I won't debate your numbers, they look fine at a glance. However, supercapacitors are quite common, actually. You'll likely find a small one in any cellphone with a camera flashbulb. They're a relatively mature technology, as well, and not unreasonably expensive for what they do. On a larger scale, a single 48V 165F Maxwell costs on the order of $1-$2k, if memory serves. That's for a capacitor that has an ESR of around 5 milliohms, and which can deliver several thousand amps instantaneously, repeatedly, with no significant negative impact. Try doing that with batteries. My company uses these things on a regular basis, probably on the order of hundreds per year, integrating them into a number of projects on a variety of scales. I just finished designing a portable unit to charge and discharge 500V 130V capacitors, measuring their capacitance and ESR in the process.
Of course, I can imagine a different approach to the ice problem. Instead of melting all the ice over the entire road, they could melt paths in the ice instead, breaking it into smaller chunks which are easier to deal with and melt faster. Also, I'm wondering if there would be a significant difference between the amount of energy necessary to actually melt that layer of ice, and the amount of energy necessary to prevent it from forming in the first place. Neither of those would make orders of magnitude difference, probably. The fact that ice might not form or stick the same on a glass road as it does on asphalt might though...