While geostationary satellites are the ideal, and obviously-correct answer if orbital altitude is permissible, they don't work so well in higher latitudes.
For these there's a less-stationary but still viable option: a highly elliptical inclined orbit such as a Molniya Orbit or Tundra Orbit, which are designed to give a high dwell time over the area of interest.
This will appear to trace a "γ" gamma-shape in the air, slowing down to essentially stationary and then reversing in the loop:
It spends 2/3 of its time in the small eye of that tail - for the geostationary Tundra orbit, that's the 18 hours you asked for.
With two or more satellites following this same orbit (a "constellation"), you get essentially constant coverage.
If, because of angling of solar panels or something, the satellites are only visible at certain times, such as at the apogee (the very tip of the gamma tail) they can then essentially look like a single stationary object, that periodically blinks out briefly and then turns back on (slightly to one side of where it turned off, but you'd have to be very accurately monitoring it to notice that).
The requested gap of a few hours could either be due to a gap in the constellation, or because to be visible they require the sun to be shining on them, and they are in the earth's shadow at that time.
However, for these to be non-technological would be a stretch. A highly elliptical orbit is feasible though unlikely for a single object, but multiple objects in a constellation, not so much. So, the tricks to make it seem extremely stationary won't work.
Against the sun or stars, though, a single object in a Tundra orbit would appear essentially stationary, rising, hanging there, and setting at the same horizontal position.
Without solar panels, it'd need to be very high albedo - clean white or perhaps crystal?