Those pictures (click on them to enlarge them) were taken about six hours apart on the day of our last spring snowstorm. The truncated-cone roof is standing-seam copper and encloses an open, unheated mechanical space.
If you look closely, you’ll see that the snow ends a bit short of the corner where the slope turns to vertical. There are guards at that location to prevent snow from sliding off the cone onto the streets below.
In the first picture, the depressions between the copper seams are pretty much filled with snow side-to-side, although the snow has already started sliding down the slope, creating some irregular curves in the pattern. Hours later, after warming by the sun, a lot of the snow has either melted or sublimated…maybe half of the volume of snow is gone. A small amount has made it past the snow guards onto the very bottom of the slope, but most has returned to the atmosphere. The thing that really catches my eye is the creation of beautiful serpentine curves in the snow on the slope. This has something or other to do with fluid mechanics and heat transfer between the air, the snow, and the copper. I doubt anyone could analyze the effect in enough detail to predict which way the snow would curve as it slowly melted and slid down the slope, nor do I care. But the randomness of the curve pattern is a good reminder of something that’s easy to forget: often the details we see of weathering damage don’t mean much. The overall pattern usually makes sense and helps us understand what’s happened in the past, but the small details, like those snow curves, are just meaningless and visually-interesting randomness.