1. IBB Calibration Update J. Rufus GIST21 Meeting Boulder, CO. 31 st March – 2 nd April 2004 © Imperial College London

2. IBB Calibration Update Ideal Method - All Heights, near Instantaneous © Imperial College London Measure a near instantaneous average V CBB and record IBB data while observing WBB. (Method used for GERB-3.)

3. IBB Calibration Update Ideal Method - All Heights, near Instantaneous © Imperial College London Measure a near instantaneous average V CBB and record IBB data while observing WBB. (Method used for GERB-3.) =Total Gain

4. IBB Calibration Update Method Used – One Height, Static © Imperial College London During GERB-2 IBB Calibration more time was devoted to SW IBB tests. Time used on these observations caused concern over the amount of data collected at 5 heights in region of rapidly changing T. Compromise employed where WBB and CBB were measured at one height (#3 Pixels 97-158) Use radiometric cal around IBB trials to provide gain and MODEL offset (measured from CBB)

5. IBB Calibration Update All Detector L IBB Data – Static Method © Imperial College London A surface plot of the data shows a noisy profile with occasional low ‘spikes’ Also the noisy pixel 123 is visible as an elevated ridge. Percentage of anomalous points (<100) 76 out of (256  7908)  0.004%

6. IBB Calibration Update Problems with the Fits © Imperial College London Over the temperature region for which we have data a linear fit is sufficiently accurate. The fit gradient is within expectations but the intercept gives rise to structure across the detector. Since T op  T cal we may also have problems with linear extrapolation

7. IBB Calibration Update Attempted Solutions © Imperial College London Truncate removes low T dominance Average combat noise Fit T 4 improve extrapolation

8. IBB Calibration Update Extrapolation to IBB Operating Temperature © Imperial College London Fitting becomes a problem because of the mismatch in temperature between the calibration and operating regimes.

9. IBB Calibration Update Detector L IBB Data after Averaging over  T=0.04K © Imperial College London A surface plot of the averaged data shows a profile with numerous ‘furrows’ indicating varying offsets between pixels The truncation of the bottom 25% of the data results in a loss of only 0.4K in temperature coverage. Altering fitting function will not remove structure in data.

10. IBB Calibration Update All IBB data for Height 3 © Imperial College London Static Method Instantaneous Method Ht 3 only

11. IBB Calibration Update CBB Measurements from Contemporary Radiometric Calibration © Imperial College London Observations of the CBB signal in the height overlap regions shows A bulk positive drift A very small uncertainty, < 0.05% Several anomalous pixels which deviate from the general trend Discrete jumps in the trends for individual pixels at random points

12. IBB Calibration Update Source of Fluctuation © Imperial College London V CBB s measured during Radiometric Calibrations either side of IBB Trials do not exactly correspond to those during. The magnitude of the effect explains the deviation of the two methods and the increased structure in L IBB in the full pixel method.  L =  V CBB / G ~ 0.4%

13. IBB Calibration Update Correction based on Average Difference Observed © Imperial College London

14. IBB Calibration Update Correction based on Average Difference Observed © Imperial College London

15. IBB Calibration Update Correction to Shape Over Height 3 © Imperial College London Correction factors influence on the smoothness of radiance profile: Black – Uncorrected Red – Corrected Blue – Instantaneous But this data includes pixel 123 which the correction smoothes significantly.

16. IBB Calibration Update Correction to Shape Over Height 3 - Revised for Pixel 123 © Imperial College London Picture still improving over the whole height range with 123 removed.

17. IBB Calibration Update Correction Over Height 3 – Data Dependence © Imperial College London Do better in regions of high measurement, corresponds to longer observation time thus more stable V CBB, and C

18. IBB Calibration Update Correction Over Height 3 – All T Mean © Imperial College London This is borne out by average over the whole data set along the temperature axis. Using the correction determined from comparison of the average of the IBB trial V CBB averages (taken over ~ 6 hours) and the average of the radcal V CBB s (taken in 4 2.5 hour windows over 3-4 days) an excellent correction is produced.

19. IBB Calibration Update Correction Over Non Central Heights? © Imperial College London GERB-3 fits show high degree of homogeneity. Use these as witness sample to correct GERB-2 heights outside height 3. Back up this approach by using data from moving table observing shroud. GERB-2 GERB-3