Determining the nominal roll angle for stable intervals
The spacecraft time boundaries for intervals (usually several obis in length) of constant nominal roll angle (srai) should be determined.
We found the necessary information in a file 'rh*.ao', which has a binary extension table for each obi. The primary header contains the nominal roll angle for the observation. In the first table header, you will find the nominal roll angle for that obi (which should be the same as for the field). However, successive tables may, or may not have the same nominal value. Since nominal values should change by at least one degree (it may be 3 degrees, which is what we found), small differences (< 0.1 deg) are of no consequence. The table headers also give min and max values of ASPECT_ROLL. Normally, when these values differ greatly during an obi, it is likely that the pointing drifted off and those affected data were rejected. If in doubt, you can examine the table itself, and evaluate ASPECT_ROLL which is the instantaneous offset from the observation nominal. Note that nominals are given in radians; offsets in units of 0.5 deg.
The necessary information is packed in the `rh*_anc.fits' file. When unpacked with the PROS tasks 'rfits2pros' or 'rarc2pros', the actual roll is given in 'rh*_asp.tab'; but this is a long table and it is not normally necessary to follow the details of the roll angle. The relevant quantity of interest is the nominal roll angle for each obi, given in the file, 'rh*_obi.tab'. Alignment of obis
It has been well documented that some observations suffer from aspect solutions that jump back and forth between two solutions, and that such behavior is often associated with the change of obi. Unlike the question of determining the nominal roll angle, we know of no a priori indication of this problem (although the guide star configuration may be the cause, or at least associated with this condition): it is necessary to examine the source position for each obi. For your convenience, we have provided a script to do this, 'obidiv' (see 'Additional Tools/Obi Analysis', below). In some cases, you will find that the effective PRF for each obi is quite good, and that a simple shifting of obis and restacking (qpshift followed by qpappend) will achieve a marked improvement in the final PRF without doing any 'de-wobbling'.
It has been seen that an inadequate acquisition of guide stars can cause a significant degradation of the aspect solution. When only two guide stars are followed by the star tracker, the aspect solution is usually poor. It has been noticed that even when the star tracker has acquired three guide stars, the observation can still have a poor aspect solution. Therefore, we recommend that the observer performs a preliminary analysis of the guide star configuration. The guide star record is at the end of the 'anc file'. It can be viewed with 'fv' (ftools). You will find 7 items. The last table (item 7) is called 'GSTAR' with 1 column and some number of rows. The first part gives the space craft clock time, then there is a digit e.g. '9', and then follows the number of each star used (e.g. 78174945 9 9200090 9090110 9270130 0 0 0). It is thus possible to define intervals when only 2 guide stars are present, and deal with them separately or delete those time intervals. The same information is unpacked by rfits2pros, producing a file, rh700881n00_gst.lst. Although this file ends in '.lst', it is actually a table file, and so can be viewed with 'tprint'.