It is an odd situation, I agree. The rise, however, is nowhere as great as the depression that some people observe on their circuits.
Here, in the UK, the SNRM will slowly sink during the hours of darkness as long distance RFI from Eastern Europe gradually becomes more noticeable.
Looking at the plot you have shown, I would be more inclined to check what is happening with the clock on the system that is running DSLstats. It seems, from the way the traces zig-zag, that the time appears to "run backwards" for brief moments. I've seen similar, once before, but I cannot recall what was the problem nor how it was fixed.