From my personal experiences, I have found the same phenomenon occuring for months after the last time I changed a component in the water chain and in doing so, introduced visible air into the system. After you remove the visible air from your system and top it off, there is still air in your system. My famous "gut feeling" is that it exists in two components.
The more obvious is pockets trapped in places where you can not see them, like the two humps on top of the typical chevette heater core. Unless you pick up your PC and rotate it it in all planes while it is running, chances are you will have these remote pockets. Water absorbs air. If it didn't, the fish would all die. That air, though invisible, when suspended in water adds (minutely) to its volume. As your water trap percolates out the air, the water picks up more air from those pockets and the pocket size decreases. That process continues until the pockets are absorbed and replaced by water. That process makes your level go down.
Once all the pockets have been eliminated, you still have the air, suspended in the water, to get rid of. Depending upon the temperature, altitude and presence of other chemical compounds in the water, that process is a slow and subtle one. Eventually, your water reaches equilibrium for your altitude and temperature and no more air will come out of it. At that point, as I said, some times months later, the rate of water drop will slow considerably.
Unfortunately, we rarely leave our system undisturbed long enough for the water to reach equilibrium. I know I have only done so once or twice since I went to water cooling. As such, I try to check my water no less than once a week. If you use a T-trap and let the water level drop to where the flow sucks air back into the lines, you have to start the process all over.
That's my "laymans" spin on the issue of dropping water levels in a closed loop system.
Hoot