Fortunately, they have county-by-county data (and hourly data if you so desire) that spells out the design criteria for heating systems;

https://www.energystar.gov/ia/partners/bldrs_lenders_raters/...

Though heat pumps are unique in that they produce less heat as the colder it gets - A few hours of -15º every few years shouldn't be the primary consideration in spec'ing a system that still produces 75% of its heat in that worse case.

I live in that blue area and ran through all of the math and considerations recently - I pulled the hourly temperature data for 6 years. Of the 94,000 data points in that period, a total of 26 hours were below -5º: https://imgur.com/a/P7A3kan

Meh. I live in Wisconsin. It's fine.

> Our modeling finds that even if Focus incentivizes 800,000 heat pumps with electric resistance backup (10 times the number of heat pumps as it did furnaces in the past four years), the state will still be able to meet its electricity demand with currently operating power plants, even on the coldest days. Depending on the efficiency of the heat pump, in-state winter generation capacity would still exceed peak demand by 1,400–4,300 MW on the coldest day.

https://rmi.org/three-questions-wisconsinites-are-asking-abo...

The dataset is typical winter low for 1984-2014. Earth is warming thanks to climate change, and the lowest lows have notably warmed. See, e.g., https://xkcd.com/1321/. In my 12 winters of living in the light blue zone, I think I've spent like a grand total of 36 hours or so in sub-0°F temperatures, and that was pretty much the one period mentioned in the xkcd comic.

You can simply add backup resistive heat to cover the occasional 4AM drop to -15F. It’s more expensive to operate, but as long as this is rare it doesn’t significantly change your bill.