The following JPEG contains my answer.
[Photo of AC unit 2 meters from rig]
I've seen this before in a different setting. It's to be expected unless extremely controlled exterior temperatures and constant or zero airflow are are used, by, for example, enclosing the exterior of the rig in a water jacket that is kept at as constant of a temperature as possible by a thermostat. Almost all labs have some kind of HVAC connection to keep the inside environment at habitable temperatures and humidity, and almost all labs have one or more exterior walls, windows, or roofs that couple the exterior environment, diurnal solar heating and night time radiation cooling, and convection into the rig's experiment room.
What can be very helpful is if the AC unit and its fan blower unit on/off times are registered in the collected data so that the obvious change in external air convection can be at least noted on the charts if not (through extreme calculation gymnastics) filtered away. One idea is to cool off or heat the room before the experiment starts, then turn off the HVAC for the data collection period, collection room temperature as part of the experiment, and leave the door closed with the room empty during the data collection period, so that only free convection and radiation would occur.
All this is a real pain in the behind and is essentially a type of experimental noise, lowering the signal to noise ratio. That is why the ideal experiments need signals so large that you can drive a 18 wheeled truck through it -- so that the estimation of the environmental effects can be grossly simplified with minimal calculation work, without significantly lowering the probability that the positive result was experimental error.