a chain of HII regions 6kpc long, indicating intense knots of young stars. Similar to bar in NGC 7479. - The inward streaming of gas in the bar is seen in the HI velocity field. - The leading side of the bar has shock fronts, which can bee seen as broad dust lanes emerging near the nuclear regionand curved along the leading sides of the bar. - Inside the corotation radius, the gas accumulates at the inner Lindblad resonances (at 1kpc), which coincide with a nuclear ring or spiral. - HI absorption data reveals a fast-rotating nuclear torus with a radius 10" or 500pc (1600 ly). Most of the hotspots are also found at this radius. This is likely the radius of the inner Lindblad resonance (ILR). Forbes, Boisson, Ward, 1992, Mon. Not. R. Astron Soc, 259,293 "NGC 1808: AGN or starburst?" - The starburst can explain the nuclear activity without invoking a black hole (AGN). - U/I color map shows that hotspots A,C,E appear blue (closer), while B and D are highly reddened (embedded in dust and obscured). Hotspot D is the nucleus. - Hotspot positions and UBVRIJK mags given. They become less well defined at longer wavelengths. - Starburst is less than 50 million years old. - The compact radio sources found by Sakia must be SNRs - An off-nucleus compact 20cm radio source exists between hotspots A and C, assumed to be a powerful SNR -- do we resolve it optically? It must be 50-200 times more powerful than Cas A. - Late note: Axon's 3.6cm map resolves the nucleus with a size of 0.435 x 0.222 arcsec2 (about 25pc or 80ly assuming d-16.4 Mpc). Tacconi-Garman, Sternberg, Eckart, 1996, AJ 112, 3, 918 - K-band 2.2 micron (0.6") imaging of NGC 1808: IR observations of super star clusters. - Reveals many compact (under 0.7") near-IR continuum sources, which are likely clusters containing a few to several hundred massive (~30 solar masses) red supergiants. The inferred cluster masses are between 100,000 and 1 million solar masses. - Conclude that the the clusters are young (~10Myr) and decaying rapidly. - A 370pc long IR bar at PA=155d, and evidence of inner spiral or ring-like structures are present, and aligned with the extended 6kpc bar. - Most of the near-IR light in the resolved nuclear source is likely produced by giants and supergiant stars. - Distance = 10.9 Mpc = 36 million ly (Ho=75), 1"=53pc - Krabbe 1994 estimated the total rate of star formation in the circumnuclear region to be about 11 solar masses per year. - List positions and K flux for 26 hotspots and the nucleus. They divided a smoothed image by the original to get good positions for the brightest spots and fainter spots (a "high-pass" filtering). - Estimate that most knots are about 20pc - Resolved nucleus: FWHM of 0.79" x 0.64" or 27x12 pc - Sakia's 1950 coordinates of the nuclear source: 05 05 58.49, -36 34 36.7 - The prominent IR sources do not coincide with the radio sources. The radio sources probably correspond to the random locations of the most recent (last 200yrs) supernovae. - Holtzman et al has concluded that all the clusters seen in starbursts environments with HST must be young globular clusters.