2000-2005 Plate Scale Calibration Changes

In 2002, a new plate scale calibration was performed. That work is documented here. We have used the method laid out in that work to perform new calibrations using the data from 2002-2005. In brief, we performed a Powell minimization of 10 of the 19 calibration coefficients (leaving temperature fixed) using as our source data the expected and measured distance between available star pairs in science observations.

We came at this problem with two goals: 1) to track changes in the camera assembly calibration over time and 2) to compare the calibration presently used by ground pipeline processing tools to the present state of the camera.


Changes Over Time

To track changes over time, we generated a new "baseline" set of coefficients using the 2001 data. These coefficients are referred to as "2001newbase" in the tables below. For each year, the star distance deviations calculated with newly fit coefficients have been contrasted with the deviations consistent with use of the 2001newbase coefficients. The newly fit coefficients were used at an estimate of the ACA housing temperature for the year, and the 2001newbase coefficients were always used at 14.5 deg C.

2000 : Temp = 14.15 deg C

2001newbase coeffnew fit coeff
CHISQ/N= 0.0542CHISQ/N= 0.0470



2001 : Temp = 14.75 deg C

2001newbase coeffnew fit coeff
CHISQ/N= 0.0565CHISQ/N= 0.0565



2002 : Temp = 14.75 deg C

2001newbase coeffnew fit coeff
CHISQ/N= 0.0711CHISQ/N= 0.0695



2003 : Temp = 16.35 deg C

2001newbase coeffnew fit coeff
CHISQ/N= 0.0916CHISQ/N= 0.0795



2004 : Temp = 16.55 deg C

2001newbase coeffnew fit coeff
CHISQ/N= 0.0735CHISQ/N= 0.0685


2005 : Temp = 16.75 deg C

2001newbase coeffnew fit coeff
CHISQ/N= 0.0940CHISQ/N= 0.0866



Conclusions

While there have been measureable changes in the camera as seen by fitting new plate scale coefficients, even the largest expected offsets at the corners appear to be less than .30 arcseconds when the data from 2005 is compared to the data from a baseline generated by the 2001 data. These changes are to be expected due to the gradual increase in ACA housing temperature.


Evaluation of Current Ground Processing Calibration

Here we have compared each new coefficient fit by year with the coefficients currently in use by the ground processing system. Each calibration was redone using the ground processing coefficients as the seed points. For each year, the ground processing coefficients and the newly fit coefficients were used at an estimate of the ACA housing temperature for that year.

2000 : Temp = 14.15 deg C

ground process coeffnew fit coeff
CHISQ/N= 0.0503CHISQ/N= 0.0470



2001 : Temp = 14.75 deg C

ground process coeffnew fit coeff
CHISQ/N= 0.0578CHISQ/N= 0.0565



2002 : Temp = 14.75 deg C

ground process coeffnew fit coeff
CHISQ/N= 0.0714CHISQ/N= 0.0698



2003 : Temp = 16.35 deg C

ground process coeffnew fit coeff
CHISQ/N= 0.0865CHISQ/N= 0.0792



2004 : Temp = 16.55 deg C

ground process coeffnew fit coeff
CHISQ/N= 0.0950CHISQ/N= 0.0682



2005 : Temp = 16.75 deg C

ground process coeffnew fit coeff
CHISQ/N= 0.1122CHISQ/N= 0.0866


Conclusions

The evaluation of the ground processing coefficients (compared to newly fit values) was designed to quantify the extent, if any, that aspect products might improve if newly calculated coefficients were used in this ground processing. The expected-measured distance plots and the vector difference plots do illustrate two important points: 1) as years pass from the creation time of the ground processing coefficients, greater improvement would be seen by implementing new coefficients (this can most clearly be seen by the difference in CHISQ/N for each set of plots) and 2) the current ground processing coefficients are still mostly within .20 arcseconds of "ideal" new coefficents even for the 2005 data.





Aspect Information main page



Comments or questions: Aspect Help

Last modified:12/27/13