I'm guessing that rooftop is radiating heat from the attic for hours after the sun goes down, and with little or no wind to sweep it away, it is affecting the temperature reading.
I doubt that the shingles alone have as much thermal inertia as some of the rocks and boulders around my station have, but, yes, maybe the interior of the garage acts as a heat reservoir through the night to keep the ISS warm.
On the otherhand, maybe the air ~20 ft off the ground (rough height of the ISS there) is warmer than that closer to the ground. The night time outbound radiation here in the high desert on a relatively calm (less than 2-4 mph winds) is incredible. Surfaces such as the ISS's radiation shield become much colder than the surrounding air so fan-aspirated air is actually chilled passing over the shield. I've tested them with an IR thermometer. The bare sandy soil and even nearby boulders chill down rapidly on most nights
Then there is the possibility of night time, stratified air density currents bringing cold air down from the mountain with the coldest air closest to the ground. Higher winds could cause turbulance and mixing of the cold ground air with the warmer air above.
If the latter two factors are the cause of the conditions here (and I believe that they are), then these results might not be as applicable in England or other lower elevation, higher humidity sites.
