ISSA Explanatory Supplement
II. IRAS SKY SURVEY ATLAS OVERVIEW
A. Changes and Improvements in Atlas
- Improvements in Relative Calibration
- Zodiacal Foreground Removal to Permit Coaddition
- Destripers to Stabilize Detector Baselines
- Oversampling to Improve the Representation of Spatial Information
- Improved Pointing Information
- Particle Radiation Removal
- Known Asteroid Removal from the Coadded Images
- Full-Sized Detectors
The processing that created the ISSA images was designed to correct several problems that limited the sensitivity and usability of the SkyFlux images. These problems include the effects of the photon-induced responsivity enhancement (Main Supplement §IV.A.8), also known as hysteresis, which degraded photometric accuracy around bright regions such as the Galactic plane at 60 and 100 µm; variations in detector responsivity and electronic offsets that produced prominent striping in the SkyFlux images; and spatial and temporal variations of the observed zodiacal foreground producing steep, artificial gradients in the SkyFlux images, which obscured faint sky features and prevented co-addition of the individual HCON images. Finally, the 2' pixels of the SkyFlux images just critically sampled the resolution of the time-ordered detector data for the 12, 25 and 60 µm bands, making interpretation difficult without further interpolation.
The combination of all improvements reduced the residual stripes to the level of the intrinsic detector noise and largely eliminated interference from the zodiacal foreground. The removal of the zodiacal foreground emission increased the sensitivity over the SkyFlux images by roughly a factor of five. The destripers reduced the detector-to-detector noise by factors of 2-3 at 12 and 25 µm and 1.5-2.0 for 60 and 100 µm. This results in images with similar noise in the in-scan and cross-scan directions. Coaddition provides an additional factor of sqrt(3) improvement over individual HCON images. The co-added images reveal faint structure at 12, 25 and 60 µm totally invisible in the SkyFlux images. Details of the quality of the ISSA images will be found in Chapter IV.
A.1 Improvements in Relative CalibrationWhen looking at uniform sky, equal-sized detectors within a band will give different measurements due to variations in detector baselines and responsivities. With perfect calibration, these variations are removed and images of the sky appear uniform. Any imperfections in calibration result in detector-to-detector striping in the images. Calibration enhancements for ISSA, described in §III.A.2, reduced the detector-to-detector stripes by roughly a factor of ten at 12 and 25 µm relative to the calibration used in the SkyFlux images. No calibration changes have affected the IRAS point source calibration.
A.2 Zodiacal Foreground Removal to Permit CoadditionA foreground predicted by the zodiacal emission model described in §III.C.2 and Appendix G was removed from the time-ordered detector data, permitting useful co-addition of the individual HCON images. The subtraction of the zodiacal model resulted in a five-fold or better reduction, compared to the non-zodiacal-removed data, in gradients and artifacts due to changes in zodiacal foreground during the IRAS survey. However, some effects of the zodiacal foreground remain in the data. Since the zodiacal emission model is not perfect, insufficient foreground was removed in some places and too much was removed in others. Residual foreground removal errors for ||>50° are 3-5% of the original background, 0.5 MJy sr-1 at 12 µm and 1.0 MJy sr-1 at 25 µm over scales of 10°. For 50°>||>20°, the residuals are 1.0 MJy sr-1 at 12 µm and 2.0-2.5 MJy sr-1 at 25 µm over scales of 10°.
The zodiacal emission model assumed a physical dust distribution which did not include the dust bands. The dust band emission remains in the data and produces artifacts in the images at low ecliptic latitudes, the ISSA Reject Set.
A.3 Destripers to Stabilize Detector BaselinesStripe noise due to residual baseline fluctuations, residual zodiacal foreground and uncalibrated responsivity variations was reduced with two destripers. The first destriper globally compared all of the survey data at 1.2 million points on the sky (Emerson and Gräves 1988). At each point, the global destriper attempted to match each detector to the average of all other detectors in the same wavelength. This was accomplished by applying a slowly varying baseline correction to every scan of every detector. The assumption of global destriping is that the average of all IRAS measurements, after zodiacal foreground removal, of a particular point on the sky is the best estimate for the brightness at that point. The second destriper, known as the local destriper, used a similar comparison of each detector to the average of all detectors to make further baseline adjustments. This destriper used only the data in a single 12.5° field. The local destriper was able to accomplish about a 10% reduction in cross-scan RMS left after global destriping. The two destripers are described in §III.C.3.
A.4 Oversampling to Improve the Representation of Spatial InformationThe 1.5' pixel spacing in the ISSA images improves the sampling interval by 25% over the SkyFlux images and obviates the need to further smooth the time-ordered data.
A.5 Improved Pointing InformationImprovements in the pointing reconstruction for the IRAS survey contribute slightly to improved resolution in the ISSA images (§III.A.1).
A.6 Particle Radiation RemovalSignal processing on board the IRAS satellite attempted to remove from the IRAS detector data noise spikes due to high-energy protons and electrons. However, many small spikes and vestiges of large spikes remained in the SkyFlux images. A deglitcher removed this noise from the ISSA images (§III.A.3).
A.7 Known Asteroid Removal from the Coadded ImagesKnown asteroids listed in the IRAS Asteroid and Comet Survey (Matson 1986) were removed from the data prior to making the co-added images. This eliminated a major contributor of nonconfirming sources (§III.C.4) in the co-added images. Asteroids remain in the individual HCON images.
A.8 Full-Sized DetectorsWhen flux measurements are converted to surface brightness, point sources measured by undersized detectors appear too bright. Thus, to eliminate photometric problems associated with combining different detector sizes, only the operative, full-sized detectors (Table II.A.1) were used in making the ISSA images. The SkyFlux images used 3/4-sized as well as full-sized detectors.
|12||23 24 25 28 29 30 48 49 50 51 52 53|
|25||16 18 19 21 22 40 41 42 43 44 45|
|60||08 09 10 13 14 15 32 33 34 35 37|
|100||01 02 03 04 05 06 07 56 57 58 59 60 61|