IRAS Explanatory Supplement
V. Data Reduction
D. Point Source Confirmation
D.2 Overview of Seconds-Confirmation
D.2.b Position Reconstruction
Table of Contents | Index | Previous Section | Next Section
Detections in a single band were processed in time order. Each detection had its position in the Sun-referenced coordinate system computed. At the same time the photometric uncertainties, modeled as a white-noise Gaussian random variable in the logarithm of the flux and a constant but unknown calibration scale factor error in the flux, were obtained from lookup tables as functions of signal-to-noise ratio and correlation coefficient.
The position reconstruction was performed by searching the pointing history for records with time tags bracketing that of the detection. The matrices for transforming from the focal plane coordinates to the Sun-referenced system and to the 1950.0 mean ecliptic system were obtained by cubic spline interpolation. Each detector was characterized by a unique unit vector in the focal plane coordinate system. The transformation matrices yielded the polar, azimuthal, and twist angles defining the position of the detector slot on the sky in the Sun-referenced and 1950.0 mean ecliptic systems. The cross-scan half-width of the slot defined the uniform error component, denoted Lz, in that direction. The cross-scan limit cycle reconstruction uncertainty was modeled as a Gaussian random error, z. These and the in-scan errors discussed below were the only significant uncertainties involved in the confirmation decision process. After all confirmation and refinement processing was completed the absolute position angle errors, which were estimated by the pointing reconstruction program but which cancel out of the confirmation decision, were taken into account.
The uncertainties in the scan direction which affected the confirmation decision were a time-interpolation error and the detection timing uncertainty. The latter yielded an angular uncertainty when multiplied by the scan rate. These were taken to be independent Gaussian errors with a combined effect of a zero-mean Gaussian random error, denoted (X)y, with a variance equal to to the sum of the two error variances.
Flux and uncertainty, time and uncertainty, signal-to-noise
ratio, detection correlation coefficient, and detector number.
Telescope boresight angles in the Sun-referenced
system and uncertainties, scan rate, time tag, sines and cosines
of the angles in both the Sun-referenced and 1950.0 mean ecliptic
systems, and rates of these angles, all sampled at one-second
Status (on/off) and geometrical models in image space.
Point-spread function model.
Model of pointing reconstruction error due to time
interpolations; photometric error as a function of signal-to-noise
ratio and correlation coefficient for each band.
Confirmation acceptance, search windows, etc.