Swift UVOT UV-grism wavelength accuracy in the clocked mode

A Statistical overview of the wavelength calibration accuracy

The errors in the wavelength calibration are summarised on this page, while the verification details using the individual calibration observations are on subsequent pages. The error analysis is based on grism plus lenticular filter observations.

For version 1 a detailed analysis was done taking the new calibration file and reanalysing the data, measuring the line positions, and comparing to the correct positions. For the version 2 (June 2013) update, the anchor positions and dispersion relations from the calibration file were compared to those measured directly from the observations.

If further updates were made they will be mentioned at the bottom of this page.

Version 1 wavelength calibration (2009)

../_images/uvclocked_anchor_accuracy_w_ins.png

Figure 1: The wavelength scale errors have been split in two parts. The first is the offset of the whole scale due to errors in the anchor position and is shown as a histogram in the inset top left. The second is due to errors in the dispersion and is shown as a histogram in the top right. The data are based on the calibration spectra taken at locations all over the detector. The locations are shown in the main graphs by their anchor positions as blue dots. The error in anchor position is smallest near the default location for placing targets and largest near the edges and are shown as contours labeled with the offsets in Å.

The inset in the top left corner shows that at most times the anchor point that was predicted using the lenticular filters taken just before and after the observation, for the first version of this calibration has a mean offset of 15.9 Å averaged over the calibration observations. One outlier shows that occasionally an even larger offset occurs. Ignoring this offset, the anchor is found to cause an error which can be as large as 33Å.

The contours give a better picture of where the calibration for the anchor position results in an offset in the wavelengths. In the corners and near the edges, only a few observations were available for the calibration, so the values should be taken as an indication of the size of the error, rather than an accurate value that can be used to improve the wavelength scale knowledge.

Looking at the second inset “accuracy”, and disregarding the systematic wavelength scale offset from the top left inset, wavelengths from the dispersion relation are found to typically fall within 14 Å (about 4 pixels) of the correct value. The “accuracy” is the RMS dispersion of the dispersion errors over the whole range as shown in the accuracy plots in more detail, and not the error themselves.

This figure shows the errors in the wavelength calibration. The blue dots in the main figure are the anchor positions (at 260nm in first order) of the calibration spectra on the detector. The contours are a fit to the accuracy of the dispersion expressed as the RMS error between measured and predicted wavelength for known lines in the spectrum.

The inset in the top left corner shows that at most times the anchor point that was predicted using the source position in the lenticular filters taken just before and after the observation has a systematic offset of 15.9 A. (This offset was corrected for in a final revision of the wavelength calibration file). Disregarding the systematic offset, the anchor position typically falls within 23 A (about 7 pixels) of the mean position. This is sufficient to avoid making misidentifications of the larger spectral features which have a FWHM which is comparable in magnitude.

The inset in the top right corner shows the accuracy of the adopted dispersion from the scaled zemax model after removal of the anchor point offsets. The histogram shows the root mean squared measure. The contours in the main figure show the overall variaton of dispersion accuracy over the detector. The main result is, that the RMS error in the wavelengths in a spectrum is typically less than 16 A, which is about 5 pixels. The variation of the accuracy of the dispersion relation over the detector can be seen from the contour plot below. The contours are the RMS values of the wavelength errors of known lines in the calibration spectra. There is only a small variation over the face of the detector present, with the largest discrepancies in the top left and bottom right corners.

Detailed plots of the wavelength accuracy and of the spectra used

In order to allow better understanding of the capabilities, the data obtained by verification of the calibration spectra with the calibration are provided, as well as plots of the accuracy of fit for the individual observations. See the detailed spectra and accuracy plots.

Version 2 (June 2013)

The correct Zemax model parameters were used. The observations were reprocessed since the software had much improved since the last version was made, and stored in a Python dictionary. This allowed a much more straightforward check of the resulting calibration file and intermediate products. Since there was a remaining mismatch to the scaled Zemax models at long wavelengths, and to some extent at the shortest, a second term was added to the scaling of the zemax model. That term was held constant over the whole detector, being small and difficult to determine to a larger accuracy with the currently available data. The zemax model was scaled in two steps, first a large scale factor to all model positions that brought the model very close to the observed dispersion at the default (centre) position, and with a small positional shift to line up the observed drop in sensitivity with the model (as discussed in the flux calibration). The differences between observed anchor position and intermediary model were next fitted to a bispline of order 1(in x)+2(in y), and used to derive corrected anchor positions for the zemax model. The dispersion scaling relative to the anchor was next addressed. The initial scaling of all positions had already given a better dispersion in the intermediate model. A further (dispersion) scale factor was derived which varies slowly over the detector and is modeled by a bispline function. Over most of the detector that correction is within 2%. For each model anchor point one scale factor plus a small higher order term were then used to scale the dispersion. The number of adjustable parameters is much smaller than the observations used. Therefore the new calibration can be used with the calibration observations to get an estimate of the accuracy of anchor position and dispersion.

../_images/anchor_accuracy_20130613.png

Figure 2: The X, and Y are the measured offsets between the correct anchor position and the position extracted from the calibration file. The offset in pixels was multiplied by 3.2 to get a wavelength offset estimate for this plot. Using the typical angle of the spectrum on the detector image, the wavelength offset in the dispersion direction due to the error in anchor position was determined (dark blue).

Although the range of offsets is similar as in version 1, more observations lie very close to the correct anchor point.

Within the range up to 4000Å, we find that the mean error of the offset (due to anchor misalignment) is 4.5 Å, while the standard deviation of wavelength errors is 16.7 Å, which are the numbers to compare to the version 1 calibration. The errors are not seen to be larger than 22 Å.

Updates to the calibration

Due to an error in the Zemax model generation used for the first version of the wavelength calibration, the dispersion was not very good at long wavelengths. This has been corrected in Version 2 (June 2013) of the wavecal file. The error analysis of Version 1 above is still valid.

../_images/wavelength_error_20130613_a.png

The error in the wavelength (due to dispersion) sampled over all calibration observations covering the detector. One outlier has been clipped.

For each calibration observation, the difference in wavelength was derived using the correct dispersion for the observation, and the one derived using the calibration file. One outlier was removed (with errors up to 200Å). The errors are dominated by those at large wavelengths,

The total error is a linear combination of that from the anchor and dispersion, so lines may in some cases be displaced by ~40Å.

Technical Documents

  • Swift UVOT Calibration report. Zemax optical models for the UV-Grism study: Zemax pixel scale factor and positioning. MSSL 2008.
  • Swift UVOT calibration report: Zemax optical models for the uv grism study. Second order wavelength scale for sources near the axis.