| Summer 2005 Research |
Data analyzed for this project were collected at the Pine Bluff Observatory near Madison, Wisconsin, in the winter of 2004-5. For calibration purposes, data were periodically collected from a cerium-neon (CeNe) lamp throughout the course of each night’s observations.
Using a series of IDL programs a small section of the "raw interferogram" was selected for data analysis. The white rectangle indicates the region of data analysis. This region is chosen because it avoids most of the sharp transitions from light to dark areas, which could cause problems with the Fourier transformation.
After the interferogram is cropped down, a Hanning Function is applied to the image in order to smooth the transitions at the edges of the image down to zero intensity, and then a Fourier amplitude transformation is performed on the image. The Fourier transformation produces a "power spectrum" which is then used to produce the spectrum to be analyzed. The light colored line running across the "power spectrum" is the signal obtained from the observation. The rows of bins that contain the signal are selected and the relative intensities are added together to form the spectrum.
Bins vs. Relative Intensity
Since a filter was used to allow only a small wavelength range, and some of the interesting data fell near the edges of the filter (filter profile shown below), the spectrum was divided by the filter to correct for this. The dashed lines indicate the portion of the spectrum that contains regions of interest to [OII] 3727Å observations. The dotted lines indicate the region which was used for the calibration spectrum.
Once the calibration spectrum had been obtained from the "raw interferogram" the task was to covert the bin scale to a wavelenth in Angstroms scale. In order to obtain an accurate Angstrom per bin ratio, seven bright peaks from a calibration spectrum were fit with Gaussians, using Voigtfit (a Gaussian fitting program), to determine the centroids of each peak. Once the centroids were accurately located, known emission lines of cerium and neon were used to determine the corresponding Angstrom measurement by comparing three of the centroid locations to our best guess of the corresponding emission lines within our spectral range. The Angstrom per bin ratio obtained was then used to predict the position of the other centroids. The process was repeated, and the Angstrom per bin ratio was adjusted until the centroids of the spectrum peaks and the known emission wavelengths matched. An accurate dispersion was impossible to determine until the discovery of iron emission within our cerium-neon calibration spectra. After known emissions of iron were accounted for the Angstrom per bin dispersion accurately predicted the positions of the four other peaks (the green line) with known centroids, as well as five additional fainter peaks.
The red crosses indicate the centroid position of seven peaks. The data points all falling on the best fit line lends confidence to the dispersion of ~.008Å per bin (shown as "pixel" in the plot).
A correction also needed to made due to an instrument drift which caused observations to be slightly offset from one another. Tracking the positions of several peaks which appeared on each of the calibration spectra throughout the night enabled us to determine the bin drift over time of the CCD and thus correct for it.
Instrument Drift ~.039 bin per minute
A similar procedure was used to identify peaks on the observational data collected. Observations of the night sky were all shifted, using a best correlation IDL program, to the observation spectrum that was taken closest in time to the CeNe calibration spectrum. The observations were then averaged together to increase the signal to noise ratio and then fit with Gaussians, using the same VoigtFit program, to determine the centroids of several bright peaks. The dispersion obtained during calibration was then applied to the average spectrum to ascertain the wavelengths of the emissions.
Bin vs. Relative Intensity
The dotted lines above indicate the area to which the best correlation program was appplied. The thick red line is the aveage of the observations of this region for the night.