The intended amounts of food cached for winter seem far too small relative to the populations they must support and uncertainty of winter’s duration. Does this suggest that the primary strategy is to buy food, with the winter stores as more of a backstop?

You raise a good question, and all I can say is 

GRRM seems to think it’s enough.

Well, that’s not exactly true, there’s a bunch more I can say: 

There’s an underlying world-building problem here, which is that the multi-year seasons don’t really make sense when you consider the ecology of the life cycle of flora and fauna. If winter was just unrelenting night and cold and nothing else, you’d expect 100% die-off as seeds wither in the frost and animals run out of plants to dig up from the snow. (Either that or there are some truly baroque evolutionary adaptions that you’d think we’d have heard about by now) Likewise, it doesn’t matter how much you store and how cool your cellars are, there are hard limits to how long you can store food in a pre-modern context. 

So the way that I’ve rationalized it is that the seasons are really closer to climate cycles than what we think of as seasons – summers are extended warm periods, winters are mini-ice ages. While agricultural productivity is going to be much much higher in the “summer” than in the “winter,” it’s not the case that there’s no growth at all during the winter.  

Because even within the “winter,” you’re going to get variation in temperatures – your “false springs” and “spirit summers” – that allow for short bursts of agriculture productivity. Those little bursts are vitally necessary to stretch out your supplies, replenish fodder for whatever livestock and game is still around, repair some of the damage done by malnutrition, etc. 

But I would imagine that those are very chancey – if the lull in the snows and the cold ends before you can harvest whatever crop you’ve been able to get into the ground, you’re going to lose it all.