It was then believed, wrongly as we know now, that there were few if any sharp features in the corona. (This was due in part to the poor off-axis MTBF of the soft X-Ray imager on Skylab, and partly to the low spatial resolution of coronal imagers up to that date - 1991.) It was also believed that corona changed only gradually, although I cannot imagine why anyone thought this. Accordingly, the coronal imager on SOHO was only supposed to produce a few images a day, in its role as a finder telescope for the spectrographs.
After the remarkable soft X-Ray images from the Yohkoh SXT began to appear, it was obvious to solar physicists everywhere that not only did the corona contain very sharp structure indeed, but that the corona varied on every time scale on which it had been measured. Unfortunately, very little in the design of the EIT telescope allowed for rapid operation: there is no heat path out of either the sector wheel or the shutter motor other than via diffusion through the telescope structure itself. Since the front of the telescope is heated by the Sun, that leaves the back end of the telescope: that is, the CCD camera itself and its radiator. This is an extremely inefficient way to cool these motors and worse, there are no thermistors on the motor housings themselves, so when the the CCD cold finger temperature is elevated by a fraction of the degree Celsius, we have no way of knowing how hot the motors have become. There is a real concern that we may be causing the Demnum lubricant in the motors to creep out of the bearings every time we run a sequence of normal exposures (i.e., using the shutter) every 1 - 2 minutes.
It should be pointed out that the spacecraft engineering team responded heroically in finding unused telemetry bandwidth to boost the LASCO/EIT share to approximately 7.9 kbps from its original 5.2 kbps (of which 1 kbps was to go to EIT) in submode 5. This allows a CME watch at 12 m cadence at half resolution (ffhr). In sudmode 6, which unfortunately deprives us of the opportunity of co-observing with SUMER, the LASCO/EIT bandwidth goes up to 15.8 kbps. (Note that these figures do not simply add up to the nominal LASCO/EIT bandwidth plus that "lost" by SUMER, as the frequency of the packets polled from each instrument is tightly constrained.) For more information on SOHO telemetry assignments, see:
Another, obvious constraint on EIT image cadence is due to the speed of the LEB hardware. Using Rice (lossless) compression, it takes approximately 6 minutes of CPU time to compress a full-field, full-resolution (fffr) image and move it into the telemetry buffer. The lossy compression schemes (ADCT and h-compress) tried so far have not produced noticeable improvements in those times without introducing unacceptable artifacts into the images. (This is due to both the dynamic range and sharpness of the features in the images, as well as the size of the pixel blocks [32 x 32 pixels] on which the compression algorithms act.... but that is a separate discussion.)
Subfield images take less time to process, though there is some overhead that makes the time saving less than linear. In general, however, small subfields (5 x 5 pixel blocks or less) are generally not CPU-bound; that is, EIT can obtain them as often as exposure time, shutter reliability, and thermal concerns allow.
The highest cadence possible with the shutter is approximately one image every 35 s (allowing normal shutter settle times, &c.), though we have begun to notice shutter irregularities when running this fast. At that cadence, we can obtain subfields up to 5 x 5 blocks in size.
LASCO also runs polarization and color calibration sequences daily/weekly that reduce the total number of instances in an EIT CME watch. Similarly, we obtain photometric response (degradation) measurements weekly, and participate monhly in intercalibrations (IC 001) with CDS, TRACE, and sometimes SUMER.
B. Instrument constraints
1. Thermal constraints
See section I.A, above for a description of the
problems with the EIT thermal design.
In practice, we have seen the cold-finger temperature begin to rise after < 1 hour of high-cadence (< 5 minute), and so limit the length of such "high" cadence image sequences to 1 hour. More modest cadences, even with much larger subfields, are still possible (e.g. 8 x 8 or 10 x 10 block subfields every 8 - 10 minutes). Recent trends, however, indicate that even the 6-minute cadence achievable for fffr's in short-exposure sectors (171 and 195 Å) when LASCO is idle (e.g. when the LASCO doors are closed for spacecraft thruster and CDS door ops) may be causing temperature creep in the back end of the telescope. We will watch this trend carefully to determine if we need further to constrain the cadence of observing.
Once the telescope has become heated by the frequent shutter operations, it takes many hours (~ 6 - 12) for the telescope to equilibrate again, so we run only one such hour per day.
Given the thermal constraints described above, we generally limit high-cadence (defined for now as δ t < 6 minutes) observations to one contiguous hour per day.
b) shutter
As noted above, a 35 s cadence is apparently the fastest we can operate the shutter, but this figure may have to be revised upward given recent experience. The EIT shutter has already operated several hundred thousand times, and including ground testing, is nearing its rather modest design life of 1 - 2 million operations.
c) LASCO synoptic plan
As noted above, the regular LASCO synoptic plan is driven by the requirements of amassing CME statistics with instrumentation of unique dynamic range and angular coverage. While the LASCO team has proven extremely flexible in allowing us to schedule high-cadence, subfield observations, each hour we do so represents the loss of an hour's worth of synoptic data for both C2 and C3. It is clear that while we can do this for some number of days per year, we cannot do so indefinitely without seriously aliasing the LASCO measurements.
d) Planning
The regular LASCO synoptic plan/EIT synoptic + CME watch has allowed a great simplification of the LASCO/EIT planning process that has in turn allowed reduction of the ground operations staff for the two instruments from three to two full-time personnel. Any less regular or more varied plans put additional stress on the only two people who really know how to operate the instruments; this is not a negligible consideration, especially as current funding trends and the start of new work on STEREO at the LASCO home institution do not make it likely that the level of trained, capable staff will increase.
Clearly this mode mixes great opportunities with significant risk: we can obtain good time resolution (as low as 13 s for 171 and 195 Å, though only over limited fields of view (still with the normal tradeoffs), but only for as long as the telemetry buffer is not full (40 minutes to 1 hour, for the subfields used so far).
Once the telemetry buffer fills, (i) the cadence is unpredictable, and (ii) the LEB cannot be interrupted in case of emergency --- very risky. The risk is posed by the fact that the shutter is open for that entire time.... and closing the shutter "manually" is not an operation that is performed all that often. A flare in 304 Å, in particular --- anywhere on the unblocked ~ 2/3 of the detector, not just the area read out --- could cause damage similar to or greater than that incurred in 1996 July, when the shutter inadvertently was left open for several hours and an active region near the limb left its signature on the CCD. The published literature, and back of the envelope calculations, indicate that a ~ M1 flare results in the enhancement of 304 Å emission by a factor of > 1 million during the impulsive and early gradual phases- perhaps 10 - 20 minutes. We really don't want to find out what the long-term effects of such an occurrence would be, and so we use shutterless mode sparingly, and only when solid, scientific justification is present.
Thus, while we have made a conscious decision to accept the risk of such an occurrence, and the consequent, drastic heating of the sector motor and telescope, four times a day for synoptic observations (in which the sector wheel movements occur at 6-7 minute intervals), we run high cadence, sector-wheel-changing sequences only with extreme reluctance, and then only when the EIT planner, the investigator, and a LASCO/EIT ground ops person are present, to insure the best chance of quick recognition and correction (LEB reboot) of such a problem. This means that such programs can only be run during normal, GSFC working hours and during near-realtime contact with the spacecraft via a ground station that is reliable for uplink (i.e., not DSS-27).
Note also that a LEB reboot requires several hours of LASCO/EIT operator work and lost synoptic data.
Fortunately, we have a backup in the IMOC: the Flight Operations Team has been able to monitor for the existence of a sector wheel hang since late 1997/early 1998. This is of limited help, however, if the sector wheel hang occur while SOHO is out of contact with a ground station (as has happened once, fortunately not too long before the next ground station contact).
EIT is the currently the only instrument capable of obtaining full-disk (and well above-limb), EUV images of the Sun --- the TRACE mosaics do not represent the appearance of the solar corona at any given time, but rather a composite of many different snapshots, taken over at least 45 minutes. We have been obtaining a unique, synoptic set of observations since 1997, and extended time series (either ffhr or fffr, depending on available telemetry) of CME watch images. These utilize EIT's unique capability as a full-disk imager.
TRACE and EIT are complementary: higher spatial resolution, higher cadence, limited field of view, sometimes frequent changes of targets vs. lower resolution, lower cadence (except for large areas), full-Sun field of view, most time spent on whole-Sun observations. There are exceptions to both of these characterizations (e.g. TRACE limb sequences during eclipse periods or EIT shutterless operations), but they are in general a simple fact of the design goals and limitations of the two instruments. It clearly does not make sense to operate either, much less both, of these instruments in extremely non-optimal ways for extended periods of time.
We have, therefore, some non-hard-and-fast groundrules for EIT observing: