When Hinode goes into eclipse, the pointing mode changes from gyros-plus-UltraFineSunSensors to gyro only (because it can't see the Sun any more!). In this case, there was prediceted to be a drift in the pointing. This gets corrected some time after Hinode comes out of night.

In the one case i've looked at where a mode change occurs (from gyro to gyro+ufss), the repoint takes of order 15 seconds. This was taken from the shift seen in 20070509_074627, a HH---N03 raster, with nominal 5 second cadence (more like 6-and-a-bit). This repoint happened between 07:53:27.275 (expo start time, admittedly) and 07:53:47.117 (expo start time). The corresponding NGT_EXIT happened at 07:45:30, so there is just over 8 minutes until the repoint due to mode switch. Since the few datasets I've looked at so far (all from the same date, from the merged telemetry) indicate that EIS twilight isn't over until about 10 minutes after NGT_EXIT, this repoint delay is not such a worry for us. The bigger worry is the length of night, including twilight, which cannot easily be calibrated out.

Nights typically last 16 to 17 minutes in the EUV (no transmission), with the total time that obscuration effects happen (either twilight or true night) clocking up 23 to 24 minutes per orbit (or a quarter of one orbit). That's now, too. It'll be interesting to see how serious it gets in the deepest part of eclipse season.

For now, I'd recommend leaving 10 minutes on either side of the NGT markers in the OBEV file if you want untainted data, or just steering clear of data within 10 minutes of those markers.

One use for these data might be to use them as dark current calibration. There is effectively no input to the detectors (unless the 3p-1s transition is strong in geocoronal emission) so anything we see will be stray light.