Winter in the desert means really big changes in the weather, even over the course of a day. It's not unusual for us to have a morning temperature of 5F/-15C and late afternoon temps of 38F/3C. High pressure sits over the region for days, causing temperature inversions and freezing fog (the source of the rime in the picture) with really low visibility. I remember one winter a few years ago, flying an ILS into Salt Lake City when the visibility was 1 mile and finding that I was a little distracted by having so much to look at outside the airplane. When the visibility is 1/2, it takes longer to ask "Are those approach lights or touchdown zone lights?" than it takes to reach the ground.
The fog is so thin that the chance of "airframe icing" during the approach is negligible: your time in the clouds is best measured in seconds. It's also common to have a low ceiling (200 to 400 feet) with good visibility, with cloud tops only 2500 feet above the ground. That puts the tops at some of the intermediate altitudes for instrument approaches, and it's always a little eerie to fly an approach in severe clear with the wheels dragging through the clouds, especially at dawn.
In these conditions it doesn't take long to climb through the clouds, and even if you did pick up some ice coming through it would sublimate quickly in the intense sunshine above. Still, a lot of pilots with good instrument skills hesitate to launch into these conditions, not because it is dangerous, but because it is illegal unless the airplane is certified for flight in "known icing conditions."
Now, the FAA has clarified their interpretation of "known ice" in a letter from the Chief Counsel's office. (You can read about it here.) The short of it is what will be called the "reasonable and prudent" rule: if a reasonable and prudent pilot would consider that the current conditions imply negligible risk from icing, then "known icing conditions" do not exist.
I have mixed feelings about this. On the positive side, I may now launch through the morning fog, confident that there is severe clear a few miles away, sometimes. Or consider a day with a scattered-to-broken layer at 1,200 feet AGL. We had one of these the other day, and my student and I flew a circuitous route toward clear air that kept us in class G airspace so we could fly at 1,000 feet AGL (this is impossible in much of the world, but we can do it). We flew our maneuvers in the clear, and descended for the return home. The clouds were only about 1,000 feet deep, but by the old rule that would have been known icing conditions and we would have been in violation. With the new interpretation, I would fly an instrument approach and be done with it.
On the negative side, a few years ago a friend was giving some instrument instruction in a 172 near Rexburg, Idaho. The cloud cover was far from thin, but there were no pilot reports of icing. My friend and his student did the instrument approach into Rexburg, which is based on the Idaho Falls VOR; see the chart to the right. They got no ice, and took off IFR, planning to fly the ILS 20 into Idaho Falls. It's only 20 miles or so, but they didn't make it to Idaho Falls. They had so much ice that they could not hold altitude, and stalled attempting to land in a field. They were both badly banged up, and the airplane (which, by the way, was the first airplane I was ever paid to fly) was destroyed.
I'm pretty sure that my friend (now an RJ captain, his license suspension forgiven by his company) would claim that he had been reasonable and prudent: he had just flown the route without ice. But there was still too much ice for that 172.
So what is a reasonable and prudent pilot to do? First, use your head: you can't climb through 5,000 feet of ice in a 172. You can almoist surely climb through 500 feet, though. Maybe you can climb through 2,000 feet, and maybe not. What else?
Check pilot reports, either in your preflight briefing or at AviationWeather.gov. If there's a control tower, call them and ask what others have reported. You need to do this, because pilot reports don't always make it through the system to the pilot. I once flew an Archer from Washington National Airport to Griffiss AFB, NY, at night. I was on top and the air was smooth. At every handoff, I asked if there were pilot reports of icing, ready to make a new plan if there was going to be a problem. "No ice tonight" was the universal reply.
Finally, NY Center handed me off to Griffiss Approach. "Any ice tonight?" I asked. "Oh, yeah, everyone's had ice tonight." The controller was almost laughing. Grrr.. (Actually, it wasn't bad until I landed and found braking action nil on the ramp.)
Finally, you can check the Experimental Icing Forecasts at Aviationweather.gov.
Even with this, you need a plan. Ask yourself what you would do if there is more ice than a reasonable and prudent pilot would expect?. Knowing that there is clear weather a few miles away, or that the surface temperature is well above freezing, or that the tops are low makes launching reasonable.