If you were to move it to the 20, would you still allow 2 pt conversions from the 2?
I think you need to move it further out. The improvement in kicking accuracy in the NFL has been borderline insane over the past 20 years and shows no sign of slowing.
Look at it another way. What do you want the percentage to be? 99%? 95%? 90%? Lower?
Set the percentage and then look at the distances you're talking about.
In the NFL this year there were 253 field goal attempts < 30 yards. 6 were missed (a third of those by Jay Feely). That's good for 97.6%. Between 30 and 39 there were 295 attempts. 22 were missed. That's 92.5%. And, as with the shorter kicks, the majority of those misses are due to a small group of kickers.
The NFL doesn't publish more detailed kicking stats than that (collated, anyway), so with just those numbers and knowing about what the curve looks like, the average of the two numbers seems a likely estimate for the 37 yards a field goal from the ball spotted at the 20 would correspond to.
And that's today. 5-10 years from now that number is almost certainly going to be significantly higher.
As for the influence of weather, I was actually able to dig up a nice article on that (
pdf link) and the conclusion (page 8) is that at 37 yards, the effect is minimal. Neither rain nor wind really do much with those short kicks. The biggest effect was actually with "cold" games, but even then the effect was very small.
In other words, the weather would have to be pretty exceptional to nudge the field goal percentage from 37 yards below 90+%.
Those numbers do support keeping the 2 pt conversion from the 2-yard-line, of course.
Poll
Author