How do hot and warm pixels affect line fitting?#
As of February 2009 warm pixels account for around 11 % of the EIS CCD pixels, while hot pixels account for an additional 0.6 %. Hot and warm pixels, along with cosmic rays, saturated pixels and pixels affected by dust absorption are flagged by EIS_PREP as missing pixels and the recommendation from the EIS team up to now has been to not include them in your data analysis.
The number of warm pixels is now so large, however, that inspection of images formed from Gaussian line fit parameters (intensity, width, velocity) reveals many 'streaks' that result from distortions of the Gaussian fits due to the neglect of pixels in the line profile. The following questions thus arise:
- is simply ignoring warm pixels the best way of dealing with them?
- to what extent are derived line fit parameters affected by warm pixels?
- is there a critical number of warm pixels for which accurate line fit parameters will no longer be accurate?
A report has been prepared to investigate these issues and is available through the link below as a pdf file.
The following recommendations are made:
- Scientists should always use interpolated intensities and errors for missing pixels when analysing EIS data. Significant errors may result if missing pixels are simply ignored, particularly as the number of warm pixels on the detector increases.
- The method for interpolating the intensity of missing pixels in EIS_PREP should be modified.
- The error bars assigned to interpolated data points should be estimated through comparison with non-missing pixels, but modified according to how the interpolated intensity value was derived.
- Accurate re-creation of line fit parameters appears to be possible even up to the case of 30 % warm pixels. This corresponds to a lifetime of around 6 years considering the current rate of increase of warm pixels on the EIS detectors.
The method described above has been incorporated in EIS_PREP as an option: it is called through the /refill keyword.