The plot below is from the question Why do the geosynchronous TDRS satellites have this distribution of inclinations?, and the explanation says
Inclination and mean motion (rev/day) for TDRS satellites from the first TLE found in the first week of each calendar year. Dot for TDRS13
The blips in mean motion are probably movement from one station to another, and the step downs (from 1.003 to circa 0.99 rev/day) are probably decommissionings to a higher graveyard orbit.
Interestingly, TDRS 8, 9 and 10 were inserted into orbits with substantial inclinations (6 to 8 degrees) that decreased with time over several years, before starting to increase again.
I'm guessing that the orbit doesn't "bounce" or suddenly change rate of inclination, but instead the way inclination is defined in the TLE it's probably always positive to save space (one less character) and the zero crossing shows up in another parameter in the TLE.
Question: How can I use more information frmo the TLEs to recover a sign for the inclination -- convert from 0 to 180 degrees to -90 to +90 degrees -- so that TRS numbers 8, 9 and 10 will cross zero circa 2010?.