Документ взят из кэша поисковой машины. Адрес оригинального документа : http://xmm.vilspa.esa.es/calibration/documentation/CALHB/node946.html
Дата изменения: Tue Sep 20 19:45:21 2011
Дата индексирования: Mon Apr 11 03:38:37 2016
Кодировка:
Поисковые слова: rainbow

Calibration Access and Data Handbook

Generation from In-Orbit Data

Ideally the in-orbit PixToPixSens file will be generated from internal calibration measurements using the fullframe high resolution engineering mode. However due to their large data volume the fullframe high resolution data won't be available frequently enough to base the in-flight generation of the PixToPixSens file on the high resolution data. Therefore a different strategy has to be adopted.

The SAS task omflatgen will be used to compute the PixToPixSens files from low resolution internal calibration exposures (tracking off, BPE on), which are predominately acquired during slews. The analysis can be outlined as

1. Multiple exposures may be co-added to achieve high enough statistics.
2. removal/flagging of bad pixels
3. calculation of average CCD frametime in the used exposures
4. calculation of average CFRR,
   i.e. the average counts per pixel per CCD-frame
5. fpn?
6. check dark counts and exclude bad pixel?
7. calculation of the local normalization mask(?)/value for each 8x8area/pixel
   e.g. by calculating the average within a sliding box of TBD width.
8. local image normalization
   by dividing the pixel content by the local average value
9. eventual exclusion of large excursion regions and recalculation of step 4 and 5
10. global normalization
    e.g. by division with the average of the local normalization values or by normalization to values of a reference area/pixel.

If the non uniformity of the calibration lamp illumination is known with sufficient accuracy, the inverse of the illumination pattern can be be used for uniformity correction and the normalization steps can be omitted in the data analysis. The non-uniformity of the calibration lamp illumination can be derived from a comparison of the smoothed high resolution flatfield with the large scale flatfield.

In case not sufficient high resolution flatfield data are available the low resolution data must be expanded into high resolution PixToPixSens data.
This can be achieved in three different ways, depending on the availability of data.

• If high resolution data are available and provide sufficient statistics, these data are used to calculate the PixToPixSens file. A pixel expansion is not required.
• If high resolution data are available, but the data do not provide sufficient statistics or they are out of date, each pixel of the low resolution data is expanded into four subpixels by applying a weighting factor to each quadrant of the old pixel. The weighting factor is derived from full resolution images.
• If no high resolution data are available or for the initial calibration procedure low resolution data are expanded into full resolution data applying a weighting factor of 1.0 to each quadrant of the old pixel.