If the average annual temperature has been increasing over time, wouldn't you expect warm Septembers and warm winters to correlate (or warm any warm month with warm winters for that matter) during a warming climate? The majority of the cold-september-cold-winter cases are from before 2000 whereas more of the warm-september-warm-winter cases are from after. Would this not just be evidence that years as a whole are getting warmer?