Helen, it sounds like you have a very well thought out approach to an interesting problem.
My concern would be pointing any type of meter directly at the sun. I'm not sure if the circuitry involved in any meter would be able to deal with a gain of this magnitude. Reflected sunlight from a roof sounds ok, but I would hate to see you ruin a good meter on a purely academic question. It seems to me that metering an incident specular reflection and then an incident value of full sunlight would make more sense than trying to deal with the sun's full illumination.
Since we (typically) don't take pictures of the sun itself, an incident sunlight reading and a specular reading would suffice.
In any event, please post your findings as I'm curious about the results. tim