It's almost comforting to hear that the offsets on SRMs are affected by the temperature, as I've had that some issue with Powertaps and needing having to zero out when the temperature changes.
the SRM (like the quarq) uses strain gauges to measure the deformation in metal as a factor in the calculation of power. metal is affected by temperature and can affect zero offset if temperature changes during a ride (or, for example, if your bike is stored in a warm (cold) place and you take it outside to ride in cool (warm) temps).
it is important to note that while all systems are affected to a degree, the amount they are affected is quite important.
as a rule of thumb, for the SRM or quarq a 3-unit change in zero offset reflects ~1W @ 90 rpm. so, if you see the ZO changes by 5 before and after a ride, that is below the resolution of these units. on the other hand, if there is a 60-unit change (as can happen with some brands/types of power meters) that is a different matter and can cause problems with your data.
the real problem when you have a unit like that is you don't know if there is a steady drift throughout a session, or if it was +100 30 minutes in, -100 60 minutes in, then +60 at the end, for example.