worked to accurately (~0.1 pixel) determine the final calibration details
of the ACIS-S geometry based on HETG data sets.
The result is
a new geom file for the ACIS-S chips,
http://space.mit.edu/ASC/docs/memo_geometry_acis_s_1.0.ps.gz ( pdf version ),
that includes slight (<0.02 degree) rolls of the chips about their normals and slight (~< 0.2 pixel) changes to the chip-to-chip offsets (CHIPY direction) and spacing (CHIPX direction, aka "chip gaps").
Here, this new geom file and recent LSFs were used in processing Capella data sets and measuring observed line wavelengths. Plots comparing the measured with "theoretical" rest values of wavelength were made. Variations in overall wavelengths between the approximately yearly observations are well correlated with the relative Capella-primary to Earth velocity, indicating very stable wavelength scales for HEG and MEG spectra.
Additionally we can compare the HEG and MEG wavelength scales and these appear to differ by 40.2 km/s (+/- 5 km/s); a slight increase of the MEG period --- from 4001.41 A to 4001.95 A --- is suggested to bring these wavelength scales into agreement.
The issue of setting the absolute wavelength scale (to better than its current +/- 100 km/s range) will require some decisions on which lines/blends to use to set the scale and their 'real' wavelengths; this is deferred to another time.
Line Wavelength Results
Many of the HETGS Capella observations were reprocessed using this new geom file, producing PHA files for inspection.
Custom ISIS s/w was used to fit the PHA files, including a nominal RMF, and output simple FITS files containing theoretical line wavelengths and measured line wavelengths.
These FITS files were then processed with an IDL routine to make plots and measure the measured-over-theory wavelength ratios for each grating (+/- orders effectively averaged) and observation. The table below summarizes the outputs and provides plots of wavelength "error" for the MEG and HEG spectrometers. Plus and minus order measurements are indicated on these plots by *'s and diamonds.
|Obsid||MEG ave offset|
|HEG ave offset|
|Sep 24 1999||1103||-169 , 'abslam.pdf||-40 , 'abslam.pdf||129|
|Mar 3 2000||0057||87 , 'abslam.pdf||193 , 'abslam.pdf||106|
|Feb 11 2001||1010||-6 , 'abslam.pdf||132 , 'abslam.pdf||138|
|Apr 29 2002||2583||35 , 'abslam.pdf||179 , 'abslam.pdf||144|
|Sep 27 2003||3674||-121 , 'abslam.pdf||56 , 'abslam.pdf||177|
|Sep 10 2004||5040||-65 , 'abslam.pdf||45 , 'abslam.pdf||110|
HEG-MEG Wavelength Scale Offset
The plot below shows the tabulated HEG and MEG wavelength offsets (in ppm) for the different Capella observations. It can be seen that the two offsets track each other pretty well indicating that this is a real observation to observation wavelength change. The effect of the Earth-to-Capella-primary relative motion is also indicated on the plot at the observation times and largely explains the variations seen. The second figure below shows the residuals after this expected Doppler offset of the primary star is removed. It is clear that there is a systematic offset of the MEG wavelengths compared to the HEG values and that the HEG values appears to agree well with the expected offsets. It would makes sense, then, to slightly adjust the MEG period to bring it into agreement on average with the HEG values.
Specifically, the MEG period is adjusted from 4001.41 A to 4001.95 A. Note that the amount of this change, 134 ppm, is very similar to the amount estimated in Marshall, Dewey, and Ishibashi ( SPIE 2003, astro-ph/0309114) of ~ 160 ppm. (The Capella velocities had not been accounted for in that paper and so, there, it was proposed to adjust the HEG period because the MEG values looked more absolutely accurate.)
This plot shows that the wavelength offsets of the MEG(diamonds, lowest curve)
and HEG("*") are very correlated from observation to observation -
suggesting a simple wavelength scale (or period) offset between them.
The dashed line ("+") data are the expected Doppler offset
due to the Capella-primary motion relative to Earth.
[ Available in full page pdf format as well. ]
Here, the average wavelength offset values have been corrected by the expected
Capella-primary relative velocity and expressed as an equivalent velocity.
[ Available in full page pdf format as well. ]
Absolute Wavelength Calibration
Given the spread in wavelength offsets of individual lines/blends, figure below, any further adjustment of the HEG-MEG wavelength scales for improved absolute wavelength scale is best held off until obtaining a community consensus on how the absolute scale should be set, e.g., which specific lines/blends are the "standards" for wavelength scale and their values.
This plot (others like it are linked in the table above) shows the wavelength errors
for the Obsid 1103 MEG lines measured; the pattern of deviations seen here
is similar in other observations. This indicates that the assumed "theoreticsl" wavelengths
of the measured lines/blends are not accurately know (at least to us!)
Determining an absolute wavelength scale therefore requires some descision on which
lines/blends are to be the standards and their rest emission wavelengths.
[ Available in larger pdf format as well. ]
For reference, the various software routines used here (but perhaps not all that are needed) are in the directory SWused, the principle routines are get_bish_data.i to read in the PHA files, bp_extract.i which calls bp_fit_counts.i to do the line fitting, and bp_examine.pro to make the summary plots. The routine start.i is loaded first to set up various infrastructure including "region" (rg_) and "bump" (bp_*) data structures. Again for reference, specific commands are:
unix isis isis> .load start.i isis> .load get_bish_data.i ; data set selected in the file. isis> .load bp_extract.i unix idl IDL> bp_examine, 'Bish04_bp/acisf01103_002N002', 'meg' IDL> bp_examine, 'Bish04_bp/acisf01103_002N002', 'heg'