Snow load and drifting
A roof in central Connecticut carries snow for weeks at a time, not for an afternoon. And it never carries it evenly: snow slides, drifts and piles against anything that interrupts the slope, so the load concentrates in a handful of predictable places while the rest of the roof carries almost nothing.
Where the weight actually goes
Three places, on almost every house. Against a wall where a lower roof meets a taller section — snow sliding off the upper roof lands there and stays. In the valleys, where two slopes funnel their accumulation into one channel. And at the eaves, where the last few feet of roof overhang unheated space and the snow there melts last.
The consequences are slow rather than dramatic. Sustained load in the same spots compresses the covering and works fasteners loose. Where a drift sits against a wall for weeks, meltwater has time to find its way behind flashing that would shed a passing rain without trouble. And on the older housing this region is full of — hand-framed roofs with rafters sized by eye rather than by table — thirty winters of that is visible as a sag along the ridge line, readable from across the street.
What helps, and what does not
Nothing about the covering changes what snow weighs. What changes is where the water goes when it melts: an open metal valley sheds a wet load far better than a woven shingle valley, ice-and-water membrane carried well past the eave line protects the strip that melts last, and step flashing at every wall junction — layered into the courses rather than sealed over the top — is what survives a drift sitting against it in February.
Raking snow off a roof is a common instinct and mostly a bad one. It is dangerous from a ladder, it damages the covering when done in a hurry, and it treats a symptom. If a roof genuinely cannot carry its normal winter, that is a framing question for someone qualified to look at framing, not a maintenance chore.