[Cushman]

Really? Even assuming that we absolutely need to heat up the lights to solve this problem, we only need to do it when there is enough snow on the light to be a problem, which means at most only during snowfall, realistically much less frequently than that. Even in New England during winter, you've gotta be talking under 10% of the time that this has to happen.

You seriously can't come up with a better way to solve this problem than replacing a very efficient, very robust light source with a very inefficient, very delicate one, secure in the knowledge that as it runs all year round, some tiny fraction of the huge amount of waste heat it generates will go towards melting snow?

Although I'll note in your defense that "incandescent lightbulbs" are more accurately described as "heat lamps", and one potential solution might well involve a specialized heat lamp which only activates when the light itself is obstructed.

[ars]

Don't forget you need some kind of snow sensor - or at least a temperature sensor.

Or some kind of centralized switch.

Totally doable of course, but it does make them more complicated, more expensive and less reliable.

You would need to calculate the tradeofs. It's not automatic either way.

[Cushman]

A traffic light is already half of a snow sensor by virtue of being a light. If it's covered with snow, the snow will reflect light back. Add a few photoresistors and you're in business. It'd be the cheapest part of the whole project.