So, where is the Sun?

A coronagraph blocks out the disk of the Sun in order to observe the faint emission coming from the surrounding corona.  This means that there is an uncertainty in knowing exactly where the Sun is.  There is also an uncertainty in the value of the roll angle, the angle between the "top" of the image and solar north.  Operationally, we command the pointing legs on LASCO to move the entire LASCO optical box to make the corona appear to be symmetric around the occulter.  This can be done only for one of the three telescopes and we have selected C2 as the primary determinant of the pointing, because it has the most stringent pointing requirements.  The last time the legs were moved was April 1996.

To calculate the position of the Sun and the roll angle we use stars.  This is possible for each C2 and C3 image because each image has many stars.  (For C1 we must use the passage of Alpha Leo which occurs every August to determine the same information.)  An IDL program has been written to map the position of the stars and the Sun onto the focal plane of C2 and C3 for the SOHO orbit.  The output of the program allows us to determine the geometric distortion of the optics, the row and column position of the Sun center and the roll of the image with respect to solar north.   In addition, it provides an experimental determination of the vignetting function.

Here are plots of the Sun center and roll angle determined from star positions:



Also provided are plots from LASCO pointing eyes (BoreSighters and Pointing Error Sensors) and SOHO roll:

List of times when SOHO was rolled 180 deg: nominal_roll_attitude.dat


To determine the Sun-center, roll-angle, and correct observation time, several problems had to be solved. The most important problem was the effect of timing errors on the overall pointing determination. The final correction method addressed several difficulties, including knowing the Sun location, the roll angle, the time drifts, (sometimes up to 5 min), and pointing variations due to temperature changes.

The uncorrected absolute time accuracy has three different behaviors during the mission. From launch until June, 1996, the time required no correction. The second period, from June 1996 until the mission interruption in June 1998 showed steady drifts of up to 5 min until a command to transfer the spacecraft time to the instrument was sent. After the mission interruption the on-board time adjustment no longer worked.

Two methods were considered to correct observed time drifts. The first utilizes the calculated X-position of the Sun center from the stars and corrects the time to place the Sun-center at the proper location. The second method compares the time of the EIT housekeeping and science telemetry packets every 10 minutes during real time data-downlink to the ground receipt time. The difference between these times is the onboard time-offset within about 15 seconds. The packet time-offset calculations were not collected before September 1997, but from that time to the SOHO interruption (June 1998) the two methods agree reasonably well, in contrast to the post- interruption (December 1998 - Present) calculations. In this period the EIT packet time offsets are considered to be more accurate.

The final solution uses a method that combines the two methods to correct for both time-offset and pointing. Prior to June 1996, no correction is needed. From June 1996 to June 1998, the time correction is inferred from the C2 Sun-center positions. From September 1998 to the present the EIT packet time offsets are used.

The C2 and C3 pointing and roll corrections are calculated from corrected image times and star-position predictions. In the Level-1 image header, the original observation time (DATE-OBS), sun-center values (CRPIXi), and roll angle (CROTAi) are replaced with the new values. The image is not corrected for roll, except for rectification when SOHO is in the inverted position. The time difference between corrected and uncorrected times are kept in the header as history.

In the plots, you can see some gaps in the data.  The large ones beginning in mid 1998 and extending to early 1999 are due to the loss of pointing of the SOHO spacecraft and then later to the failure of the gyroscopes.  In this interval, the orientation of the spacecraft was not maintained with solar north "up".  This can be seen in the full scale roll angle plots.

From the analysis we find that the C2 and C3 have a slight roll offset from each other.  C2 is offset from solar north by +0.50 degrees and C3 is offset by -0.24 degrees.  Notice that there is a slight variation (<0.1 degrees) throughout the mission.

The plots from star positions show a lot of scatter in the X and Y positions of the Sun.  However the variations are so large that we must be very careful before believing them, since we know that the spacecraft doesn't experience such large variations, and our own pointing eyes (the PES and boresighters) and the images themselves don't show those excursions.  The small-scale variations may be due to uncertainties in the stellar ephemerides, inherent limitations in measuring sub-pixel position variations, or unaccounted-for instrument optical effects. On a larger scale, the data do indicate periodic, small excursions which are presumably due to thermal distortions in the instrument (not in the spacecraft).  We therefore use a 50-point median value for sun center and roll.

For February - April 1999, the temperature of the LASCO optical box was being cycled on a weekly basis to try to diagnose the problem with the C1 Fabry-Perot caused by the extreme cold temperature experienced during the loss of pointing.  This shows up as a pointing variation during this period.

In October 2010 the computation of attitude values was updated and a few errors were found in the software and corrected. While the effect the correction was very slight in C3, in C2 there was an offset of about 0.15 pixel in the X-cen value. Plots showing the overlap of the previous values and new values can be seen here.

Updated 10/19/2011 by NR