Kristen Adams
We recorded three to four images of each section of the nebula. Each group of images was identical in the data recorded; thus, after reduction, the individual images of each group could be combined directly to yield a higher intensity signal output. To reduce the data, I performed the following ~tandard CCD procedures:
1. Bad Pixel Corrections
2. Overscan Corrections
3. Zeroing and Trimming
4. Flat Fielding Corrections
I will briefly describe the purpose of each correction and any unusual proceedings in applying it to my data (as applicable). The individual frames referred to, as well as a schematic showing how they relate spatially to each other and a foldout mapping of the center frames, are located in the appendix.
Bad Pixel Corrections
Dr. Mclean explains that bad pixel corrections are necessary to compensate for random, localized pixel defects either inherent in the CCD material (for example, at the semiconductor interface) or induced by manufacturing processes. Bad pixels also occur when a pixel becomes “saturated;” i.e., when too many photons are absorbed by one pixel before the charge levels are cleared from the semiconductor by the pulsing voltage (Mclean 124). When this happens, a charge can “trail” across several pixel columns, obscuring the data recorded in the affected pixels. Additionally, bad pixels can occur at “traps;” Dr. Blouke, et. al., define traps as interfaces in the CCD where small potential barriers keep charge form transferring properly (cited in Robinson 463). Trapping unfortunately occurs most commonly when using CCDs to observe faint objects, Drs. Eccles, Sim, and Tritton report (Eccles 164).
Luckily, bad pixel corrections are fairly easy to apply. I identified three columns (two of which spanned the entire length of the CCD images) which were present on all of the images. I also located a row of bad pixels which spanned the width of the image on frame 12 (the center frame). Using the IRAF computer system, I corrected for these errors by extrapolating from adjacent pixel intensity levels on both sides of the affected pixels. The extrapolation followed a linear fit in projecting the intensity levels across the gap produced by the bad pixels. The resultant images appear uniform, indicating that the extrapolation produced an accurate fit.
Overscan Corrections
Drs. Johnson and Wall state that the overscan technique corrects for the inflation of the output intensity due to sources other than the object of interest. They usually result from design and fabrication parameters on the CCD chip itself which produce a bias level in the CCD readout (Johnson/ Wall, 283). This bias value is analogous to instrumental noise in an electronic circuit; it is the read- out produced when no signal input is present This bias severely limits the low-level signal sensitivity; it is thus a significant problem in nebula detection.
IRAF images are recorded with an accompanying graph of the bias value across the image. Using the IRAF overscan interactive fitting process, I matched the recorded value to a polynomial and used this polynomial to subtract off the bias value from the images. I fit almost all of the images with a three to four degree polynomial, although I used as high as a seven degree polynomial to accurately fit the more complicated bias patterns of a few of the images. Fitting the bias interactively to each image, I was able to correct fairly accurately for the error.
Zeroing and Trimming
Zeroing and Trimming corrections are used to calibrate the images. Zeroing involves subtracting a constant intensity from each image to force the intensity levels of the data to conform to standard scales. Trimming eliminates the areas of the images which do not correspond to real, accurate data. These areas arise mainly from the superposition of the input data, which is limited by the circular aperture of the telescope, on the square surface of the CCD plate. To correct for the zero counts, we programmed the computer to record special images of the sky in which the telescope was not focussed on any specific object I then subtracted these images, called zeros, from all of the data images. I trimmed all of the images (again, using the computer) to a standard 1550 x 1550 pixel image. This is an adequate trim for the 36″ Burrell Schmidt telescope which was used to take the data.
Flat Fielding
Flat fielding is one of the most important CCD corrections. A CCD plate, even if exposed to a perfectly uniform light source, will not record a perfectly uniform image due to a non-flat field. Dr. Mclean explains that non-flat fields arise from impurities outside the CCD (such as dust), small variations in pixel response capabilities (this is usually a manufacturing error), doping of the semiconductor material, or variations in pixel sizes. He warns that these errors are random and strongly wavelength dependent (Mclean 132-3). They can significantly alter even a strong signal.
I observed several severe flat fielding problems in my data. Primarily, I observed a strongly darkened area on the upper right corner of the center and left-hand frames (frames 6, 7, 8, 11, 12, 13, and 14 ). This area was so dark as to totally obscure all of the data in the affected region. In addition, I found a strong vertical gradient to exist on all of the frames; this gradient caused serious boundary problems when I tried to fit the images into a coherent map.
To correct for the errors, we recorded separate “flat fields” for each of the filters we used to map LBN434. (Separating flat fields by filter prevents errors due to the wavelength dependency of the effect) Flat fields are images without objects; i.e., they are images of a uniform light source. We made these flat fields by recording three to four images (per filter) of the sky before many stars were out, moving the position of the telescope between each image. (We moved the telescope to avoid recording the same stars in the same places on the images; such stars would add constructively when the individual frames were combined and therefore appear as objects in the images.) By superimposing the three corresponding flat fields, each of a different part of the sky, I obtained a single flat field frame for each filter which was effectively free of any objects. I then divided each of the nebula images, pixel by pixel, by the appropriate flat field (according to the filter used in making the image).
Unfortunately, due to instrumentation imperfections, these images still showed a marked vertical gradient and obscured corner area after the flat fielding. To fully correct the error, I was forced to result to more creative means. With the help of Dr. Moody, I found that I could obtain a better flat field by combining a standard flat field with a nebula image using a “make-sky-flat” routine on the IRAF system. This routine subtracts the stars on the image and then uses the gradient inherent on the image as a first order correction to the flat field. Although somewhat unconventional, I found that this method “personalized” the standard flat fields to each image, greatly enhancing the data. Using the revised flat fields, I was able to eliminate enough of the obscured comer area to see the data behind it. The new flats also significantly improved the gradient problems.
Final Analysis
As a final step, I combined identical images to enhance the signal intensity. I did this by marking star coordinates on each of the frames and then shifting the frames to ensure that the coordinates matched exactly; such a match is obviously imperative for a superposition of data. These procedures prepared the data for juxtaposition (to create the mapping) and analysis.
MAPPING
I fitted the frames together as shown on the schematic (the first page of the appendix; a full copy of the appendix is located in the BYU Honors Office). The nebula can be clearly seen spanning the heights of frames 11, 12, 13, and the lower portion of frame 14. It has an irregular shape with an approximate center in the upper center of frame 12, corresponding to -49.94 ° galactic latitude and 95.90° galactic longitude. It extends slightly into the upper right portion of frames 16 and 17, and the lower left portion of frames 8 and 9. (Due to the gradient problems which still remain, it is more difficult to see the nebula on these outer frames; however, its presence on the center frames is more than enough to establish its existence and to suggest that it extends into these outer frames.) The foldout of the center frames shows a continuum of the bulk of the nebula. (The red lines were added to emphasize the boundaries which are more clearly visible on the computer display than in the hard copies in the appendix.) When the frames are properly fitted together, the entire nebula can be inscribed in a square with sides about 2,050 pixels long. This corresponds to about 4.5 minutes of arc or about 1.5 degrees. These dimensions indicate an area of approximately 2.25 square degrees for the nebula. These results are in fairly good agreement with Lynd’s preliminary data.
Two significant features became evident in the process of mapping the nebula. First, I found that the Abell2657 galaxy cluster (marked in red in the lower middle section of frame II) is clearly obscured by the nebula. Second, I noted the presence of two stars in the lower left portion of frame 11 which appear to be enveloped in an elliptical, nebula-free region (marked in green on frame 11). Such patterns are typical of stars which are being created from the gas and dust of the nebula; if further study confirms that this is indeed an actual phenomenon, its presence would strongly support the validity of the existence of LBN434.
ANALYSIS
Nebula LBN434 is important for two main reasons: first, because of where it is, and second, because of what it covers. Its location is significant because it lies in a region which is believed to be free of nebulae, based on the Burstein-Heiles study. Astronomers have accepted the results of the Burstein-Heiles study to be completely accurate; they have used it as a guide in their work without verifying the validity of the results. However, my study of LBN434 indicates that a nebula does indeed exist in one of these forbidden regions. Moreover, it lies directly on the border of the 50° cone which defines high galactic latitudes and which is supposed to be nebula free. If a nebula of significant extent, such as LBN434, can exist in this region, it is possible that nebulae may exist even in the highest galactic latitudes. Based on my findings, I believe that, while the Burstein-Heiles study is useful as a guide, it should not be elevated to the status of a “truth,” nor accepted as infallible. Clearly, as telescopes and recording devices become more sensitive and able to probe more deeply, we should be constantly and carefully checking past assumptions against new possibilities.
LBN434 is also of significance because it covers the Abell2567 galaxy cluster (marked in red on frame eleven, located in the appendix). This cluster has been widely studied; it is part of an important analysis of galaxy morphology by Dr. Alan Dressler of the Carnegie Institute of Washington. In this study, Dr. Dressler grouped galaxies and clusters in both spatial and red shift coordinates to determine galaxy densities according to type (Dressler, March 1980, 352-355). Although he studied Abell2657 and found total and bulge magnitudes for the galaxies which make up this cluster, he did not realize that it was shielded by LBN434. The fact that a nebula is obscuring some of the light which is reaching earth from this cluster unfortunately means that the measurements of its magnitude and distance which Dressler used are incorrect.