spacer link to MAST page spacer logo image spacer
link to STScI page

TUES Data Reduction & Analysis

A description of the data processing procedure from the project is below.

ORFEUS II Echelle Data Reduction

By J. Barnstedt


1. Steps of data processing

A) Extraction

All orders were extracted automatically from the images, i.e. during extraction a centering in y-direction (cross dispersion direction) was applied according to the center of gravity for each individual echelle order. The extraction is done in x-direction (main dispersion direction) by summing up a well defined number of pixels in y-direction. The number of pixels used for extraction in y-direction varies with the echelle order and is indicated in the header of the data file for each echelle order ("cut width"). The center for the extraction in y-direction follows a straight line along each order and is located on whole pixel numbers.

Some echelle images show tilted absorption lines within the strip of the echelle orders. In such cases the extraction was done summing up the pixels tilted by 45 degrees, which results in a significant increase in resolution.

back to table of contents

B) Background

Between the echelle orders a line of 3 pixels width was used to estimate the background. With exception of orders 40, 41 and 42 the background was calculated as average of the strip above and below the corresponding order. For the first three orders only the background values below each order were calculated. The so calculated background was smoothed twice with a width of 21 pixels: First a median filter was applied and as a second step a boxcar smoothing was used. This smoothing works fine if the counts within the background pixels are not too low. For very low counts within the background field other smoothing methods might be more satisfying.

Within very broad absorption lines (e.g. Ly-alpha) the background subtracted is in general overestimated. The reason might be that by far the strongest contribution to the background comes from stray light of the echelle grating:
This stray light is scattered exactly in horizontal direction on the detector while the echelle orders run slightly tilted across the detector. So within very broad absorption lines stray light is reduced within the echelle orders, and almost normal in the background extraction fields.

back to table of contents

C) Pixels 511/512

Due to arithmetic rounding errors at the calculation of the photon position within the echelle electronics an artefact is observed, especially in the middle of the detector image, which leads to a higher intensity in one of these two pixels while the other shows a corresponding loss of intensity. This means, that the pixel border between these two pixels seems to be somewhat shifted. This effect is eliminated by applying an averaging to these two pixels.

back to table of contents

D) Correction of the blaze function

The efficiency of echelle gratings has a maximum for one direction of diffraction (blaze angle), while the efficiency is reduced as a function of the deviation from this angle of diffraction (blaze function). The optimized diffraction direction was pre flight adjusted to point to the center of the echelle detector. What we find now is that the center of maximum efficiency is different for each echelle order. Furthermore we found that both the position and the width of the blaze function differ from observation to observation.

Therefore it was neccessary to introduce an individual blaze correction for each observation. Often only the overlap region between two adjacent orders could be used as a criterion for a good blaze correction.

back to table of contents

E) Wavelength calibration

The wavelength calibration was calculated from the positions of 814 interstellar absorption lines from 12 different echelle images. Using these data we determined the parameters for the dispersion function. Radial velocity values were not used for the calibration, the zero position of the dispersion function was determined seperately. The accuracy is better than ±0.005nm, i.e. better than the optical resolution of the instrument.

We used the position of the Ly-alpha geocoronal emission line as a reference for the absolute wavelength zero position. We found that for different observation blocks the position of the Ly-alpha line differed up to 0.006 nm. For this a wavelength correction for each observation block was applied.

Further wavelength corrections are:

  1. Heliocentric radial velocity correction
  2. Radial velocity correction due to the satellite's orbital velocity.
The given extractions use the on board integrated images, for these images only the average of the orbital velocity component during the observation was used for correction. In future we will extract the spectra of the individually registered photons for which the orbital radial velocity correction will be applied to each individual photon.

An additional wavelength error might occur, if the target was not exactly centered within the 20" diaphragm. The maximum resulting uncertainty is ±1.2 10-4 as a relative wavelength error. This error was not corrected up to now.

To overcome this problem we will try to get a hint about the image position within the diaphragm from the count rates. Rapidly changing count rates indicate that the image was close to the edge of the diaphragm. In these cases, we can correlate the ASTRO-SPAS pointing data with the count rate fluctuations and thus find out the x/y-coordinates of the image in the diaphragm. Anyway this correction procedure will be complicated and tedious.

back to table of contents

F) Correction of loss of sensitivity in the detector edges

The corners of the detector image and the left edge show a loss of sensitivity which is probably due to a reduced efficiency of the repeller grid in front of the detector. The electrical field in front of the MCPs (about 50V/mm) is used to force those photo electrons back into the MCP channels, which are released from the areas in between the channels. This improves the quantum efficiency by about 30% (causing also a loss of 10% due to shading of the grid in front of the detector). Probably due to an inhomogenious field at the borders of the detector the efficiency of the repeller field is reduced, and there is a rather sharp step visible between lower and normal sensitivity. This is visible in some images as a circular shaped area. We estimated a loss of about 25% and corrected this by applying a "smooth" step function. The position, width and height of the step was estimated for each order from the sum of all echelle measurements. The actually used correction values are listed in the column "EDGE_CORR" (see below).

A detailed flat field correction was not applied. The reason is, that the optical light path of the spectrometer cannot be reproduced in our laboratory. This however would be essential for an exact estimation of the flat field behavior of the detector. Any other correction methods are too uncertain to be useful.

back to table of contents

G) Absolute flux calibration

We used a HST archive model of G191B2B calibration. The calibration was additionally checked with a model of BD +28°4211 (R.Napiwotzki). We guess an accuracy of ±10% for the flux calibration. This is valid, if the object was fully centered within the diaphragm. There are, however, some observations for which the object was not completely centered. The reasons are probably some temperature drifts of the telescope causing a shift of the alignment. In some cases also slightly wrong coordinates of the target might have led to a decentralization.

Observations with badly centered targets are identified by their strongly varying count rates. The flux calibration was calculated for the maximum observed count rate for the corresponding target (also from other observations of this target, if neccessary). This count rate was scaled with the registered count rate in the integrated image. The corresponding count rates are documented within the file headers.

The file headers contain a maximum count rate and an actual (average) count rate of the lower electronic threshold. The third value is the registered count rate in the actual image. The actual count rate of the lower threshold and the registered count rate differ for the following reasons:

  1. The electronic upper threshold suppresses pulses too large to be processed.
  2. Electronic dead time effects and dead time effects from the on board processor reduce the electronic efficiency.

The above page was written by Jürgen Barnstedt from the ORFEUS Group at Tübingen and is reproduced here for the MAST archive with his permission.