However, there is a great deal which is not clearly explained within the article or even the paper itself. I did give the idea a bit more thought on the bus last night and think that I now follow the idea. However, when the paper announces that it is counter intuitive to expect something to cool when taken from the shade and placed in sunlight this it to the extent that real scrutiny of the idea is still needed.
Here is my latest understanding of how this is possible:
A perfect mirror would reflect 100% of EM in both directions.
Objects radiate heat based upon their temperature
If you placed this mirror atop a hot source such as building then the mirror is reflecting the same radiation back into the building which is then the same as if the outside temperature was the same as the building.
This is used for situations where you are trying to minimise heat exchange such as a thermos flask or a space blanket.
So if you have a situation where the outside temperature is hotter than the building then neither conduction, convection or radiation can possibly cool the building.
The sunlight will interact with the surfaces it illuminates and generate heat. The surface will continue to get hotter until the outgoing heat is in equilibrium with the incoming heat. If you block the incoming sunlight then you create a potentially lower equilibrium.
If the sunlight was very crudely represented as 2 blue and 1 red and this was matched with the radiation from the hot surface as 3 red lights. Then if you have a material which reflects the blue but lets the red through then you would now only have 1 incoming red light but 3 red going out. So the surface can cool until it is at level which only needs one red light.
So this mirror can allow a surface that was above the new equilibrium to cool. Because it is being tested where the building has been illuminate by morning light until 10am already this then means that the roof is hotter than the temperature that it would be if continuously covered by a mirror. This can then cool the roof before it eventually warms as the sun continues to heat.
The practical limits of the temperature at which this operates as cooling rather than retarding the warming is limited to the initial covering of the surface which has very limited practical applications. And in winter it too would continue to cool at certain temperatures so less effective for year round than a pure mirror.
Some applications such as perhaps bus-stops or cars might be useful but the temperature range is very limited and for fixed structures it may be to increase the efficiency of some devices which generate heat such as traditional
air conditioners but most of the temperatures would be preventing an object hotter than the air cooling via convection.
The only real interesting applications might be for Geo-engineering arctic ice applications as the heat is dissipated to the atmosphere or a few specific chemical plants.
The roof scenario would almost certainly be better served by a stronger reflector and then using the light as either CSP or PV.