next up previous contents
Next: 9.3 Spectrum Location and Up: 9 Low-Dispersion Flux Extraction Previous: 9.1 Noise Models

9.2 Background Flux Extraction

The background flux that is present in IUE images is composed of contributions from several sources, including the null pedestal, particle radiation, radioactive decays within the detector phosphor, halation within the UV converter, background skylight, and scattered light. The integrated effect of the last three sources varies in a complicated manner across the target depending on the spectral flux distribution of the object observed, whereas the general radiation and null components vary slowly across the vidicon tube. In practice, the average background flux in the low-dispersion SI is computed within two regions on either side of the spectrum and this average flux as a function of wavelength is fitted with a sixth-order Chebyshev polynomial.

The two background regions that are examined are each 7 pixels wide and, for large-aperture data, begin at a distance of 13 pixels from the predicted center location of the spectrum. For small-aperture data the two regions begin at a distance of 8 pixels from the center of the spectrum (see Figure 9.1).

Figure 9.1:  Locations of the extraction and background fitting regions within the low-dispersion SI.

The narrow large-aperture extraction region corresponds to the 13-pixel point source slit height, while the broad region is the 23-pixel extended, trailed, multiple, and flat-field source slit height. The entire wavelength space of the low-dispersion SI is included in the analysis. Before calculating the average flux as a function of wavelength for each of the two regions, each spatial image line within the two regions is examined individually in order to identify flagged pixels. The FN values of flagged pixels are temporarily replaced with the FN value of the closest unflagged pixel. Telemetry dropouts in the background regions are identified during this process and $\nu$ flag values of -4 are set in the output 1-D flag spectrum as well as retroactively set in the 2-D low-dispersion resampled flag image. This flag replaces the original value of -8192.

At this point a difference spectrum is computed from the average FN as a function of wavelength for each of the two background regions. This difference spectrum is used to detect the presence of cosmic ray hits that may occur in either of the two regions. The Chebyshev polynomial that will be fitted to the background data is sensitive to broad features such as grazing cosmic ray hits. Since we are only concerned with relatively broad features, the difference spectrum is smoothed with a 7 pixel wide boxcar function. The mean and standard deviation of the smoothed difference spectrum is then computed and any pixel locations that deviate by more than 2$\sigma$ from the mean have their weights for the polynomial fitting step set to zero. All remaining good pixel locations are given equal non-zero weights in the polynomial fit. Pixel locations that are beyond the long-wavelength edge of the camera target area are also excluded from the polynomial fit. Once the Chebyshev fit has been calculated, the values in the 1-D background spectrum that correspond to locations outside the camera target area are replaced with the fitted value of the last valid pixel.


next up previous contents
Next: 9.3 Spectrum Location and Up: 9 Low-Dispersion Flux Extraction Previous: 9.1 Noise Models
Karen Levay
12/4/1997